Ninety-three food engineers from 45 food-manufacturing companies and 29 partner engineering firms convened Feb. 4-7 at Fort Lauderdale, Fla. for Food Automation 2001, Food Engineering's second annual conference addressing automation issues in food manufacturing. An additional 17 vendor representatives demonstrated hardware and software solutions for food-manufacturing applications at the concurrent Automation Expo. Attendees toured the National Deli, Inc. plant, a unit of ConAgra, where plant manager Steve Conrad discussed his plan for tripling plant capacity in existing space at no capital cost by re-engineering automated equipment.
Engineers and operations executives from food-manufacturing companies, together at some sessions with their systems integrators, shared experiences in automating their processes or integrating the plant floor with information systems. Conference highlights follow.
Integrating the enterpriseWilliam R. Friend, former vice-president of the J.R. Simplot Food Group and principal of the consulting firm WR Friend & Associates, keynoted the conference by addressing the need to integrate manufacturing with the enterprise.
"ECR (Efficient Consumer Response) is dead!" Friend declared. Reasons: complex, fragmented IT architectures at both manufacturer and customer levels; consolidation of manufacturing and retailing companies; government regulations requiring trace and recall procedures; the growing complexity of business relationships, including co-packing and third-party distribution; increasing demand for customer-specific standards in both retail and foodservice.
These factors plus costs are driving food manufacturers and retailers toward collaborative forecasting and planning with standardized supply-chain information. "Competition is developing between supply chains rather than between customers," Friend observed. Enterprise Application Integration (EAI) software such as Microsoft BizTalk, Tibco and WebMehods are emerging to integrate both the enterprise itself and B2B transactions.
To integrate the plant floor with the supply chain, Friend advised, manage systems integration as a "change process" and:
- Stick with the same operating systems and databases where possible. Equipment and software vendor selection is critical. "Don't buy integration problems. If you can't be consistent, be compatible."
- Develop an integration strategy for plant systems. Consider data-warehousing for interfaces. Web-enable legacy systems.
- Develop the capability to import ingredient information from suppliers and forecasting data from customers, and to export production information to customers and co-packers.
New business models based on the web are changing the rules of competition, Friend continued. Leadership must therefore come from the top, with cross-functional teams incorporating IT, purchasing, manufacturing and marketing to execute the strategy. Management must build on investments already made, demand ROI from e-business projects, and "take the downturn in the [e-commerce] hype cycle as an opportunity."
Justifying & managing the projectDavid Watson, director of engineering at Pepperidge Farm, addressed justifying the automation project with case studies of robotic applications at Pepperidge Farm bakery plants. The economic justification for robots was based on estimates for direct labor replacement; productivity and efficiency; ergonomics and reduced worker-compensation claims.
Robots were introduced by the U.S. automotive industry during the 1970s to reduce labor costs, Watson observed. When ergonomics emerged as a science during the '80s, robots began to protect workers from repetitive-motion injuries. As labor grew short during the '90s, robots started substituting for people unwilling to perform repetitive tasks. Repetitive-stress injuries in the U.S. soared by 1,400 percent between 1982 and 1994, but declined during the late '90s as companies focused on eliminating or reducing worker exposure to repetitive motion.
Watson reviewed various methods of justifying a project:
- Return on investment (ROI), a simple percentage;
- Cash payback (CPB), typically two to four years in the baking industry;
- Internal rate of return (IRR), which measures the average rate of return the investment is expected to yield over its life, assuming reinvestment of cash flow at the project's IRR;
- Economic profit (EP), which measures residual income after subtracting all capital costs.
Watson cited several examples of how these criteria were used to justify robot projects. Among them:
- A robotic cookie-sandwiching machine installed in 1998 at a capital cost of $2 million resulted in annual savings of $1.3 million for an IRR of 41.3 percent and payback in 1.5 years;
- A robotic cookie-cupping machine installed early in 2000 at a capital cost of $3.2 million achieved annual savings of $1.2 million for an IRR of 27.8 percent and payback estimated at 3.26 years.
