INTRODUCTION

The development of MIVAC technology is reaching the critical point for commercialization. The application areas of this innovative technology have gone far beyond the original intent, i.e., dehydration of grapes. Regardless of the timetable of its commercial applications, economics of the operation will ultimately be a dominant factor. Thus, the objectives of product development and equipment design are obvious: how to develop a quality product with minimum cost, minimum power and shortest time. For example, can a product of similar quality be processed at 750W instead of 1500W? Or, can a quality product of the same quantity be obtained in one hour instead of two hours?

MIVAC is a relatively new processing technology (Clary 1994). To this date, product development has been conducted in a heuristic manner. In order to shorten the product development cycles, there is a need to reduce the ³search space² and optimize data collection and analysis. It is with these considerations that we set out to further examine the temperature measurement in the MIVAC unit.

To meet the prescribed objectives, there is certainly a need to understand the behavior of this process in more definite terms ­ in other words, how the microwave interacts with various products under different conditions. However, thus far there is not sufficient data available to answer this question intelligently. This is mainly because of the difficulty of measurement in a microwave field and in the vacuum condition. Data collection and analysis is an essential element of the product development of the MIVAC pilot operation. Temperature measurement is a critical variable for microwave power control and recipe conformance. Therefore, reliable temperature data during the MIVAC treatments must be examined closely to ensure successful product development.

Previous studies by Cheng (1996) indicated that infrared (IR) readings were lower than those measured by a thermocouple when water was used for calibrations. The discrepancies are more significant near the water boiling temperature (212°F). A preliminary comparison between an IR detector and a fiberoptic measurement also showed similar differences. However, those tests were conducted under atmospheric pressure. The discrepancy may be attributed to the ³fogging² effect on the IR detector in the presence of steam. Nevertheless, it does raise concerns as to how reliably the IR detector measures under a vacuum.

"Benchmarking" for temperature measurement in the MIVAC unit presents a technological challenge due to the presence of microwaves and the vacuum condition. Infrared technology allows temperature sensing from a distance. However, it can be error-prone if not carefully calibrated or focused appropriately. The best available technology is fiberoptic probes. Following are brief descriptions of these two technologies.

INFRARED TEMPERATURE MEASUREMENT

An infrared (IR) detector is capable of reconstructing a thermal image on the surface of the product (Berek and Wickersheim, 1988). However, an IR detector has certain limitations. Only surface temperature is measured, and the correlation with internal temperature may be difficult to determine, particularly in the presence of berry skin or protective membranes. The IR detector also requires that each object of emissivity be known for accurate temperature readings (Bengtsson and Lycke, 1969).

Emissivity is a surface property of a material. When radiant energy strikes the surface of an object, it is absorbed, reflected or transmitted. This can be simply summarized as:

a + t + r = 1

where a is the absorbtivity, t is the transmitivity, and r is the reflectivity of the surface.

According to Kirchoff's law, all the energy absorbed would also be emitted for a 'blackbody' at the equilibrium temperature. That is a=e=1. However, most real life surfaces are not 'blackbodies,' and e is always less than 1. It is also important to notice that emissivity is also a function of temperature and wavelength. For accurate temperature measurement with an IR detector, the emissivity of the product must be determined and accounted for.

Practically speaking, the emissivity of an object can be defined as the ratio of the radiant energy emitted by a body to the radiant energy which would be emitted by a blackbody at the same temperature. In short:

e = W_o/W_bb

where W_o = total radiant energy emitted by a body at a given temperature T, and where W_bb = total radiant energy emitted by a blackbody at the same temperature T.

Fruits, vegetables and foodstuff are subjected to structural changes upon heating. This means the surface properties, such as emissivity, reflectivity, etc., are strong functions of the temperature. For example, a grape berry may have a value of e = .55 before MIVAC treatment. The value of e may increase to .8 or .9 at the end of the treatment, depending on the process condition. Furthermore, it may increase in a rather non-linear fashion. The implication of this example is that the emissivity setting for the IR detector should be continuously adjusted during the heating process in order to obtain accurate temperature readings. From the system control point, this is very undesirable.

The IR detector used in the MIVAC unit is a MIKRON M67 Infraducer. The Infraducer has a field of view of 15:1. In other words, it averages a target area of about 1" diameter at a distance of 15" from the sensing point. The quality of air may greatly affect its accuracy of reading. Fortunately, MIVAC always operates under a vacuum. Air quality is not an issue in this application. However, temperature readings for any product size that has a major dimension less than 1" may inherently incur measurement errors. False temperature readings may easily cause over-treatment or under- treatment when automatic control system is used.

