While control engineers in the food industry have always been concerned about the nuances of control system theory and how it can be applied to their latest brew, cooking oil process or packaging line, their management has come to expect a lot more of today’s process control systems. Managers want more control of their entire business. According to Kevin Zaba, director of processor initiative at Rockwell Automation, food processors are in an increasingly competitive business where companies struggle to stay at the front of the pack in production efficiency and market share. Different factors play into this challenge. As energy and raw material costs increase, food manufacturers are faced with keeping their production costs down. This is particularly challenging in an industry where consumer demands create short product life cycles.
On top of changing demands, as food manufacturing moves from local production facilities to global-scale manufacturing, product consistency is very important. With manufacturing shifting from a ship-to-order to make-to-ship business, ingredients and finished products no longer sit until needed. Rather, they are used on an as-needed basis, ensuring nothing sits unused in the warehouse. Zaba says supply chains today are demanding that manufacturers be able to switch their processes quickly and efficiently when necessary.
According to Steve Ryan, director of Proficy process solutions at GE Fanuc Automation, “Today’s process control systems should supply visibility into the production process and be able to optimize performance.” In addition, genealogy tracking and compliance with regulations like HACCP should also be achievable through the process system. Standards-based solutions and open technologies allow OEMs and end users to have a consistent communication architecture.
Standards and open technologies can help, but they’re often slow to be accepted and implemented. According to Rick Van Dyke, Frito-Lay group manager controls & MES systems and OMAC OPW (open, modular architecture control/packaging workroup) executive chairman, an OMAC Packaging Workgroup automation survey found that more than 60% of packaging lines have not yet been networked. More than two-thirds of packaging lines have no reliability data collection, and only 7% of packaging lines are integrated into ERP systems. Not exactly good news for networked process control.
Not surprisingly, connectivity gets the short shrift when funds are low. According to Matt Ruth, director of the food and beverage group at Advanced Automation, documentation, training and connectivity are cut when client projects are low on funds. With the decreased number of personnel in a plant, Ruth says companies realize that training and documentation are more important than ever, so therefore, connectivity gets left behind. With limited connectivity among systems, processors can lose sight of important process details.
According to Ruth, a processor’s biggest business return occurs when all systems work together. While the art of process control has been mastered in the last 30 years, vital business factors such as overall equipment effectiveness (OEE), uptime and constraints on the process and bottlenecks get lost in the overall noise of operations, i.e., making product. Today, the focus is on the communications connecting the brew house, bottler, case packer, labeler and the office. Process control is no longer limited to the distributed control system (DCS) in the brew house.
The changing definition or role of process control is clear. “Process control systems allow a company to more accurately and efficiently execute its business,” says Patrick Pilz, president/CEO of CSB-System International. “The accuracy improvements lead to quality improvements and the efficiency gains to cost reductions.”
According to Claus Abildgren, Wonderware marketing program manager, today’s process control is all about having an aggregate view of all the enterprise’s resources and showing how to get the most out of them. It’s as much about the business processes of manufacturing as the physical processes themselves.
A good example is Pernod’s new aperitif plant in Thuir, France, which bottles Suze. According to Jérôme Leroux, production director at Pernod, “Due to business demands, the new production and performance management system had to be up and running in only nine months.” Pernod’s daily transactions range from receiving raw materials to developing a finished product that can be used to create Suze and other aperitifs. “The new [Wonderware] system has improved the plant’s profitability and capacity, achieving a consistently high product quality,” Leroux adds. “Based on the plant intelligence provided by the industrial application server, we can analyze all relevant process parameters for sustained high performance.”
Still using old tools?
There are still some processors using old-fashioned tools. According to Gavin Clements, LINK business manager at FMC FoodTech, many processors are indeed using static, standalone tools such as Microsoft Excel as a method of recording both product and process data from production lines. These methods tend to be labor intensive, as they require data to be captured via paper on the production floor and then entered into Excel back in the office. This means any changes to improve a process can only be historical, so when referring to QA and food safety, these may be too late.The good news, according to Rudy Westervelt, president of Power in Learning, is that large vendors (Siemens, Rockwell Automation, Invensys, etc.) now have standard platform architectures on which to build “holistic” systems. These systems take information from plant floor controls, receiving, batching, processing, finance, distribution, etc. and coordinate information needed at all levels of an organization. However, he warns, “Processors should have a three-, five- and ten-year vision for their companies. They should have a financial investment plan for new technology and process controls.”