Alan Milton, project manager for Unilever's Loders Croklaan specialty fats & oils plant at Channahon, Ill., described how his team managed and executed a nine-month project to replace process controls in the 30 year-old plant (acquired from Durkee in 1998) while maintaining production during changeover (FE Nov., '00). Instructed by Unilever to achieve Y2K readiness, Loders Croklaan recognized an opportunity to unify its controls for both continuous and batch processing and sharing collected data with its sister plant in The Netherlands. "Upgrade was not an option; we had to build a new IT system," said Milton. Operators and maintenance people were part of the team. The team first defined project requirements, including a detailed description of processes, computer system functionality, process and instrumentation diagrams (PIDs) and database needs. After surveying potential suppliers, the team partnered with Honeywell to execute the project. Software for batch and continuous systems was completely rewritten by Loders-Croklaan and Honeywell engineers, with software implementation executed in India. The integrated, distributed solution consists of five high-performance process managers (HPMs), six GUS (graphical user stations, or HMIs), and advanced application nodes for automating batch processes and storing data, all integrated via a new Ethernet LAN. Factors which helped accelerate the project and continue production during changeover: Factory-acceptance tests conducted at Honeywell's facility in Cincinnati before installation; GUS stations aligned side-by-side with existing operator stations before cutover; I/O cutovers for each process unit scheduled and sequenced over a two-week period; software modifications to the old system to keep it running during cutover validation. The team commissioned the new system in three weeks while keeping pace with the plant's 24-hour/7-day per week production schedule.
Project case-historiesPeter Fabbri, plant manager for Kohler Mix Specialties Co. (a unit of Michael Foods) and Philip Frechette, president of system integrator JCS Controls (Rochester, NY), presented the flexible batching system for ice-cream mixes retrofitted by JCS into the Kohler plant at Newington, CT, a former H. P. Hood facility. Software customized by JCS improves upon the S88 batch standard by flow-charting recipes open to various vendor platforms including Rockwell, Intellution and Wonderware. The Kohler system is based on a Rockwell RSView32 server integrated via Ethernet with a PLC-5 and MS SQL database. The S88 Master Recipe resides in the server, the control (executable) recipe in the PLC.
John R. Miller, president of Seiberling Associates (Roscoe, IL), reviewed the automation upgrade installed as part of the total renovation and expansion of the Parmalat USA fluid-milk plant in Atlanta. The seven-month project doubled plant capacity without disrupting production, and includes a new HTST pasteurization system, three new fillers and a new CIP system, all centrally controlled by an Allen-Bradley RSView HMI integrated with a PLC 5/80 and PanelView operator-interfaces. The system controls receiving, batching, pasteurization, and CIP, logs the data and prints reports.
R. Tim Gellner, senior engineer at GE Automation Services (Pensacola, FL) addressed issues involved in planning an expandable automation project at the Kroger Co. bakery plant in Memphis, since acquired by Flowers Industries. Project scope included control of flour-handling, brew batching and automatic dough batching processes; ingredient-usage and yield reports; and integration with third-party scheduling software. Software was developed using Wonderware, Rockwell RSLogix and RS Linx and an SQL server on a Windows NT platform. The process is controlled by an Allen-Bradley PLC 5 integrated with IBM PCs housed in stainless-steel enclosures as HMI interfaces. Because Flowers plans further upgrades, planning for system flexibility was critical from the start.
Steve Kellogg, project manager at Pretzels, Inc., described the fast-track automation project for replacing process control of 23 packaging lines installed at the company's new $15-million plant in Bluffton, Ind. following destruction of the original plant by fire on Christmas Eve, 1997 (FE Sept., '99). Assisted by Shambaugh & Son (Fort Wayne) in a crash design/build project, Pretzels, Inc. started-up partial production just six months later and the new plant was in full production by Oct. '98 -- on-time and under budget. Recent secondary-packaging improvements include a new case-conveying and palletizing system which handles up to 1,500 cases per hour.
