New directions

There was a time when almost any analysis of automation and control systems, whether for packaged machinery or end-to-end production lines, would at some point consider the pros and cons of PCs versus PLCs (programmable logic controllers). Latterly, however, arguments have largely coalesced around a general acceptance that PCs are stronger in terms of advanced programming, networking and heavyweight data crunching, whereas PLCs score on robustness and ease of use (think of the IEC 61131-3 languages, for example). But few care anymore. Most of the big automation and control suppliers now offer both PC and PLC functionality in factory formats right down at the backplane level, so why would you? It's much the same with that other old chestnut: PLC plus SCADA (supervisory control and data acquisition – historically favoured particularly by the utilities, food and beverage plants etc), versus the process industries' mainstay: DCSs (distributed control systems). The former grew up servicing the need for flexibility, integration with many device types from different manufacturers and remote access. DCS developers, on the other hand, focused on consistency, standardisation, a single database and configuration control on the fly – everything to keep a process plant running reliably 24/7. Again, the likes of Siemens and Rockwell, hitherto very much in the PLC/SCADA camp, have long since broadened their product and service offerings to include DCS-type capabilities, while Emerson, Honeywell etc have done the reverse. And both also have the software environments to match. That said, the debate appears now to be moving in some new and slightly surprising directions, given factories' understandable preoccupation with robustness, reliability and security. Latest thinking, for example, revolves around mobile access, distributed processing and even cloud-based systems. Robert Harrison, professor of automation systems at world-renowned Warwick Manufacturing Group, describes the impending new order as "cyber physical systems", harnessing ultra-distributed computing. Why the interest? Partly because of the sheer potential of exploiting automation intelligence – sharing information, for example, with electrical networks and BEMs (building management systems), and so redefining production optimisation. But also for its promise of new levels of granularity, enabling 'plug and play' with proven micro-control functions, and hence relatively easy plant reconfiguration in response to real world events and opportunities. "Clearly, there will be safety and security implications with processing power distributed around large networks," agrees Harrison, "but they're not insurmountable. And the potential to cut costs and improve operational flexibility by sharing real-time information between systems and people – including via mobile HMIs – is immense." He also makes the point that the more decentralised the control system, the less the risk of a single point of failure. And he should know. As head of Warwick's automation systems group, Harrison is an acknowledged expert in flexible control systems for the automotive industry and also has considerable expertise in sectors as diverse as packaging, aerospace and the water industry. In a sense, his view of future automation systems isn't wildly different from today's: We're all familiar with small distributed PLCs and other intelligent devices linked on networks. The issue is how far is it sensible to take that and, for him, current research is around going all the way to software component-based logic functions looking after, for example, individual actuators. Making this work is about visualising and then managing the physical and functional engineering complexity. That, in turn, requires virtual commissioning tools, designed to reconfigure functions on-demand, as well as remote monitoring tools for the run-time environment. "We're working with automotive assembly lines on two separate approaches," explains Harrison. "One concerns 'open' Ethernet systems that support different vendors' PLCs. The other is a longer-term project, using a SOA [service oriented architecture] system, with defined 'services' on an Ethernet or some other lower-level network. That could be configured, for example, to enable the integration of other devices, such as tablets and smart phones. It could also support 3D augmented reality systems to help maintenance engineers improve lifecycle support and ultimately also productivity." Making either of these approaches work as described, however, will require plant and factory PLC- and PC-based systems – which have historically been closed to the outside web world – to open up. And that, in turn, requires new standards, such as AutomationML (like the Internet's XML language, but for control systems). "We also need to be able to separate out safety-critical systems from the rest, before sharing information. It's plainly very important not to affect safety-critical embedded control behaviour," Harrison reiterates. But that's not a big deal: Segmenting networks with switches, firewalls and DMZs (demilitarised zones or perimeter networks) is, after all, standard stuff. Some of this is already happening. Jez Palmer, business development manager with Schneider Electric, points to Ethernet now embedded in his company's latest Modicon M580 PAC (programmable automation controller). That provides yet more impetus to the now well established trend of reducing, or even eliminating, traditional multi-layer plant and factory digital communications networks, each with their own bus protocols and technologies. By moving to just one, the existing barriers to data migration come down. Both systems and people can access information previously buried in intelligent, but effectively blind, devices way down a network hierarchy. This matters: Given lower level devices' growing intelligence, there are real gains to be had. And not only in terms of remotely viewing detailed plant operations and performance (including both OEE and energy) in real time, but also sensor, device and even process diagnostics. "Think of intelligent variable speed drives, motor protection relays and instrumentation and transmitters," explains Palmer. "Using today's standardised FDT [field device tool] software drivers and the common networking environment, we can plug in and integrate with these devices seamlessly to provide supervisory, maintenance and management systems with the information they need." With the advent of smart phones and tablets, anything less than being able to tap into some of that will increasingly raise eyebrows. Even more so when Industry 4.0 (the next level of intelligent factories in which all information systems, from engineering design to health and safety, exchange information) comes to fruition. And that's not necessarily so far off, at least according to protagonist Siemens. As Simon Keogh, Siemens business manager for automation, puts it: "We're already achieving some of this – broadening automation way beyond the traditional PLC layer – with our own software businesses. For example, we're integrating our product design software Teamcenter with the production environment so that engineering design and manufacturing can exchange information. Also, there are new modules around our process design software Comos that tie into our DCS system. So if you make a modification in the DCS, it flows through to process design." Keogh concedes that most of the UK isn't there yet, suggesting that a timeframe of five to 10 years might be reasonable. But it's coming. And when it does we may yet look back and wonder at the technically and, yes, also commercially driven, limitations of earlier automation systems and approaches. So perhaps we should look somewhat differently at our automation requirements and start thinking outside that box? Paul Herron, product manager for PC-based automation at Siemens, certainly thinks so. "A small PLC might solve your immediate control needs, but you might not be getting the whole benefit available now or at some time in the future. Think about how you might use data and systems to become more efficient. As factories become more sophisticated, or you want to achieve more, you're going to need to look around." His advice: "With any automation project, make sure you're thinking about putting in the infrastructure that will enable information to flow freely through the plant."