"Different potatoes have different percentages of solids, colors and grades, and the desired properties are different depending on whether you want to freeze, dehydrate or make chips out of them. But the production challenges go beyond that," an executive at one of the largest North American potato processors confides. "The crop changes every year, but your customer specs don't."
One man's raw material is another man's finished good, and most processors in the supply chain can lean on suppliers for more consistent raw materials. But Mother Nature is the ultimate supplier, and she has the last word on the foodstuffs and livestock that enter the chain. While she will never adhere to six sigma principles, she is receiving more and more input from experts in bioscience, genetics, cereal chemistry and continuous improvement.
Vertical integration has helped the poultry segment reduce pathogen loads in flocks sufficiently to facilitate high-speed production. "With the advent of HIMP (HACCP-based Inspection Model Program), theoretically there is no limit to line speed, provided you can control variability and meet the performance standards," notes S.F. Sarge Bilgili, poultry scientist at Auburn University and a leading authority in the deployment of production and processing technologies to improve carcass quality, yield and wholesomeness. "Skin quality, meat yield, fat content and bone strength are all issues that come up, but problems usually occur in farm management, not the fairly narrow genetics we work with. Biological variation will always be there, and the biggest variable for processors is bird size. We still don't have smart machines that can adjust for size; they can only be calibrated to averages."
Deviations of up to three-quarters of a pound exist because two bell curves are at play: one for males, the other for females. Segregating broilers by sex is done, but it's a costly, manual, time-consuming process, and the slow-feathering gene that is added to the birds is associated with slower growth and a specific virus. If an automated gender-sortation system can be created, that gene can be eliminated, and processors would reap the benefit of more consistent stock. Field tests of the first prototype of a gender-sortation system were completed earlier this year, with a second-generation prototype scheduled for deployment in 2004.
The project is under the direction of Patricia Phelps, senior director of product development and marketing at Embrex Inc., Durham, N.C. Embrex already provides automated systems that deliver vaccines and other serums to broilers in hatcheries. Twenty-gauge needles inject eggs as they are conveyed at rates that can exceed 800 eggs per minute. Phelps recently developed an inexpensive assay to detect elevated levels of estrogen in the allantoic fluid of eggs incubated for 16-18 days. Chicks hatch at 21 days.
To determine gender, needles extract 20 microliters of fluid through a 1 mm incision, and a sophisticated tracking system matches the eggs in each bar-coded flat to a template that determines which egg yielded what fluid. Hatcheries like those of development partner Cobb-Vantress incubate millions of eggs at a time, so gender determination must occur quickly. Results that used to take four hours now are available in 30 minutes, "and we hope to get that down to a minute or less," says Phelps. Automatic sortation equipment then segregates the males and females in each flat.
"It's extremely challenging to devise a system with a very short assay time, throughput of 20,000 eggs per hour and at a cost that meets the hatcheries' requirements," she adds. "On a limited basis, we're running at 100 percent accuracy in gender tests and feel confident about the test. But this is a critical part of the customers' process, so we're going slow to make sure it works on a day to day basis before bringing it to the commercial market."
A pioneer in pig genetics was the UK's National Pig Development Co. (NPD) and the NPD pig is the basis of the lean hog developed by Smithfield Inc. The NPD pig also provided the gene pool at Saskatchewan-based Genex Swine Group Inc., but one size doesn't fit all of today's pork business. Genex developed six boar lines serving 1,200 sows to meet the specifications of individual customers. In October, Genex merged with Hypor International, the swine genetics arm of Nutreco Holding NV, a Dutch firm involved in breeding, feeding and processing poultry, pork and fish. Nutreco operates 120 food production and processing plants worldwide.
Since 1987, the company has linked slaughter data on individual animals back to breeding programs, greatly improving meat consistency. But much greater strides in eating quality, pathogen reduction, water holding capacity and other attributes are being attained through genetic engineering, and the impact is just beginning to be felt, suggests Benny van Haandel, head of R&D at Hypor.
"Until 1990, even the university genetics programs were qualitative based," says van Haandel, who studied genetics and statistics at Wagingen University in the Netherlands. "Then biotechnology came into play, and a lot of comparative mapping of linked chromosome areas in different animals became available. All of the university projects in human and animal genetics are linking their libraries on the Internet. Once you know the regions where you need to focus, it's simple to reach the related human databases."
Based on swine research, some human genome mapping has been amended. More frequently, efforts to unlock the secrets of human DNA are helping swine researchers identify and isolate chromosomes that dictate specific traits.
