Smarter Food Safety Relies on More Data, Compute Power
Demand is rising for improved precision in food inspection. Food producers are taking advantage of the data and processing capabilities in their systems to not only improve food safety, but also to work toward improved traceability.
ProdX allows multiple facilities to be connected to a single data point.
Mettler Toledo
When it comes to food safety, today’s food producers are dealing with a whole host of mounting challenges. A reduced use of pesticides—whether because of regulations or the popularity of organic farming—is giving rise to more toxic weeds within crops. Climate change is causing increasingly intense droughts and rains, which in turn causes a rise in aflatoxins, typically in nuts. Social media is a bigger headache than ever, where inspection slips can go viral instantly. Add the COVID-19 pandemic to the mix, intensifying an already difficult workforce situation along with the challenge of having key experts work from home.
Food inspection technology needs to be not only more reliable and more precise than ever, but smarter as well. Last year, the U.S. Food and Drug Administration (FDA) announced its New Era of Smarter Food Safety, an initiative intended to build on the Food Safety Modernization Act (FSMA) to create a more digital, traceable food system. The FDA initially planned the release of the 10-year blueprint in the spring of 2020 but COVID-19 pushed it back to July 2020. The challenges that arose during the pandemic showed just how necessary some of the actions in the blueprint are.
“It’s more important today than ever before in our history to work together to create a more digital, traceable, and safer food system,” said Frank Yiannas, FDA’s deputy commissioner for food policy and response, at the release of the plan.
Track and trace
Though FDA’s new program is largely focused on technologies that could be used within the supply chain rather than inside the factory, there will likely be a role for inspection to play from a traceability perspective, according to Robert Rogers, senior adviser for food safety and regulations at Mettler Toledo’s Product Inspection division. “Especially in the age that we’re living in today—the pandemic and restricted travel and restrictions of going into plants less—it’s totally essential,” he says.
In this climate, regulators might look for ways to better focus on the inspections that are most needed. In other words, if a food producer can document that it’s performing required tests, getting good results, and basically handling food safety the way it should, that would help provide confidence to a regulator or auditor that their time could be better spent elsewhere. In fact, in its blueprint, FDA recommends evaluating the feasibility of using remote and/or virtual inspections of companies with a demonstrated history of compliance.
See it Live at PACK EXPO Connects Nov. 9-13: Total Process and Quality Management of PET Containers, by AGR International, Inc. Preview the Showroom Here.
Mettler Toledo is beginning to do some work in that space, Rogers says, providing customers access to the type of data needed for this and other scenarios. “We have data collection capabilities through our ProdX software that allows multiple facilities to be connected to a single data point and have all the systems from those multiple facilities provide information into this,” he says.
Tomra Sorting is also putting more emphasis on its Tomra Insight, an Industry 4.0 monitoring platform designed to turn sorting machines into connected machines to keep them performing at their best, says Jeffry Steemans, product manager for digitalization at Tomra Food. In addition to helping to reduce downtime and maximize throughput on the machines, the data also helps manufacturers sort to high-quality levels to maximize profitability, he adds.
And, what food processors have found particularly important during the COVID-19 pandemic, Tomra Insight enables more employees to work from home, able to log in to a website, and see the machine remotely. Food processors are challenged by guidelines to keep the number of people on-site to a minimum. “That means that when they’re running the plant, experts cannot always be on-site,” Steemans says. “But they need to have the information on hand if something goes wrong. It allows experts to see what is happening.”
The future of food inspection technology is now
When it comes to making use of available data, Rogers says that perhaps the future is coming a lot sooner than people might think. Connected devices are important on many different levels within the food business already, he says, not only from reporting activities on inspection systems, but also for reducing downtime. But traceability will be an increasingly important use of data as the industry moves forward.
“I believe that there’s benefits to using some of the models in other industries, such as serialization in the pharmaceutical industry, where they’re able to identify what blister pack is in a pallet of cases of cartons of medicine,” Rogers says. “That sort of serialization and aggregation down to the item level, moving that toward the food industry, is something that would greatly benefit the food industry and the initiatives along the lines, certainly, of traceability.”
Track and trace technology that’s used heavily in the pharma industry is starting to trickle down into the food industry, says David Lamprey, senior product marketing manager at Thermo Fisher Scientific. “Food is starting to take that pathway, but it’s still in its infancy now. Part of the reason is because of the sheer cost to implement that process and do it across the entire horizon of food safety,” he says. “The food safety industry is dabbling and changing their whole approach from reaction when there’s an issue out in the field and doing things like recall to more prevention.”
