Searching for the "black box"
June 01, 1994
by Teresa Acklin
Research in North America seeks to develop high-tech ways to measure, assess wheat quality.
For many years, the grain industry has been looking for a high-tech way to assess wheat. While the ultimate device remains elusive, research has produced technological advances and the quest is far from finished.
In the U.S., the goal is nothing less than “automated measurement and prediction systems to determine grain grade standards and end-use quality at each step in the grain handling industry,” according to the U.S. Department of Agriculture's National Research Plan, drafted in 1993.
Relying primarily on existing and future wheat quality research projects by the U.S.D.A.'s Agricultural Research Service (A.R.S.), the plan aims to ensure consistent U.S. wheat supplies with known end-use qualities.
Development of so-called rapid quality predictive testing nicknamed “black box” technology is one of the plan's specific objectives. Ideally, the new technologies would be able to predict end-use performance at any stage of wheat marketing even at the farm gate.
Flour millers for years have urged the U.S. wheat industry to put more emphasis on wheat end-use quality and measurement. A controversy in the mid-1980s over the classification, hard or soft, of a new wheat variety brought together all U.S. industry segments to address the shortcomings of the existing grading and classification system. At that point, the search began in earnest for solutions.
The push for a black box since has expanded to cover areas beyond classification, including technologies that can assess end-use quality. In 1993, 17 wheat quality research projects were under way within the A.R.S, covering areas from near-infrared analysis to “artificial nose” technology.
But the greatest advances to date appear in wheat classification, which has been under study for the longest time period. During the next year, the U.S. Federal Grain Inspection Service plans to install in its field offices the Single Kernel Characterization System technology that may replace visual wheat classification in the U.S. as early as 1997.
The Single Kernel Characterization System
The S.K.C.S. was invented by A.R.S. scientists at the Grain Marketing Research Laboratory, Manhattan, Kansas, U.S., which owns the patent. The A.R.S. reviewed proposals from 20 companies before signing an agreement with Perten Instruments, N.A., Reno, Nevada, U.S., to commercialize the system.
The S.K.C.S. measures four wheat characteristics: hardness, kernel size, kernel weight and moisture content. The equipment is unique because it measures these traits for individual kernels, according to Virgil Smail, director of the A.R.S. research laboratory in Manhattan.
Existing hardness testers using near-infrared technology rely on a ground sample, and they provide only an average for the entire sample, Dr. Smail noted. Thus, they cannot determine the variation of hardness within the sample, which is critical to assessing uniformity, he said.
Because it measures each kernel, the S.K.C.S. provides not only the sample's average hardness, but the variation of hardness, as well as information on individual kernels.
The standard sample size is 300 kernels, or about 10 to 15 grams, and the wheat must be cleaned of dockage, broken kernels and foreign material. The automated system weighs each kernel, measures it for size, and then crushes it to determine moisture and the crush force profile. The hardness measurement is calculated based on the other readings.
The entire testing process takes about three to five minutes, and the system has a 95% accuracy rate, Dr. Smail said.
Because the equipment will cost slightly more than U.S.$18,000, its use probably will be limited to large elevators, large flour mills and the F.G.I.S.
John Giler, chief of the F.G.I.S. standards and procedures branch, said the agency planned to begin putting the system in service this year and hoped to have its field offices fully equipped by May 1995, in time for the 1995-96 U.S. wheat harvest.
The F.G.I.S. will offer S.K.C.S. analysis in addition to traditional classification, Mr. Giler said. In that way, the industry can familiarize itself with the new system and discover how S.K.C.S.-measured wheat relates to traditional wheat classes.
The agency will designate four hardness ranges, soft, hard, semi-soft and semi-hard, to “classify” wheat. The ranges will be based on S.K.C.S. hardness readings, but the definitions have not yet been established, Mr. Giler said.
He said the system would not replace the existing U.S. classification system until May 1997 at the earliest.
“The change from visual to hardness is a big change,” he said. “We have to educate producers, breeders, traders and processors about the capabilities and what the system means.”
For flour millers, the single-kernel system holds promise as a way to improve the quantity and quality of information about wheat's milling performance. James Bair, director of government relations for the Millers' National Federation in the U.S., said millers would welcome the S.K.C.S. if it could deliver precise, accurate information.
“If instrumentation can be developed to help us identify, segregate and purchase high-quality wheat to make flour the baker wants, everyone benefits,” Mr. Bair said.
Nonetheless, some U.S. millers are cautious about relying too heavily on S.K.C.S. hardness readings, at least initially. That's because current technology does not enable the system to distinguish between the traditional wheat classes, such as hard winter, spring and soft wheats.
“A couple of years ago, we had a (crop of hard red winter wheat) that, on hardness alone, would have been classified as soft wheat,” one miller recalled. “But it sure wouldn't perform as soft wheat.”
