Improving processes is the best way to boost quality, and statistical process control is showing the way.

A generation of engineers has grown up since statistical process control (SPC) became SOP in U.S. manufacturing.

SPC is the building block for total quality management, ISO standards, six sigma and a host of other methodologies to control variance and reduce waste in manufacturing processes. Food engineers have applied SPC in some fashion for a quarter of a century. Though its use is expanding, the number-crunching essence of SPC remains unchanged. "We keep adding more algorithms to meet more rigorous requirements," allows Jeffery L. Cawley, vice president at Northwest Analytical Inc. (NWA), "but the core functionality is the same."

The automotive industry was the first to apply statistical data to control processes, but many questioned the application in food, where inconsistent raw materials created variability nightmares. Food engineers overcame that roadblock, and the data mined from SPC is migrating to other users. Whether it's the line operator or the CEO, users of SPC data are gaining a better understanding and control of their company's products.

The same is true of customers. Part of the SPC evolution is delivering more user-friendly data to these data users. "A lot of our customers are asking us, 'How can we get histograms, run charts and other analytical data on what you're shipping on the Internet?'" reports a flour miller's head of quality, who requests anonymity. "Right now all they would get is a line with ingredient declarations. Wouldn't it be great to give them current averages for moisture content, protein, ash absorption and stability without forcing them to invest in the statistical training needed to read our data charts?"

Web browser access to data charts is a capability many software suppliers have added to their programs, among them NWA, GE Fanuc and Zontec (see related supplier's list). Windows-based operating systems are another development in recent years that has expanded data accessibility.

"We went through three generations of our product," beginning with a DOS-based application, says Warren Ha, Zontec's president. "Most people have moved to Windows, particularly in the last two years. Compatibility is the big advantage, and once you learn one application under Windows, the rest should be fairly self-explanatory. The user interface is much better, and you can run multiple tasks."

Rockwell Software originally offered the SPC module of its RSView32 HMI package as an add-on module, but today it's part of the core functionality. "That change was definitely customer driven," according to Darrell Walker, marketing manager. "Manufacturers insisted on the inclusion of the SPC function."

As SPC has become more mainstream, more bells and whistles have been added to the basic packages. E-mail messaging, remote monitoring and alarm systems help broaden the audience for the data, but there is the potential for undermining one potential benefit: operator empowerment.

"If we empower operators in the appropriate manner, do we really need pagers carried by managers to go off?" asks John G. Surak, a professor at Clemson University and an expert on food applications of SPC. "We're transforming these people to be able to monitor and analyze process data and take the appropriate action, including the decision not to make adjustments if a predictable cause for variation has occurred.

"You really want the operator to make the decision that a change has occurred that requires altering the process," Surak continues. "If the process is out of control, operators should go through a decision tree to determine the appropriate corrective action. If all else fails, then you call the manager."

Operator involvement is critical, concurs Syed Hussain, director of technical services at ConAgra Foods, particularly if processors are going to move SPC from a monitoring tool to a process improvement methodology. "Giving control to your operators not only makes them responsible and accountable, it motivates and increases their enthusiasm." Not coincidentally, it enhances company profitability (see related story).

New SPC applications

As their comfort level with SPC has increased, food processors have demonstrated a willingness to use it as an alternative to traditional inspection-oriented approaches to quality control. The stability of the process dictates the frequency of SPC measurement. If those measurements indicate product is within an acceptable range, the manufacturer can feel confident that consistent product meeting customer specifications is being shipped. That's a more appealing scenario than an inspection that uncovers unacceptable product already in the pipeline.

Quick service restaurants have been a driver in getting food processors to adopt SPC-oriented production. Zontec's Ha cites the client who produces frozen dough that will be baked off as biscuits at Kentucky Fried Chicken outlets. "If the size is not constantly monitored and controlled in the factory, the stores end up with a bad presentation to their customers, and that is unacceptable," he says. An inspection-based approach would likely result in rejected shipments and, ultimately, loss of the account.

While SPC is well accepted for controlling product weight, portioning, fill levels and other physical characteristics, proponents have been circumspect in advocating SPC's role in HACCP. That is beginning to change: the Agricultural Marketing Service (AMS) of USDA is going forward with a program that will essentially require suppliers of ground beef for school lunch programs to implement approved processes for pathogen control.

The policy shift began last summer, when the department performed lot-by-lot inspections of 116 million lbs. of beef delivered by 17 meat companies. About 8.6 million lbs. failed microbial tests for coliforms, Salmonella and other pathogens, including 1 million lbs. contaminated with E. coli 0157. The results underpin the department's decision to shift from the approval or rejection of finished goods to a process-based approach that will force suppliers "to be proactive up front," according to Barbara L. Cope, an administrator in the department's livestock and seed program.

"The main purpose is to get suppliers to implement and maintain effective HACCP systems," she explains. "We're going to be monitoring companies instead of finished goods."

A similar SPC-oriented approach to young animal slaughter is being eyed for the Food Safety and Inspection Service's HIMP program.