When one of its customers decided to move forward with a track and trace initiative, Fortress Technology worked alongside them to enable its customer to pull the data it needed from its metal detectors. “It allowed them to not only say that they have a metal detector, prove that it’s working, and show when they’ve gotten rejected, but it allowed them to give their customers visibility on a product-by-product basis of the results from the metal detector,” explains Eric Garr, regional sales manager at Fortress Technology.
Track and trace will become more prevalent because of both the push from regulators and the pull from the consumer side relative to increased levels of safety, Lamprey says. The cost and complexity are worthwhile, he contends, because of the benefits of avoiding the kinds of recalls that are common today. “It’s a massive recall because they don’t know what the scope is, so they pull everything back and it gets very costly,” he says. “If you’re able to do track and trace, you know specifically where a product is and can act immediately on a much smaller scale. So, although there’s a big investment upfront, the back end is much less expensive and the brand protection is much greater, as it’s tied to social media today.”
More data, better detection
Putting aside the bells and whistles that might push the food industry into new levels of Industry 4.0 capabilities, inspection equipment suppliers are also working to improve sensitivity levels of their systems—finding smaller contaminants, distinguishing between similar contaminants, and generally discovering the issues that are giving food producers more concern about the safety of their products.
Increased data capabilities and computing power have roles to play in optical inspection technologies, which are able to detect smaller and smaller objects, says Stephan Westcott, global AIS product manager at Key Technology. “We’re able to process more and more data so we can get to smaller pixel sizes, which ultimately translates to smaller defects or foreign bodies that we can go after on the detection side,” he says.
Increased processing power also enables the technology to pull in additional channels of information, Westcott notes. “It used to be that you could only use just a couple of channels. And if you had a machine that had both cameras and lasers, they were often independent; they were making independent sort decisions,” he says. “With our latest platform, we’re actually fusing that information together so you can have both cameras and lasers and make smarter decisions about the classification of the objects you’re looking at.”
See it Live at PACK EXPO Connects Nov. 9-13: Dual Head Automatic Filling, by AMS Filling Systems, Inc. Preview the Showroom Here.
Key Technology’s VERYX digital sorters use what the company calls Multi-Sensor Pixel Fusion, combining data at the pixel level. “With the combination of that information, you’re really able to go after both product defects as well as foreign bodies,” Westcott says.
Hyperspectral imaging takes that to hundreds of wavelengths—hundreds of channels of information, essentially—to create a hyperspectral signature. The technology is not exactly new, but it continues to become more advanced and more accessible, Westcott says. And it’s well-suited to finding more contrast in defects that might not be as easily distinguishable between safe and unsafe. “Take a french fry that’s got a bit of a rotten end on it that has gone through the fryer. It’s safe to eat, but the color is a bit darker, black even,” he says. “To a legacy sorting system, it’s really difficult to tell the difference between a piece of charred black fry and a piece of black rubber.” However, because potato and rubber have completely different moisture content, the hyperspectral imaging can easily distinguish between the two.
Technologies like hyperspectral imaging or pixel fusion also provide additional tools for dealing with the increase in extraneous vegetable matter like datura or nightshade. “A nightshade berry might be indistinguishable from a pea to the naked eye, either in color or shape,” Westcott says. “In nuts, they have issues sometimes with aflatoxin. It’s an invisible defect. You can’t see it, but it’s dangerous.”
Tomra Sorting uses its Detox laser technology to identify aflatoxin in nut-related products. It uses a special optical configuration to capture the low intensity of light reflected by the fungus to detect aflatoxin contamination. This is an alternative to blanching the nuts, which is “quite expensive and also reduces shelf life,” Steemans says.
Another Tomra technology called Biometric Signature Identification heads into the near-infrared region of the spectrum to provide something akin to DNA fingerprinting of an object based on light. It is possible to identify certain chemical and molecular differences on the surface and also within the objects inspected.
A place for artificial intelligence
With the kind of processing power available today, artificial intelligence (AI) is making it into all corners of the manufacturing space. Tomra has ventured into AI and neural networks with its recent launch of LUCAi. The technology—which grew out of the company’s acquisition of BBC Technologies in early 2018—is used specifically to grade blueberries. Able to process 2,400 individual images of fruit per second, the platform can better detect specific defects, and even enables variety-specific and seasonal-specific sorting. Looking to the future, Steemans sees a place for AI in food safety—to better identify the potentially toxic solanine in potatoes, for example, which is increasing with climate change.
Just as important as finding contaminants is finding something that isn’t there. Not only will false rejects waste product and time, perhaps the most devastating effect is the loss of confidence in the detection system itself, Rogers says. When that mentality starts to creep in, operators might lower the sensitivity, leaving more opportunity for contaminants to pass through.