Connection benefits
A few years ago the DCS was a key player in process operations such as brewing or distilling. These systems used proprietary hardware and networking. Programmable logic controllers (PLCs), typically used in packaging machines, were also fairly proprietary both in hardware and networking, and they didn’t communicate directly with a DCS. The answer to this dilemma, according to Wonderware’s VP of Global Marketing, Mark Davidson, is in process control system modularity and connectivity to larger ERP systems when necessary.Zaba says there has been a tidal wave of support to the end user market to make control systems scalable, easier to maintain and more accommodating to changes -ranging from small recipe adjustments to new government regulations. He suggests processors migrate to a single control system that bridges the divide between process and discrete applications.
Even though there is a trend to get these systems more directly communicating, say through Ethernet TCP/IP networks, top layer protocols still may not allow these devices to freely communicate. But according to Mark Kettunen, project engineering director at Cadbury Schweppes Americas Beverages, the communication of key words is often enough to allow necessary communication between a PLC and another plant floor device such as a conveyor or bottling system.
According to Greg Bodenhamer, Schneider Electric manager, engineered solutions and services, Ethernet is letting engineers mix and match equipment that actually shares data and connects with quality and enterprise systems. “Today’s database and software standards such as OPC (www.opcfoundation.org),” says Bodenhamer, “give users tremendous capability to use common naming and tagging conventions for control systems and I/O as well as allowing shared data sets for alarms, and fault logging throughout the plant. Current Ethernet backbones allow food processing plants to take advantage of this ‘shop floor-to-top floor’ capability through the use of even low-end PLC equipment that has built-in Ethernet capability.”
An open plant network that several vendors support allows processors to pick and choose instrumentation. Campina Dairy in Aalter, Belgium, installed a Profibus DP network for discrete automation. Later, it installed Profibus PA technology for process automation. These fieldbus technologies provided a tightly-coupled solution using a single cable network for process and discrete control. According to Campina Automation Engineer, Paul Jongbloet, “Profibus DP and PA work well together, and that suits us because we have a mix of classic process measurements to make (pressure, temperature, flow and level) alongside many discrete functions, such as pumping operations and a diversity of on/off valve manifolds.”
The Campina plant network provides two other benefits. It provides a path for data from SCADA systems on the plant floor to the enterprise SAP system. In addition, says Jongbloet, the fieldbus system makes it easy to check parameters remotely such as auto-zeros and maintain the instruments. The use of multivariable transmitters and mass flowmeters provides a lot of useful information. High and low values can be monitored, so information about pressure or temperature shocks is available, too.
Too much data?
Once processors discover that data can be acquired, they may add sensors and transmitters to the system in an indiscriminate manner. This, according to food quality and safety consultant and Clemson Professor Emeritus John Surak, can lead to what he calls, “rich in data but extremely poor in information and knowledge.” To get knowledge, Surak says processors must combine technical information with process control data.”Rudy Westervelt often sees the same issue with his clients. “Processors have an over-abundance of data. They need assistance in turning this into useful ‘actionable’ information that has the intelligence to notify them of events that occur to evaluate the health of the operation.” This information, he adds, must be exactly what is needed at the plant floor, quality lab, maintenance department, finance department, corporate office and, in some cases, at off-site vendor locations via Websites, emails, etc.
Surak says processors can use this data intelligently by running it through a statistical process control (SPC) system to check for give-aways in product weight or volume, check for proper sanitation levels in CIP systems, test for E-coli levels, or possibly keep tabs on the alcoholic content of beer or wine. But Kettunen suggests that while many processors use SPC forensically to track, find and fix problems, SPC can also be used to evaluate what happens when a process is working perfectly. Armed with this information, engineers can stop a problem before it occurs.
Food processors who are looking to implement SPC systems should do a little homework. Look for systems than can accept information from multiple sources, including HMI screen input, Westervelt suggests. Pull information directly from scales while inputting data about the product or connect with a bar code reader to give product information tied to weight and lab results. Other suggestions: look for systems that can easily accept new parameters without major programming needs or, he adds, look for SPC programs that can export data directly into Excel.