Uncertainty also has surfaced from a marketing perspective because the traditional wheat classes, rather than hardness readings, form the foundation of the U.S. wheat pricing system.
Bert Farrish, vice-president of Columbia Grain, Inc., Portland, Oregon, U.S., said he supported the black-box concept, but needed more information to judge the merits of the S.K.C.S. specifically.
“If we can come up with devices that can grade grain and do it consistently, I think it will be good for all of us,” Mr. Farrish said. “The marketplace has been moving this way for 10 years; we're more end-user quality conscious now.”
Still, he said the system raised many unanswered questions, the first being its effects on pricing and merchandising. Accounting for the 5% error margin also is necessary to avoid discounts based on erroneous readings, he said.
In fact, the U.S. industry as a whole is uncertain how the marketplace will react and what adjustments might occur over time.
“The effects depend on whether people put more weight on the hardness numbers rather than on wheat class and vice versa,” said Jim Frahm of U.S. Wheat Associates.
Taking it one step further
The S.K.C.S. may have been designed to improve the wheat classification system, but Charles Deyoe, at the Grain Science Department of Kansas State University in Manhattan, has taken the system one step further. His research indicates that information gleaned from the single-kernel system can be used to estimate flour extraction rates.
In 1992, Dr. Deyoe tested wheat samples to determine if a correlation existed between mill yield and hardness, test weight, kernel size or kernel weight. He discovered that by combining all four factors, he could predict a wheat's flour extraction with a 95% accuracy rate.
The test was repeated on 104 samples of 1993-crop wheat with the same results. The finding means millers soon could have the ability to determine quickly and easily a sample's likely extraction rate and uniformity based on S.K.C.S. measurements.
This technology involves using a near-infrared source to reflect light off, or transmit it through, a sample. A substance then can be identified based on its reflected or transmitted wave length and the frequency on the light spectrum.
Near-infrared technology already is widely used to measure average moisture and protein in wheat and flour samples. Research is under way to use N.I.R. in a single-kernel system, which would provide much more information about uniformity than existing N.I.R. technology.
At the A.R.S. research center in Beltsville, Maryland, U.S., agricultural engineer Stephen Delwiche is studying N.I.R. as a complement to the S.K.C.S.
Although much more research is required, N.I.R. transmittance appears to be particularly feasible for single-kernel protein measurements, Dr. Delwiche said. If further research confirms the initial findings, N.I.R. protein readings could be combined with the S.K.C.S. hardness measurements to classify wheat into the traditional wheat classes.
Researchers also have been studying N.I.R. as a method to predict end-use characteristics, such as loaf volume, mixing time, tolerance and overall appearance.
But Dr. Delwiche said N.I.R. analysis of commercial flour samples had not been greatly successful in predicting these characteristics. Any additional research in this area is likely to focus on wheat's biochemical composition, rather than flour analysis, to predict baking quality, he said.
The concept of using digital imaging to assess wheat quality is under study by a number of commercial and public research groups, including the Canadian Grain Commission. Digital imaging uses a computerized method to convert pictures into numbers.
According to Stephen Symons, head, grain biology and image analysis for the Commission, researchers are using digital imaging to try to develop varieties that will produce more uniform wheat for milling and baking use.
Canadian researchers also have used digital imaging to develop a rapid fluorescence test for flour quality, Dr. Symons said. The test literally counts bran flecks, providing a quick and quantitative measure of bran contamination, which is particularly important in semolina.
At the A.R.S. research laboratory in Manhattan, Kansas, U.S., digital imaging has been successful in determining mold levels, insects, broken kernels and weeds in bulk wheat, Dr. Smail said. But further research will be needed to bring the technology down to an affordable, easy-to-use level, he said.
Digital imaging also shows promise in offering quantitative measures of crumb and texture in finished wheat products, Dr. Smail said. Some groups already have been able to bring the technology to a desk-top format for this use, and prototypes have been developed, he said.
Researchers in both Canada and the U.S. also are working on digital imaging technology to assess sprout damage. This research is designed to develop tests with more repeatability and consistency than existing testing methods.
Other efforts to create an automated, objective grain classification and grading system include research on everything from acoustical insect detection to artificial nose technology. Someday, in fact, the grain industry may be relieved of the need to sniff wheat and other grain to check for objectionable odors.
For the past three to four years, Larry Seitz, a chemist at the A.R.S. center in Manhattan, has been linking grain samples with known, identified odor problems to the odors' chemical profiles as revealed by gas chromatography. He has identified 125 primary chemical compounds and has built a database of this information. In the last year, Dr. Seitz has tested blind samples to determine whether odors can be identified accurately by checking their chemical profiles against those in the database. The results so far have been successful enough to continue the research, and a software system eventually could be developed for the grain industry.
Even though technologies like the “artificial nose” show promise, few in the grain industry expect to see wholesale changes anytime soon. Most people cite the near-decade spent to develop the S.K.C.S. as evidence that technological change will occur slowly.