Raw material variability is one of the distinguishing traits of food production, and animal-based products provide a prime example of that reality. As a result, the processes for beef, pork and poultry pose a particular challenge for automation.
The metric for judging automation traditionally has been labor savings: If a machine displaces enough manual work hours to generate an acceptable ROI, capital is made available. But the cost of workplace injuries also needs to be considered, causing meat and poultry processors to look at automation spending in a different light.
Spinal column removal from primal cuts poses particular danger to workers in a beef fabrication room, with workers’ fingers within inches of a band-saw blade as they guide it as close to the bone as possible to remove the least amount of meat. When Steve Dunivan began in-plant trials of his smart-rib butchering system, he received a particularly warm reception from the band-saw operators: “Almost every person can show you a scar,” he says.
Dunivan, owner of Midwest Machine LLC, Amarillo, Texas, has worked on automating the rib cut for four years. Early versions of the machine simply couldn’t deliver the precision the job requires. About 18 months ago, he began discussions with Hermary Opto Electronics Inc., a Coquitlam, B.C., supplier of 3D vision systems.
Hermary’s roots are in the lumber industry, explains Wes Henderson, director-business development, and its “sheet of light scanner” is hardened from years of industrial use in sawmills. By November 2013, Dunivan had written the coding to integrate the 3D system with his saw and began in-plant trials at Cargill’s Schuyler, Neb., facility. Since then, more than 1.2 million spinal cords have been precisely removed.
Hermary’s camera triangulates on an object, taking up to 1,000 scans per second. That’s more than the rib-saw system requires and would unnecessarily slow processing speed. Dunivan dialed down to 200 scans a second, enough to produce a geometric rendition to guide two saws cutting within a 1/16th in. tolerance. Anything outside that is considered a bad cut, and a 1 oz. loss in yield per piece could cost a high-volume plant $750,000 a year.
Dunivan's latest machine delivers good cuts 99 percent of the time, compared to 90 percent in the earliest versions and 70 percent with manual cuts.
Hermary is a Rockwell Automation partner, Henderson emphasizes, which simplifies data download via Ethernet/IP to MES or ERP programs for tracking and management reporting.
Enabling one operator to perform the work of two is a more tangible benefit, but enhanced precision and safety may be the greater value. “It’s an intricate, sophisticated cut, and it’s a rough job,” Dunivan reflects. “Handling 2,500 pieces a day is typical. What are your chances of having an accident?”
Machine vision also plays a critical role in a poultry system developed by Gainco Inc., though worker productivity, not safety, is the goal.
In this case, 3D imaging was an outgrowth of a robotic cutting system for chicken breast meat. The robot’s axes of motion never matched the performance of humans, but in the course of his work, John Daley, principal research engineer at Atlanta’s Georgia Tech Research Institute, discovered a way to calculate the amount of meat left on a chicken carcass, or frame, to determine yield loss.
GTRI licensed the technology to Gainco, which followed up with a prototype machine shown at 2014’s International Production and Processing Expo (IPPE) in Atlanta with a ready-for-primetime version at this year’s show.
By placing the frame on a translucent cone, Daley found that the intensity of LED lightwaves transmitted through it could be read by a near-infrared sensor to determine the amount of meat and bone present. Based on the grams of meat in five locations, a Yaeger Gore standardized score is computed of yield loss, according to John Daley (no relation), sales and marketing director at Gainco, Gainesville, Ga.
Poultry processors routinely conduct scrape tests to evaluate operator performance, Gainco’s Daley says, but it’s a manual process with a spoon-shaped knife and takes five minutes or more. Meat remaining on the frame will be recovered downstream in the MDM process, but its value is significantly lower than the white breast meat captured on the cone line.
He characterizes his rapid yield analyzer as an instructional tool, providing quick feedback to operators who exceed the target loss level and identifying where on the frame the most meat is being left.
A handful of processors field tested the prototype design and made recommendations before Gainco fabricated the washdown ready version with an IP69 enclosure displayed at IPPE this January. International visitors were unimpressed, but half the domestic poultry companies expressed interest in conducting trials, says Daley, a reflection of the high volumes and efficiency focus in U.S. poultry processing.
Bones of contention
Different segments of the electromagnetic spectrum come into play in end-of-line inspection. Whether they use metal detectors or X-ray units, meat and poultry processors must deal with the trade-offs between sensitivity and false positives, including overlapping products, observes Todd Grube, inspection product manager at Heat and Control Inc., Hayward, Calif.