Controls Networks and IT Networks Need to Work Together for Effective Supply-Chain Management
As supply-chain management casts an eye toward demand visibility and production performance, the inevitable collisions between IT and manufacturing network engineers become more pronounced. But two companies, Coca-Cola and Procter & Gamble (P&G), are turning those clashes and conflicts into convergence and cooperation.
Jeff Kent, technology section head, controls and information, at P&G (www.pg.com), and David Bynum, principal engineer from Coca-Cola North America (www.coca-cola.com), explained this past November at Rockwell Automation Fair in Anaheim, Calif., how their enormous companies have straddled the chasm and begun to tap valuable manufacturing data.
"Internal to P&G, we have a very strong manufacturing IT presence," said Kent. "As we begin to blur those roles, the architectures they've already defined will change. On the technology side, there's still a huge amount of struggle. In our internal culture, when the IT department wants data, they come and get it. We know how damaging that can be. In these very integrated high-effort systems, we need to reduce the effort. Whether we're in Africa or somewhere else, those systems have to have the same agility."
One key hurdle at P&G is having OPC as a way to connect to the Logix platform. "There were interventions made, and we backed them off the polling rates or number of tags, or we've explored other ways to mirror the data so they can access that chassis, rather than our real running controller," said Kent. "Making Logix a data server actually means creating mirrored data so we don't hit the controller with OPC calls. We need to mirror those tags over."
At Coca-Cola, it's all about being in tune with a demand-driven supply chain and striking the right balance between efficiency, service and cost, said Bynum. "How do we promote information beyond the first consumer out to other organizations within the division so they can also see what's coming?" he asked. "What information do we need so we can be more responsive to those demand signals? The Logix platform has pushed the envelope for us. That platform allows us to take all these pieces, bring it to a central platform and serve it back out to the users. Before Logix, it was still hammer-and-chisel to put those systems in place. Now we have conversations between Rockwell Automation and IT, not just manufacturing controls. Today we have the framework in-house, so both groups collaborate. We have IT and the shop-floor expertise to get the job done, and integration to ERP has given us some momentum. We don't allow IT to just come and take whatever information they want. Creating an information layer reduces latency on the actual controller and HMI at the shop-floor level."
James Ingraham, software development team leader at Sage Automation (www.sagerobot.com), Beaumont, Texas, says most of the system integrator's customers haven't been integrating machinery with the IT network. "The coupling point still seems to be barcodes on products or pallets as they come off the production line," he says. "Even when there are nice windows into the production environment, individual machines usually aren't part of that. We rarely even get assigned IP addresses for our Ethernet equipment. We have seen a few sophisticated systems, where IT has been involved and specified gateways and segregated networks and handed out IP addresses, even allowing remote VPN access." But Ingraham says it's the exception, rather than the rule.
"Our IT department is highly involved in projects concerning control and control systems," explains Choy-Hsien Lin, development engineer, process control, Stora Enso Publication Paper (www.storaenso.com), Hyltebruk, Sweden. "There is a growing need for Ethernet communication and other high-level protocols. This increases the number of attack vectors to the systems exposing them to significant security risks. Data from our plants are aggregated in several central systems, but the ERP is only connected to specific points where custody transfer occurs."
One typical case of conflict between IT and controls engineers is when a company has an IT department that desires to do remote management of PCs, explains Stuart McFarlane, vice president, Viewpoint Systems (www.viewpointusa.com), Rochester, N.Y. "Typical cases of IT control include when IT specifies which operating system to use," says McFarlane. "Sometimes, IT will specify computer make and model when clearly that make and model are not appropriate for the industrial application or when IT specifies remote management of OS updates, novel client updates and antivirus updates. This aspect is particularly problematic because the computers installed for an application have been verified to operate and the automatic updates disturb the configuration enough that the application no longer functions correctly."
Production data is not typically collected by ERP systems, says McFarlane. "This is still an area that sounds like a good idea but is very difficult to implement and manage," he says. "The exception to this rule is the medical industry that has implemented 21 CFR, Part 11, traceability. Since it has taken it that far, integration into ERP is a smaller step. IT and plant engineers do not typically play nicely together. The most successful companies' IT organizations realize they are there to support the business in the most effective and secure way possible and are part of the solution and not part of the problem."
[javascriptSnippet]
At Sun Chemical (www.sunchemical.com) in Muskegon, Mich., the controls network is separate from the IT network, explains Francis Lauryssens, PI systems specialist. "We collect data like pressure and temperature from the machinery," he says. "Data is available on the controls network for one week for the operators on the floor, on average, so they can trend their temperatures. But it also goes through our PI server, which has a foot in the controls network and one foot in the IT network world. The two networks are completely separate. If you want to see what's going on in the controls network, that's your gateway. It's in real-time because the controls network talks with the PLC, goes through an interface and talks with the PI server." The data on Sun Chemical's IT network is used for reports and daily energy calculations.
"Some IT departments couldn't care less about the plant-wide SCADA system," says Gregory C. Kempfer, controls engineer, M.C. Dean/CIM Automation Systems (www.mcdean.com), Harrisonburg, Va. "Others want to control every network in the plant, including networks whose primary function is for the control system. We basically follow our customer's lead and adapt our system approach to demands."
More and more of Tulsa Power's customers are collecting data from their process lines via Ethernet, says Joe Crosley, senior electrical engineer, Tulsa Power (www.tulsapower.com), Tulsa, Okla. "This necessitates an interaction between IT personnel and controls engineers," he says. "Most of our experience is usually in dealing engineer-to-engineer and rarely with the customers' IT departments though."
N. Lewis Bodden, a retired control system engineer in Texas, recalls one company with more than 100 remote sites that are monitored by a central control center, where a separate database collection point (ERP) had issues with the IT department. "Each remote site monitors between three and 36 units with about 20 parameters each," explains Bodden. "Each unit has its own HMI to monitor the operation, as well as up to four site HMIs for monitoring. Locally at each site, the control system is isolated from all other networks, except for a single router configured to allow access by selected groups. Initially, the IT department wanted to have control of the control network. They didn't think that it was necessary to have an independent segment for the control system. They maintained they could provide security at the site level and computers could be assigned IP addresses by a DNS."
Ultimately, the controls engineers changed the configuration with a separate control network segment and fixed IP addresses. "The IT department then put its best people on the configuration for about two weeks to make its configuration match ours. The two weeks they spent working closely with us in our lab gave them an appreciation of our problems, and it gave us a good glimpse into theirs," says Bodden.
"Internally, we have an ERP," says Tony Lovette, president/COO, Welsh Paper (www.welshpaper.com), Youngsville, N.C. "We don't find value in scanning in our particular product set. RFID is not used in our product portfolio. We do use other companies' fulfillment engines, as requested, to check stock, manage min/max and forecast updates/changes. Complexity and customer requirements drive most of our IT investments and initiatives."