Multi-use SPCBob Hitomi, senior industrial engineer for Blue Diamond Growers, described how Blue Diamond implemented a Northwest Analytical Quality Analyst statistical quality control (SPC) package (FE Apr., '2000). Operators use statistical tools not only for continuous process improvement but to analyze data for meeting customer requirements, since nut crops can vary seasonally. For process improvement, operators use SPC to identify critical process parameters; identify the impact on that parameter by each step of the process; eliminate non-value added steps; and correct those steps which are not performing to expectations or capabilities. SPC is being applied to reducing in-shell by-products, standardizing raw-material moisture in low-moisture blanched items, and to improving blanching efficiencies such as skin removal, said Hitomi. One example: SPC identified "poorshell" as 70 percent of by-products generated, identified the electronic sorting machines as the major source "poorshell," then tested and compared the existing sortation machines with new technology. This resulted in project approval to install new sorting machines for the 2000 season. Results: poorshell generation reduced by 25 percent, good inshell loss reduced by 50 percent. SPC is "a work in progress," Hitomi added. The software is also being used to control-chart production specifications such as yields, throughputs and productivity; and to automate data collection through the PLCs, using Wonderware software and PDAs (personal digital assistants) instead of PCs to collect the data.
MES provides quick paybackPatrick Garland, director of information systems at frozen-food processor Rosina Food Products (Buffalo, NY) and Steven W. Glanell, vice-president of client services at system integrator Interwave Technology (Exton, PA), described how Rosina installed an MES (manufacturing execution system) to integrate the plant floor with the company's ERP system. "ERP has never been designed to integrate with the plant floor," Glanell pointed out. "MES did for our manufacturing plant what Microsoft Office did for the front office," Garland added.
To execute the project, Rosina first documented its original process, defined project scope, set reasonable cost/benefit and ROI expectations, and involved plant associates. "I/S took a back seat" in managing the project, said Garland "It was a plant project, not an IS project." Interwave defined project phases and deliverables, installed a Wonderware InTrack MES system with ERP interface customized by Interwave and the Rosina I/S staff, and developed a system workflow to match Rosina's production process. Touch screens, "ruggedized" for washdown on the plant floor, minimized training and "computer shock." The entire project took about 18 months from start to "go live." The system captures data from equipment (such as PLCs) on the SQL database; offers operator options based on results; tracks material consumption from raw material through work-in-progress to finished inventory; documents HACCP and integrates plant-floor data to ERP. It tracks scheduled versus actual run rates, work-center productivity including asset utilization and downtime; and displays key process indicators (KPIs) such as temperatures, hold times, test results, labor and asset utilization. Results: ROI payback of 14 months; production costs reduced by $1 million; 2 percent reduction in yield loss. Overall, "we achieved one extra day per week production," said Garland.
E-manufacturing boon to small firmsMost businesses have already committed sales, marketing, financial controls and order processing to a controlled Intranet within a company or group of companies and suppliers. "There is no reason why this capability would not be moved to the Internet," said Edward J. Bauer, vice-president of PMMI's packexpo.com and former director of packaging technology at Campbell Soup Co.
Electronic Data Interchange systems have reduced the cost of a paper transaction from $70 to $1, but EDI networks link only one supplier to one retailer and are very expensive to maintain, giving large companies such as Wal-Mart and Kroger a competitive advantage. But the Internet -- with it's ability to link many suppliers with many customers -- "will pull small and medium-size companies into the fulfillment equation?nd be the next major contributor to reducing costs," said Bauer.
On the supply side, e-procurement will generate 15 to 20 percent of these savings, "not the huge savings everyone has touted" because current commodity overcapacities will eventually balance out, making it harder for buyers to force prices down. B2B savings on the demand side, however, represent cost savings totaling 65 percent in three areas: e-supply chain management (outbound freight and logistics); e-customer resource management (sales and marketing expenses); and e-customer fulfillment - the biggest opportunity of all. This will require manufacturer and retailer to share performance measurements; to plan inventory requirements as early as possible to maximize lead times; to develop and continually update long-range forecasting models; and to deliver the product on time for the duration of the need with no out-of-stocks. "Those of you managing plants which produce a large number of SKUs realize how important each of these measurements are," said Bauer. E-manufacturing is about managing plants on more of a 'pull' basis in response to consumer demand, not the 'push' model to sustain inventories, and will eliminate non-value-added steps in sales, marketing, logistics and transportation, he predicted.