For example, breeders have understood that marbling is determined by genetics, but only recently have hog geneticists come to understand that intramuscular fat (IMF) also can be controlled through genetics. "With all the bit mapping of chromosomes, we are getting very close to the regions that affect IMF and fat deposits," van Haandel says. Likewise, a gene relating to muscularity and thought to be fixed is now understood to be a very close chromosome that can be split and controlled.
Fragmentation of the supply chain makes it much more difficult to reduce variability in beef, where outcome data seldom are available to breeders. Closer relationships between the various players are beginning to change the paradigm, however. ContiBeef LLC, the world's largest cattle feeder, struck a partnership five years ago with what is now Swift & Co.'s beefpacking operations. Qualitative data on individual animals to address the variability issue through continuous improvement is part of the focus. "Swift is excellent about providing ContiBeef with data we can use to make more informed decisions about our production practices," reports Mike Thoren, ContiBeef's vice president of operations.
Quality improvement through decreased variation has been an objective at ContiBeef for some time, Thoren says. Statistical process control (SPC) was introduced to feedmill operations a decade ago. A year ago, the company began working with the Center for Continuous Quality Improvement (CCQI) to expand the effort to establish control limits to reduce variation. CCQI defines quality as a lack of variation, explains Thoren, but the statistical aspects are sublimated to the management and teamwork principles involved. "It is very common sense and doesn't chase the sizzle that some high-profile programs do," he says. Sixty managers have been trained in the program, and at least 30 percent of the workforce will be indoctrinated in the next two years.
"ContiBeef has focused on a definition of quality that is based on low variation," CCQI founder Robert Galina says. "We've worked with most of the top pork producers on similar programs, but I don't know of anybody else in the beef group to institute systematic management in this area."
Application of SPC to reduce feed variation to boost cattle weight-gain rates and reduce feedmill costs is part of the effort, but it also requires a management shift. "In the past, people bought from whatever supplier was cheapest and could meet their specs, but multiple suppliers mean more variation," Galina says. "There are still people in manufacturing who source that way, and they're struggling."
Sajid Alavi, assistant professor in the grain science & industry department at Kansas State University, recently won a $249,000, three-year grant from USDA to determine what kind of microscopic air structure occurs in an extruded dough at a given screw speed, pressure and barrel temperature. The calculations are keyed to the glass and melt temperatures of that dough. A closed-chamber capillary rheometer to determine those phase changes will be used in conjunction with Alavi's finding to enable manufacturers to calculate outcomes before start up.
"Extrusion processes use many different formulations, so it's very hard to predict how a product will perform without a lot of experimentation," says Alavi, a food engineer and expert on advanced extrusion processing. "What is needed is a bulk approach to analysis, rather than a difficult-to-apply micro approach."
Micro aptly describes the traditional methodology for evaluating these raw materials. Protein content, ash, water absorption and other characteristics have been studied ad nauseum to help predict outcomes. Now Alavi and grad student Allen Trater are mapping moisture content and the relative amount of sugar, lipids, protein and starch in a dough, then relating those factors to the glass phase (PG) and melt phase (PM) of a dough. Then, they examine the cell structure of the extruded dough using X-ray tomography, a 3-D imaging technology. By correlating the PG and PM of the dough compositions that yield the optimal end product, Alavi expects to build a computer model that "fills in the blanks" on how to control the extrusion process for products that are more flavorful, crunchier, less susceptible to staling and other desirable characteristics.
"Over the years, many people have looked at a variable like sugar content, done empirical correlations based on regression analysis, and tried to predict outcomes," he says. "But that doesn't get to the fundamental of why it is happening. How material flows is the most important parameter," and a phase transition analyzer (PTA) can ascertain viscosity.
Three years ago, engineers at Wenger Manufacturing built the PTA (see "Shape of things to come," Food Engineering, March 2002). The conceptual framework was laid a decade ago by noted cereal chemist Carl Hoseney. PG describes the point when a dough changes from a crystalline state to a rubbery consistency. PM is the point where flow occurs. "The applications for the PTA have not been explained yet to people," notes Alavi, making it a commercial tool ahead of its time. But once PTA can be incorporated with control loops and predictive models based on the work underway at KSU, processors will have an effective tool to achieve product consistency.
Better understanding of the dough's basic composition will allow processors to make extrusion more of a science and less of an art. Similarly, better understanding of the nature of other raw materials-whether animal, vegetable or mineral-is enabling food companies to more easily control product outcomes. In an industry where quality is synonymous with consistency, reducing raw material variability is a critical first step in the march toward higher-quality products.
For more information:
Robert Galina, Center for Continuous Quality Improvement,
Patricia Phelps, Embrex Inc.,
Marc Broadbent, Hypor International,
Sajid Alavi, Kansas State University,