Alternate paths to control.
Control" on food processing plant floors is usually associated with electronic controls of automated equipment. But control of a broader, more analytical kind can be just as critical to plant profitability, as ConAgra Refrigerated Foods in Chicago discovered.
Using Quality Analyst software from Northwest Analytical, Portland, Ore., to analyze plant operations, the ConAgra plant sliced extra costs out of its consumer and foodservice meat-slicing operation.
Food processing has a natural disadvantage over many other forms of manufacturing when applying statistical process control (SPC), because the raw materials of food processing are rarely of uniform quality and size. But that doesn't mean that SPC can't make a significant difference, as ConAgra discovered.
USDA regulations require that the amount of meat in the package on average equal the amount printed on the label, but allow for variation in individual packages. That variation, however, can add up to either customer dissatisfaction or increased costs if it goes uncontrolled.
To get a handle on that variation, ConAgra Refrigerated Prepared Foods has used SPC on its meat-slicing line for several years.
The plant grinds meat to break down connective tissue, increase surface area and allow for better penetration of seasoning. The vacuum de-aerated meat is formed into circular logs and cooked to an internal temperature of 160 degrees, then cut into ready-to-serve slices.
ConAgra measured slice weight for each product and line once hourly, feeding the results into a PC. If the computer identified that more than two consecutive hourly check weights fell below the product's declared weight, then all products were held for the period since the last acceptable check weight for further inspection. That system ensured that average product weight met regulations, but it resulted in customers or consumers getting more than they were paying for, but never less.
ConAgra formed a corrective action team in 1997 of technical services, plant engineering, line operation and quality assurance personnel to address the problem. They ultimately determined that slicing equipment needed improvement to get product closer to specifications.
Tighter controls on slicing temperature and a more frequent schedule of blade maintenance greatly improved reliability. Quality Analyst had previously determined that 0.64 percent of the plant's products were overweight. After the improvements, a similar analysis showed that overweights were reduced to a microscopic incidence of one in 63 million portions, adding an extra slice of profit to plant operations.
Portion control remains the primary application for SPC software in the food industry, says Jeff Cawley, vice president of NWA. "We see a lot of uses related to fill weight, portion control or some kind of dispensing--such as monitoring the amount of chocolate spray coating inside of (ice cream cones)," he says. "The big focus is on trying to come into compliance with the maximum allowable variation or fill and trying to monitor that and give feedback to the line. There's obviously a lot of economic concern about getting that right."
Using SPC software to analyze line operations, either in offline or real-time mode, is a step short of a full-scale supervisory control and data acquisition (SCADA) system that can automatically make adjustments in process controls to offset variations.
SPC is "really servicing the function in processing where a SCADA system is too much and too cumbersome," Cawley says. "Setting up a SCADA system would often be very high in overhead and doesn't do an effective job of getting operators feedback in a small-scale testing application."
SPC works well in situations like ConAgra's, where variation can be controlled through periodic identification of non-critical problems and implementation of lasting solutions. SCADA systems are used more often to control retort and cooking lines, where raw material or other variations could cause product to be undercooked and lead to potentially more serious quality problems.
In some cases, NWA's on-line Quality Monitor program, combined with local alarm systems to notify operators of an out-of-spec condition, can provide nearly the same level of security. Cawley sees such a system most applicable for smaller operations with relatively few programmable logic controllers and relatively simple retort controls.
"It's a lot lower overhead to install, configure and maintain a work station like that as opposed to a larger distributed (SCADA) system," he says. "The paradigm of Quality Monitor is basically a plant floor work station or test station, as opposed to a SCADA system where you have multiple drop points, multiple control points and sensor points."
Modular control alternative
In another area of plant floor control, Key Technology, the Walla Walla, Wash., maker of shakers, distribution and other conveyor-belt equipment, has launched a new alternative in the PLC vs. PC-based control debate. Its Forte modular control system is a PC-based unit developed specifically to control Key conveyor systems.
"Typically conveyors have their own systems and power modules (for controls), and if you're running a plant with a whole bunch of conveyors, you can end up with a lot of different points of control," says Lisa Teske, marketing communications manager for Key Technology. "The idea behind Forte was to pull that all together for this equipment."
Forte runs in a Windows environment. The advantage over third-party-developed PC-based control is that Key has the knowledge to optimize running its own system, Teske says. Forte can also be interfaced with PLCs, external automation systems and management systems, she says.
"We do build our own control panels here and know what's inside them," Teske says. "When you hand the job over to an outside control company, there's a lot of confusion and room for mistakes."
Forte's modular approach depends on DeviceNet, which minimizes the long conduit runs for control wiring needed to link to a PLC, says project engineer Bill Davis.
The modular approach makes sense for manufacturers with at least 10 to 12 pieces of equipment they need to control, he says.