Though a lot has been written about the benefits of server virtualization, this article will address the impact that virtualization technology has on process controllers, I/O systems and process field devices—the parts of an automation system implemented as embedded systems.
An embedded system is traditionally defined as a dedicated function device containing electronic hardware, a microprocessor and specialized software. The relatively complex embedded systems found in process controllers use microcontrollers or full-fledged microprocessors. They are networked and may download their operating system and application software via network services.
With the development of multicore microprocessors, semiconductor suppliers can now support virtualization technology within a single microprocessor. In effect, the software stack of a server can now be replicated within an embedded system. Process automation, with its very long product lifecycle, stands to benefit from the broad adoption of embedded virtualization in other industries, such as automotive, aerospace and medical equipment. Process automation suppliers are now planning products that will use this technology.
ARC Advisory Group believes that embedded virtualization will have a major impact on our evolving collaborative process automation system (CPAS) vision.
Process controllers are perhaps the most complex and multi-functional components of any process automation system. To meet the extremely high requirements for reliability, they incorporate complex and proprietary means for hardware fault tolerance. Process controller software has always run on various real-time operating systems (RTOS). The next generation of process controllers will be able to exploit multicore processors and embedded virtualization. ARC believes these factors will factor significantly in the development of future CPAS controllers with embedded virtualization:
Higher performance. The additional compute capacity of multicore processors with virtualization could be employed simply to improve the capacity and/or performance metrics of a process controller.
Greater application integration. Controllers running both an RTOS and a rich OS will be able to host CPAS applications that presently can only be hosted in servers. Advanced process control and optimization are certainly applications that may fit into this configuration. Data historians and local analytics may fit as well.
System management. Process automation systems have traditionally provided their own (proprietary) system management capabilities. Given extra processing capacity and a rich OS, these could be incorporated into standards-based management protocols, which could aid in giving automation system status wider visibility in the enterprise.
Advanced networks. Several process automation suppliers have developed high-availability Ethernet networks at the controller (peer-to-peer) level. In next-gen controllers, these networks may extend to the I/O systems and perhaps even to field devices. Networks can also be virtualized, and silicon suppliers have developed technologies to do this at the embedded system level. Such virtualized network interfaces could be provisioned to support both real-time and rich OS network traffic.
New controller modularity options. Process controllers represent a combination of high complexity and low unit volume. System designers are challenged to optimize not only performance, but product lifecycle length, total cost of ownership, system bill of material costs, commonality of system hardware, and commonality of system software. We expect that in the long term, commonality and component life will weigh more heavily in this equation. This would favor new systems incorporating higher-performance processors and more common software.
Process automation I/O systems can share some of the same benefits from virtualization, but not to the same degree as controllers. The industry trend in I/O systems has been toward “smart” I/O modules that are configurable (via software or a hardware adapter) to accept multiple types of signals. This capability has proven very valuable in compressing automation project schedules by decoupling detail engineering from other project activities. Expanding smart I/O capabilities is essentially a “smart analog” technology development that will not be enhanced by virtualization.
Process field devices (transmitters, valves, drives, analyzers, etc.) have remained largely immune from virtualization benefits. This is mostly because of the age and limited capabilities of the network technologies serving process field devices (HART, Foundation fieldbus, Profibus PA). Adoption of Ethernet networking by process field devices will open up device communication to all manner of improvements and deeper integration, but adoption will be gradual. A field network consisting of a switched Ethernet infrastructure could remove many of the barriers and difficulties end users experience with managing thousands (or tens of thou-sands) of field devices.
Given the very long device product development and operating lifecycle, Ethernet field networks will emerge very gradually in the process industries.
>>Harry Forbes, senior analyst at ARC Advisory Group, leads coverage of distributed control systems and the electric power industry. He has 30 years of experience in automation, power generation, energy management, modeling and simulation, advanced control, and optimization. Prior to joining ARC, he served in a variety of marketing, sales and engineering posts, including work as a performance and automation engineer in fossil and nuclear power generation at Detroit Edison. He has a B.S. in electrical engineering from Tufts University, and an M.B.A. from the Ross School of Business at the University of Michigan.