First applied to food inspection in 1993 in a grain milling plant, GMS excels at moisture analysis. It also has the ability to provide feedback data on multiple constituents. That benefit is exploited in a system for rapid analysis of both fat and moisture in ground beef. A commercial system for ground-beef blending that promises enhanced accuracy and processing speed is under development. It incorporates a GMS system from Minneapolis-based Thermo Electron Corp. Helping to apply the unit to ground-beef analysis is Darrel T. Butler, Thermo Electron's GMS specialist. Butler began working with microwave technology 20 years ago while employed by a U.S. defense subcontractor that supplied electronic counter measures for missile detection and radar jamming. He joined a subsidiary of Baker Hughes in Houston, Texas, in 1993 as a GMS system was making the transition from prototype design to commercial scale-up. The business unit was purchased a year later by Thermo Electron, and Butler has been involved in all of the firm's GMS installations since then. He holds an electronics engineering degree from Devry University in Phoenix.
Food Engineering recently spoke with Butler about the physics of GMS and its application in in-line monitoring of food and beverage products.
Butler: The physics involve the transmission of electromagnetic energy at up to 750 frequencies in the microwave spectrum between a transmitter and a receiver, then measuring changes in the amplitude and frequency of the waves. There's no damage to the product because GMS operates on very low power, about one milliWatt-less than the energy from the typical cell phone.
Polar molecules in the material being analyzed rotate and align with the electromagnetic field. The movement of the molecules causes the microwave signal to be attenuated, or weakened, and the velocity of the wave decreases as it passes through the sample. Unlike near infrared, which is based on reflection, microwaves penetrate the entire sample. This eliminates the skin or coating effect from material buildup and provides a homogenous view of the entire sample.
The dielectric constant of the sample determines the cutoff region of the reading; after that, the signal enters the pass band region, where the amplitude of the signal is determined by the conductivity of the sample. The phenomenon is similar to what people experience when listening to a car radio. Radio waves reside next to microwaves in the electromagnetic spectrum. When driving under an overpass, an AM signal is cut off because the waves are too long to fit between the parallel plates of the roadway and the overpass. This is the wavelength at cutoff. FM signals are shorter and will pass under the overpass. The frequency of the signal relates to the amount of energy received by the radio, just as the amplitude of the microwave signal is determined by the conductivity of the sample. The cutoff region is sensitive to the amount of moisture in the sample. The strength of the signal in the pass band region indicates other constituents in the sample, each with its own electromagnetic signature.
FE: What was the first industrial application of GMS?
Butler: In 1989, two Texas A&M engineers started working with three controls experts from Fisher-Rosemount to apply the concept to industry. They had a lot of experience in petrochemical, so that was the original focus. Today, GMS is used for in-line monitoring of moisture content in oil and gas, coal and other material.
Although petrochemical was the early focus, the first big installation involved moisture analysis in milled corn. A miller was looking for a new technology because of inconsistencies in his corn cook & soak process. GMS requires a rectangular field for the receiver and transmitter. The chute through which the milled corn exited happened to be rectangular, so we mounted the transmitter and receiver on opposite walls of the chute. Signal data ran back to a PLC to control the addition of water. As a result, the mill went from variations of 1 to 2 percent in moisture content to plus or minus 0.2 percent. Analysis is done in real time, and adjustments can be made in one second.
FE: How has GMS been applied in food processing?
Butler: About 125 meters are currently in the field, measuring moisture in milled corn. GMS also is being used to measure moisture in dog food, candy-coated peanuts, dough, peppercorns and other foods. It's been applied for analysis of Brix, caffeine and acid in soft drinks; acid, viscosity, pH and salt in diced tomatoes and tomato paste; and salt and fat in peanut butter and mayonnaise.
An emerging application is analysis of both fat and moisture content in ground beef. We're working with a consortium of firms to develop a system for on-the-fly blending to a targeted fat content. Traditionally, processors have taken core samples from batches, analyzed them in a lab and then blended the batches to reach the target percentage of fat. GMS can eliminate the delay of waiting for lab results and the lean giveaway that occurs because core samples may not be representative of the mass.
GMS has the potential of allowing processors to add back moisture lost during processing. Up to 3 percent of moisture in meat is lost during processing, but because there was no accurate way to measure that loss and meter water back in, processors had to accept reduced yield. The system using GMS has been tested since last year at a beta site processing millions of pounds of ground beef, with very good results. The standard estimated error is less than 1 percent.
FE: Explain how the system analyzes a sample and derives a component profile.
Butler: It's not a direct measurement, like near infrared; it's inferred. What we're actually selling is the firmware in the electronics panel and the mechanics of the rectangular box. Coaxial cables carry readings on the electrical property changes of the material under test to the processing unit, where the slope and intercept of the cutoff and pass band regions are plotted to derive a rapid response analysis. As moisture content decreases, the cutoff band shifts to the right on the X-axis, which is a measure of frequency. You have to teach a GMS system what different levels of moisture or fat look like in a given product, and that is accomplished with multiple regression analysis to compare the plotted cutoff and pass band regions with the results of off-line analysis of the same sample. In this way, the system is trained to interpret the readings it receives from the product. If formulations change, the system is retrained in the same manner.
Ideally, you want to measure the composition of the entire product flow, and most of our applications have measured the entire matrix in line. But if the material is too conductive, no signal is detected, so a side stream is necessary. A product with high salt content being processed at a high temperature will attenuate the signal, for example. GMS has limitations; it's just another tool, and we have a lot of tools in the toolbox. But generally speaking, GMS excels at moisture analysis, and its ability to see as many as four constituents with one meter is a significant advantage over other technologies.
FE: What are the maintenance requirements of the unit?
Butler: The equipment is all solid state, with no moving parts. The only maintenance involves the seal between the transmitter and receiver in the rectangular wave guide. Those gaskets may last two years or longer, though harsh chemicals used in some CIP systems can accelerate gasket degradation significantly. For example, dairies use sodium hydroxide, acid and even steam to clean out lines, and seal life might be measured in months.
Once the unit has been calibrated, validation is accomplished with daily checks against the results from an off-line measurement tool. Results aren't going to be better than the off-line method provides, so good lab procedures are extremely important. That said, GMS provides real-time data feedback and rapid, nonintrusive analysis of several constituents. If the product's dielectric constant changes, we've got a shot at measuring it.