FIBEROPTICS TEMPERATURE MEASUREMENT

Typically, a fiberoptic temperature detector is made of mixed phosphor with a transparent binder and is formed into a small disc that is attached to the end of a silica fiber (Berek and Wicksheim, 1988). The fiber can transmit light efficiently to any location for temperature measurement. Phosphor can be made to emit light when excited by radiation of higher energy level or shorter wavelength. An example of phosphor mix is magnesium fluoger-manate activated with tetravalent maganese, which has a decay time when measuring temperatures, from about 5 ms at 200°C to .5 ms at 450°C.

The fiberoptic temperature sensing and pressure sensing unit used in this test is Metricore 2000 manufactured by Photonetics Inc. This unit employs a different fiberoptic sensing technology known as ColorOptic measurement. Light is supplied to the sensor assembly by the measurement instrument through a fiberoptic cable. The sensor itself consists of a layer of transparent material whose refractive index varies significantly with temperature, sandwiched between two layers of reflecting materials. The corresponding variation in the reflection conditions modulates the color of the reflected light. This color is in turn ³translated² by the instrument into measured temperature values.

The advantages of a fiberoptic temperature detector are 1) It can measure temperature for a specific point inside the product; 2) It does not depend on the physical properties of the measured product; 3) No calibration is required for the instrument; 4) It is not affected by electromagnetic waves or air quality in the chamber; and 5) It has the provision to measure internal pressure of the product under vacuum. However, there is one drawback for the MIVAC application: it may break the vacuum if it is not properly sealed. Also, the probe may be damaged if it is attached to the product when the turntable is running.

TEST PROCEDURES

To validate the IR temperature readings, two Photonetics fiberoptic probes were used: T51-01D and T52-01D, respectively. Both have a range of 14 to 240°F. The probes were inserted into the MIVAC chamber and sealed with silicon glue. Probes were connected to channels 1 and 2 of a Photonetic Multisensor System. The unit has 0-10 V output for both channels. These values are converted and recorded in an automated data acquisition system (Kohl 1998). Temperatures were monitored and stored with the aid of WonderWare human-machine interface software. Data were further converted into Excel spreadsheet file format (.xls).

As alluded to previously, because benchmarking IR with water under atmospheric pressure may conceivably introduce errors due to the fogging effect of steam, mashed potatoes under vacuum were used instead of water. Samples of 2.5 lb. were prepared by mixing water with a box of IdahoanR Mashed Potatoes (net wt. 155g or 5.5 oz.). It was placed in a black plastic bowl. Tests were conducted under three power levels: 500W, 1500W and 3000W. Data were taken without turntable running.

One probe (Channel 1) was placed about 1" inside the product sample and another (Channel 2) was placed freely on the surface of the mashed potato sample. An IR detector was then focused to the surface probe. Tests were conducted by drawing the vacuum prior to applying the microwave power. Channel 2 data were recorded every minute from the screen while the rest of data were recorded automatically.

RESULTS

Figures 1 and 2 are comparisons of surface temperatures measured by the Photonetics probe and IR detector @ 1500W and 3000W, respectively. During the 1500W test, the difference between the two readings did not exceed 5.3°F. The IR readings are lower than those of the Photonetic probe. While during the 3000W test, the maximum difference was 4.1°F. At the beginning of the test, Photonetics probe readings were higher than those of the IR detector. However, they crossed over toward the end of the test.

Practically speaking, these temperature discrepancies are not significant. There might be several affecting factors for the difference: the probe is measuring a particular point and it is likely below the surface of where the IR is reading; moisture boundary layer above the sample surface could cause a lower IR reading; or the sample property may change when the surface is heated much faster than the interior. Nevertheless, at room temperature readings the IR and the probe are within 2°F.

Figures 3 through 5 summarize the temperature curves at the sample surface as well as interior, at 500W, 1500W and 3000W, respectively. It should be noted here that the temperature difference is a measure of the temperature uniformity between the interior and the surface. The maximum temperature difference for each case is 16°F, 15.1°F and 6.5°F, respectively. The minimums for temperature are 1.1°F, 0.6°F and -4.5°F.