One means of keeping costs low for SPC implementation is to begin with a small network with a central server and Web-based clients. A small tomato producer implemented an Iconics system to put an end to give-aways on overweight tomato packages. According to Iconics CEO, Russ Agrusa, the producer was afraid of being a tad underweight and being fined by the FDA, so it overcompensated by adding too much product. To keep costs low, the hydroponic tomato producer installed a single TrendWorX32 package on a server and ran thin-client Web browsers on five other machines connected to the weighing system. The single-server/Web-client approach significantly cuts down the costs compared to licensing full client packages on the HMIs, and standard technology (OPC) connects the weighing system to the server.
It’s probably safe to say that there is a technology available to solve most any process control problem. To be competitive, food manufacturers need knowledge about their processes-from plant floor to supply chain-to make informed decisions. While process control systems can collect any amount of data that’s out there, filtering out the critical information a processor needs takes the careful planning of not only the plant engineering staff, but also the IT staff, operations and management.
For more information:
Kevin Zaba, Rockwell Automation, 414-646-3832,kjzaba@ra.rockwell.com
Steve Ryan, GE Fanuc Automation, 800-433-2682,stephen.ryan@ge.com
Matt Ruth, Advanced Automation, 610-458-8700 (ext. 274),mruth@advancedautomation.com
Patrick Pilz, CSB-System Intl., 619-640-0436 (ext. 203),pilz@csb.com
Claus Abildgren, Wonderware, 949-727-3200,claus.abildgren@ wonderware.com
Gavin Clements, FMC FoodTech, 425-867-6729,gavin.clements@fmcti.com
Rudy Westervelt, Power in Learning, 909-563-8700,rudy@powerinlearning.com
Mark Davidson, Wonderware, 508-549-6433,mark.davidson@ips.invensys.com
Greg Bodenhamer, Schneider Electric, 919-855-1116,greg.bodenhamer@us.schneider-electric.com
John Surak, 864-506-2190,jgsurak@yahoo.com
Russ Agrusa, Iconics 508-543-4600,russ@iconics.com
Sidebar 1: Wireless, the hot topic
Wireless technology promises to extend the usefulness of plant information systems at the sensor level and the user level. For example, imagine a tank, a silo or even a QA lab across the railroad tracks at a large plant. Wires can’t be easily run, and if they could, it would cost a small fortune in union labor to establish a wired connection. Now, consider a wireless-based level transmitter on the tank or silo, or a wireless gateway connecting the QA lab to the rest of the plant. Wireless transmitters offered by the major process suppliers (ABB, Emerson Process, Honeywell, Invensys, Yokogawa, etc.) were shown in abundance at the recent ARC Strategies Forum. On average, it appears that the addition of wireless to already existing, popular transmitters may be between $500 and $1,000, but the added cost may be a wash when compared to installing wiring.In addition, wireless sensor-based transmitters that are capable of working with mesh technology can operate over an extended range as mesh routers work similarly to the older concept of a radio repeater base station. In this case data is passed around several devices as needed to insure both coverage and integrity until it arrives at the final gateway, which passes the data into the wired system.
Another useful application for the wireless sensor is on machines where the flexing of wires leads to short life and frequent replacement, and the use of infrared is impossible due to lack of line-of-sight. Wireless holds potential benefits, but it will take some time before it wins the hearts and minds of engineers. Kettunen isn’t convinced that wireless devices aren’t totally immune to interference caused by large motors, switchgear and other wireless devices. Today’s wireless systems don’t seem to have the security issues that their predecessors had a few years ago. But the one practical challenge, he says, is whether a plant has technicians who can troubleshoot both Ethernet and wireless technology.
Wireless systems are especially handy in remote troubleshooting and where a plant technician needs access close to a machine without a wired HMI. “The application where we see wireless as most effective is the pocket PC or PDA on the technician’s hip on the plant floor,” says Ruth. For example, the PDA has all the information associated with the process and is readily available to make decisions right at the machine. With SPC, the technician can use the PDA instead of paper and pencil. The value is getting the technology out closer to the people and the problems.