Some metal detector suppliers have turned to simultaneous multifrequency technologies to help squash those false rejects that can occur, in particular, with high-conductive products, such as meats or fresh baked breads. Metal detectors can be challenged by product effect—temperature variations, or high salt, iron, or moisture content that can create signals falsely seen as contaminants. “These product characteristics with their high conductive effect and variability can limit the achievable performance,” Garr says.
See it Live at PACK EXPO Connects Nov. 9-13: Reduce Labor Costs with Integrated Coding Technology, by AT Information Products, Inc. Preview the Showroom Here.
Fortress Technology’s Interceptor combines multiple frequencies to better understand the content of the product. “Different frequencies are going to have different strong points, so combining different frequencies, you can use that information more effectively to reduce false positives and improve sensitivity,” Garr says.
Mettler Toledo’s Profile Advantage metal detector uses similar technology—which Mettler calls multisimultaneous frequency—to improve contaminant sensitivity in meats and other difficult foods. “Historically, those were challenging applications because of the wide product diversity that occurs during a normal production run—diversity in temperature, diversity in moisture, and things like that,” Rogers says. “That resulted in turning down sensitivities of the inspection device to deal with the diversity of the normal product conditions. The Profile Advantage introduction really got rid of that. Now you can run at the highest of sensitivities.”
Thermo Fisher Scientific is tackling issues associated with product effect with its new Sentinel 1000 Selectscan metal detector. The system allows the user to choose the best frequency for its application from 50 to 1,000 kHz.
Trying to dial in to particular product effects—is it wet, does it have salt in it, what’s the temperature—can complicate measurements for food producers. “The more you get into that side of it, the more complex it is to pick up components in that because it generates electricity as you’re running a charge through it,” Lamprey says. “Selectscan allows the customer to set the detection levels specifically where they want based on their manufacturing process, which ties back to an increased probability of detection, which obviously drives back to the consumer safety side and then the brand protection side.”
That ability to set the application at a particular frequency also enables Selectscan to better find the metal that’s of concern—whether ferrous, nonferrous, or stainless steel. “People want to really dial it in to specifically what they want to find,” Lamprey says.
Metal detection foiled
An evolution around fresh, shelf-stable, ready-to-go foods has driven a need for single-use foil packaging, Lamprey notes. This poses challenges for metal detection. “Frequency selectability is really important for foil film,” he says. “And then if you really want to make sure that you’re maximizing detection, you move that foil packaging into the X-ray technology. That’s one of the biggest changes we’re seeing in the industry right now in packaging.”
There are other reasons X-ray technology might be needed over metal detectors—namely, contaminants in the food stream that aren’t made of metal, including rubber, plastic, stones, glass, and more. “They often find golf balls in a potato harvest because people go out in a field and hit golf balls,” Lamprey says. Chicken and fish bones can also present challenges. “The issue with that level of technology is the density of the product. So you need an X-ray that has the capability of picking those things up as they move down through the process.”
Though X-ray is becoming part of the food safety strategy for food manufacturers, the price point and increased complexity have made adoption slow. “The technology is five to 10 to 15 times the cost of a metal detector,” Lamprey says, noting not only the need for lead curtains at the system’s entrance and exit, but also the complexity around detection and imaging algorithms. “It’s really the capability and the science behind how you’re creating images and how you’re differentiating between density in things like chicken with bones in it and being able to show that image differentiation.”
Holistic process
Various inspection technologies have their benefits and limitations. A metal detector won’t find glass. And an X-ray system doesn’t work in gravity-fed applications.
Inspection requires a holistic process—balancing the needs with the technologies available, as well as mitigation techniques. “It’s important for companies to understand the potential risks and hazards within their products in their process and also understand the capabilities and limitations of systems as to what they might implement to help prevent or deter that from entering the product,” Rogers says.
Inspection tools won’t eliminate the problems. “They’re going to notify you when the problem is there, and it’s the procedures and processes that you build around the technology that makes a big difference in preventing any future contamination concerns,” Rogers emphasizes. “Ideally, my systems never go off. That would be the goal. If every system that I put into a facility never goes off, that means everything is perfect, right?”
PACK EXPO Connects – November 9-13. Now more than ever, packaging and processing professionals need solutions for a rapidly changing world, and the power of the PACK EXPO brand delivers the decision makers you need to reach. Attendee registration is open now.
Special Report: Essential tools for effective sanitation
Today’s food processors are faced with an ongoing labor shortage and pressure to increase production to meet market demands. That means less downtime for cleaning while adhering to strict food safety guidelines. How can a manufacturer overcome these hurdles to stay profitable?
Back to Basics: Understanding Conveyors for Food Processing
Discover how modern conveyors enhance food processing—boost efficiency, ensure safety, and save space. Explore the latest tech and tips to optimize your operations.