When the first issues of Control were published in the late 1980s, Ethernet had already begun its conquest of the enterprise networking space. But in the industrial world, concerns over determinism held it at bay. The higher expense of MAP/TOP and other token ring-based networks, together with relentless performance improvements to Ethernet technology resulted in its top-down conquest of the enterprise information space that ultimately included process control.
Meanwhile, at the instrumentation level, the process automation industry continued to haggle over its own network standard that would be compatible with the extensive installed base of 4-20mA twisted-pair copper wire used for analog process signals.
Indeed, the decade of the 1990s saw us industry reporters spill a great deal of ink over the “fieldbus wars,” in which rival camps of instrumentation vendors did all they could to control—or stymie—progress of the standards work. All that effort ended with the IEC approving not one, but eight different options—acknowledging those many various protocols that had been commercialized, but doing little to narrow end users’ dilemma.
Fast forward to 2008. Ethernet and its various derivatives have conquered all but that industrial last mile—that final stretch of connectivity to a given temperature transmitter or proximity switch. At this level, as an industry, we still deploy a variety of industry-proprietary protocols.
But the technical reasons for not having a single, unified network architecture from top to bottom are falling one by one. Security considerations notwithstanding, the day is coming when that desktop computer in the CEO’s corner office, as well as the distant pressure transmitter atop the distillation tower, exist as TCP/IP addresses on the same Ethernet architecture. Ethernet chipsets are now cheap enough to be embedded in the lowliest devices, even as present day device and fieldbus protocols are modified to run on Ethernet-compatible physical media, even simultaneously with other protocol stacks.
The process industries, of course, have come to expect loop-powered instrumentation, but continuing advances in power-over-Ethernet standards promise to overcome even this last limitation to delivering device power over the same wire as the digital Ethernet signal. Meanwhile, lower-powered devices themselves, developed in part to support wireless instrument networks, will further accelerate this continuing trend. We’re in the early stages of wireless adoption, and at times they seem a lot like the early days of fieldbus: Committees work toward a consensus standard continues even as vendors and early adopters demonstrate the technology’s game-changing potential. For process control applications, among the biggest issues are network integrity and reliability, for which self-organizing and self-healing mesh topologies appear the most viable approach for process manufacturing environments.
So, twenty years from now, what will be the networking state of the art in process manufacturing plants? Faster networks? Certainly. Increased reliance on commercial networking technologies? Undoubtedly. More wireless applications? Inevitably.
The best bet is that 2028 will look a lot more like 2008 than one might think. For example, a just-completed survey of Control readers indicated that a full 54% of survey respondents still use mostly 4-20mA analog hardwiring for instrument communications. Only 12% reported using primarily digital buses. So despite all the advances in digital field communication technologies over the years, more than half still rely on technology that matured some 40 years ago. By their very nature, disruptive advances are impossible to forecast, but from it’s straightforward to extrapolate a networking future that’s faster, more wireless and, with some help from the supplier community, increasingly easy to set up and manage.