While AMS's new program should reduce the amount of "distressed meat entering the system," Cope doubts it will have much impact on overall beef supplies. SPC proponents take a different stand: suppliers who buy into the concept and change their inspection-based mindset will shift from a process-control to a process-improvement orientation. That will add value to their finished goods, and that value will be reflected in the company's bottom line.

"Technical people understand the value of SPC, but they need support and resources from management if they are going to expand its application," insists NWA's Cawley. "Management may want to scrimp on data collection, but data collection serves a vital role. It's minding the business in a meaningful way."

"Fact-based management needs SPC," agrees Surak. "It is the key to process improvement based on statistical thinking." The three tenets of statistical thinking are: all work is done in a process; all processes show variation; and the key to marketplace success is reduction or elimination of the causes of variation.

Better bells, shinier whistles

Collecting data no longer is an issue, suggests Craig Thorsland, visualization manager at GE Fanuc. "What companies really need is a system to help them manage that data better," he says. "We've taken the math behind SPC and added quality alarms that are not tied to a specific terminal. Once you make the transition from doing manual data collection, you free up the user's time to identify what's really important. SPC is just a tool for the analysis at the heart of six-sigma methodology."

In recent years, GE has adopted the six-sigma quality initiative popularized by Motorola. As a practical matter, six sigma (two defects per billion processes) is more realistic for discreet processes than food production. "We can achieve perfection in sanitation," believes ConAgra Foods' Hussain, "but for a meat product where there is so much variability, it is not realistic to go for even two sigma." His SPC-based quality goal: 1.3 sigma, or about 63 out-of-spec products per million pieces produced.

Real-time data monitoring by operators, engineers and upper management can help, believes Ha. And wireless is the next step in making that real-time data accessible, he says. "Wireless communication is the next direction the technology is taking," and his firm recently introduced a wireless control system for SPC monitoring.

"It's going to take time before companies go completely wireless, but there's a lot of interest," he adds. "There's tremendous potential in the food industry, where the washdown environment argues against PCs on the plant floor."

Regardless of what hardware and software a company selects, Surak says the key criteria for evaluating SPC systems are ease of use and ease of integration. "To me, database conductivity is the most important consideration," he says. "If the software looks like a proprietary database, you could be wed to that supplier forever."

"Highly manufactured products were early adopters of SPC, but now we're seeing it in raw commodities," concludes Cawley. "Food doesn't have to feel ashamed about where it is at."

The application of SPC in food manufacturing is spreading. Now companies are using the resulting data and their understanding of their critical process parameters to slash waste and enhance profitability.

A guide to SPC suppliers

Software is a volatile market, and programs for statistical process control are no exception. Several SPC suppliers cited two years ago in Food Engineering are no more. Among them: Bradley Ward Systems, a firm that specialized in the automation needs of food processors.

Fortunately, many excellent SPC packages remain. Process control has become so fundamental in manufacturing, SPC has been incorporated as a basic component in comprehensive automation systems. SPC packages include:

  • NWA Quality Analyst from Northwest Analytical Inc., Portland, Ore. Developed more than a decade ago, QA generates a wide variety of variable-control and attribute-control charts for process improvement.

  • from the Foxboro Co., Foxboro, Mass. Real-time data monitoring and alarm status notification on monitored charts are two touted features in this I/A series software.

  • Cimplicity OpenProcess from GE Fanuc Automation, Albany, N.Y. Process control component of GE's monitoring and control software offers open architecture.

  • RSView32 from Rockwell Software, West Allis, Wis. SPC module is part of the core functionality of Rockwell's HMI package for plant-floor information management.

  • Synergy 2000 from Zontec Inc., Cincinnati. LAN-based quality management package produces a wide range of calculations for control limits and process capability.

  • Minitab Release 13 from Minitab Inc., State College, Pa. General statistical package for data analysis and quality and process improvement.

  • SAS/QC from the SAS Institute, Cary, N.C. Powerful enterprise-wide application that generates Shewhart charts, Ishikawa diagrams and other control charts for reliability assessments.

Cost savings only one benefit of SPC

Weekly raw-material savings of $10,000 are a quantifiable benefit of a recent statistical process control project at ConAgra Refrigerated Prepared Foods, Downers Grove, Ill., to reduce weight variability in sliced meats, but the payoff from continuous improvement goes well beyond cost savings.

Greater employee involvement has produced additional process improvements, resulting in higher quality products, greater productivity and enhanced plant capabilities, according to Syed A. Hussain.

The SPC project involved fully cooked and sliced muscle meats sold in 12.5 oz. packages to restaurants. "During initial process startup, statistical analysis revealed that the process was stable but the variability made it incapable of meeting specifications," explains Hussain. He used SPC software from Northwest Analytical Inc. to help control those other specs, such as the number of slices of meat per package.

Hourly weight checks are performed to verify that the average weight meets or exceeds the declared net weight; two consecutive weight checks below that level result in a product hold. Consequently, "the process frequently exceeded the upper specification level but never fell below the lower specification level," Hussain says. "The problem was identified and a corrective action team formed."

Sharper slicing blades and tighter control of the chill process were among the changes recommended by the quality improvement team. Team members' involvement in the project also boosted their job satisfaction and productivity, he adds. Workers now regularly recommend process improvements that have a direct impact on corporate profitability, Hussain says.