It is interesting to observe that microwaves were powerful enough to penetrate the sample at lower powers, e.g., 500W and 1500W. It caused rather large temperature differences between the interior and the surface. Under both cases, one may expect fairly non-uniform temperature distribution inside the sample. In other words, heating is more uniform throughout the sample at higher power level.

Superimposed temperature curves at the three powers are shown in Figures 6 and 7 for both IR readings and the probe readings. The general heating characteristics are surprisingly different: at 500W, there is a significant "hump" at the beginning of the heating process. It indicates that there is a cooling stage while moisture being driven out. Similar humps were observed in previous tests with grapes at low powers (Clary 1994 and Cheng 1996). The likely explanation is that heating is fairly localized at the low power, and gas pockets may be formed in the early stage, thus drawing heat from its neighborhood. Such a hump is not present in Figure 7. The strength of applied power was clearly manifested in the "climbing" slopes of these temperature curves as well as where the "plateaus" settled.

Aside from temperature curves, it is interesting to present the product weight as a function of time during the tests. These are summarized in Figure 8. The fairly "flat" weight reduction at the beginning stage indicates a more constant dehydration rate. Though toward the end of the 3000W test, the weight reduces significantly. However, the slope of this curve is rather constant, indicating the rate of the weight reduction is a constant.

For comparison purposes, a thermal efficiency was calculated for these tests. Thermal efficiency can be considered as an estimation of the dielectric loss factor (Chang 1996). It is calculated by determining the ratio of the energy required to remove the total amount of moisture (in this case, the total weight reduction) to the amount of electrical energy applied during the process. The calculated thermal efficiencies are 13% @ 500W, 9.5% @ 1500W and 7.8% @ 3000 W, respectively.

CONCLUSIONS

For the first time, we are able to measure the temperatures inside samples with the aid of the fiberoptic temperature measurement unit. These results indicate that the temperature differences between the interior and the surface are significant, particularly at lower powers. It is recommended that such differences be studied for future product development in relation to physical structures, size, shape, economic constraints and desired taste.

The tests at 1500W and 3000W indicate that the maximum differences in surface temperature readings between IR and fiberoptics were about 5°F at the beginning of the test. The differences reduced nearly to zero toward the end of the test. These results confirm the limitations described earlier for the infrared temperature detector. The IR temperature detector works well at moderately high power while temperatures are uniform throughout the product. The implication is that IR measurements may serve as a reliable temperature control when samples are small in size, have a low moisture barrier and a light load. Generally speaking, since MIVAC load always decreases in the course of treatment, using an IR reading to control power does not appear to be a problem.

Nevertheless, the fiberoptic temperature instrument is a useful tool for product development, should more precise temperature control be required. As time progresses, recipes of various new products would begin to emerge. Precision temperature control would become necessary. IR temperature measurement must be "bench-marked" by the fiberoptic temperature sensor in order to provide repeatable results.

The automated data acquisition system proved to be an excellent product development tool. Not only did it save time and drudgery in recording data, it reduced transcribing errors and allowed us to scrutinize data at various levels. With the capability of the fiberoptic sensor to measure pressure, it is further recommended that tests should be conducted to determine the possible uses of this information for future recipe development. Characteristics of the temperature curves should eventually be correlated with respective products. Such analysis may ultimately reveal some insights for product development.

REFERENCES

Bengtsson, E. N. and Lycke, E., 1969. Experiment with a heat camera for recording temperature distribution in foods during microwave heating. Journal of Microwave Power 4(2):48-54.

Berek, H. E. and Wickersheim, K.A. 1988. Measuring temperatures in microwaveable packages. Journal of Packaging Technology 2(4):15-21.

Cheng, C. C., 1996. Temperature distribution and Measurement for Grape Puff in the MIVAC System. M.S. Thesis, California State University, Fresno.

Clary, D. Carter. 1994. Application of microwave vacuum and liquid media dehydration for the production of dried grapes. Ph.D. Dissertation, Michigan State University.

Kohl, R., 1998. An automated data acquisition system for the MIVAC unit. Senior project of Department of Industrial Technology, CSU, Fresno.

Yen, M. and Clary, C. D. 1995. Why is the grape puff puffy? An analysis of MIVAC temperature curve. California Agricultural Technology Institute Research Bulletin #951101, CSU, Fresno.

---

{ CATI , VERC , VERC - Research Publication }

 

Copyright © 1998. All rights reserved.
CALIFORNIA AGRICULTURAL TECHNOLOGY INSTITUTE - CATI
College of Agricultural Sciences and Technology
California State University, Fresno