Snapshots & moviesThe topic "modeling and simulating your process" might be subtitled "diving for dollars" observed Malcolm Beaverstock, manager of advanced control and simulation at General Mills. "We've attributed more than $5 million in savings to modeling and simulation over the last four years -- and we're just starting!"
Modeling offers "a snapshot" of a single piece of equipment or a unit operation, while a simulation shows a "movie" of dynamic, interactive operations and manufacturing material flow, said Beaverstock. Modeling and simulation provide value through gaining a basic understanding of the process; as a basis for making strategic decisions such as the project feasiblity; for new-process design, checkout, troubleshooting and training ; and for operational improvement. Cost avoidance contributes about 50 percent of the savings.
The scope of a modeling/simulation project can be a single machine, a processing and/or packaging production line; an entire plant; or a supply chain "to determine its impact beyond the plant," said Beaverstock. The simulation database must consider: materials, including their physical properties; equipment, including capacities, rates, personnel requirements and changeover times; changeover by product and package size; inventory targets; schedules, shifts and performance measures. Extend, a simulation package used by GM's Betty Crocker plant in Toledo, replicates supply-chain interactions within 5 percent accuracy. Other simulation packages used by General Mills include Imagine That and SDI Industry Pro for simulating processing and packaging lines. In summary, said Beaverstock, simulation demands process understanding, encourages innovation, reduces risk and increases communication. Adopting a U.S. Air Force slogan, "it's better to sweat in simulation than bleed in war!" he declared.
Automated maintenance cuts costsRobert G. White, president of software supplier Bakery Innovative Technology (BIT) Corporation, described an automated maintenance diagnostics system which integrates process control, management tools, preventive maintenance, "maintenance-on-the-run," predictive failure analysis and specialized reporting into a real-time management-information system which puts "the entire plant at your fingertips." The system displays the status of every piece of plant equipment; presents specific audio/visual alarms allowing the operator to zoom-in quickly on the fault; presents diagnostic information on the failure; and presents key performance indicators such as production, downtime and failure reports.
An ROI analysis was conducted in several plants where this system is installed, White reported. On a breadline producing 7,200-7,800 loaves per hour, for example, maintenance savings per downtime hour were $1,700. Since a typical breadline is down 80 to 100 hours per year, White concluded, savings would equal $136,000 to $170,000 per year.
Predictive retort controlModel predictive control of a retort process was described by Jeff Cleek, assistant vice-president for technology at the Ameriqual Group (Evansville, Ind.) and Richard D. Eleew, technical applications manager for FMC FoodTech (Madera, Cal.).
Ameriqual, a co-packer supplying prepared foods to several major food manufacturers as well as MRE (Meal, Ready-to-Eat) combat rations to the U.S. military, packages and processes foods in retortable plastic Omnibowls, retort pouches and glass jars. For the past two years, Ameriqual has applied FMC's NumeriCAL On-Line predictive modeling software to control processing of up to120 loads per day in its 12 retorts.
NumeriCAL, which runs on FMC's LOG-TEC controller, corrects process retort deviations on-line in real time, adjusting time or temperature to compensate for the temperature-deviation effect on microbiological lethality in the slowest heating zone and/or the fastest cooling zone in the retort. Accepted by FDA and USDA, the system tracks lethality through every stage of the retort process and immediately documents the process. The model applies maximum thermal credit for come-up and cooling times in calculating Fo value to reduce both heating and cooling times.
Prior to applying NumeriCAL, Ameriqual based its retort model on the widely-used Ball formula, or general method. Although the Ball method is much more conservative and less flexible, said Cleek, "when we compared NumeriCAL to the general method we were surprised at how closely the two approximate each other."
The NumeriCAL method delivers a uniform product and minimizes worker supervision, human error and overprocessing, Cleek reported. It accurately utilizes come-up and cooling profile data while optimizing product quality, throughout and energy costs. Like the Internet, said Cleek, "once you have on-line correction, you wonder how you ever lived without it."