As control system boundaries blur, engineers face growing network complexity
Key Highlights
- PLCs, DCSs, SCADA and safety systems share overlapping capabilities, architectures and networks, making traditional system distinctions less meaningful.
- Industrial protocols like OPC UA and MQTT are replacing serial and proprietary networks, changing system designs and required skills.
- As legacy experts retire and systems become more complex, fewer individuals fully understand end-to-end control system architectures.
It’s becoming increasingly difficult to differentiate between the different forms of industrial control systems (ICS) including safety systems, which are integrated into several distributed control systems (DCS) or programmable logic controllers (PLC). PLCs have, for some time now, advanced CPUs with high-availability configurations that can handle complex control strategies once reserved for a DCS. Meanwhile, DCS architectures are becoming increasingly modular with distributed or remote I/O and SCADA systems, including edge computing and local logic execution. Both trends blur their earlier boundaries.
Nearly all control systems now rely on Ethernet-based, industrial networks and communication strategies, such as OPC UA and MQTT, instead of proprietary or serial protocols. Colleagues of the same vintage as myself tell me many new graduates aren’t familiar with serial communications, which may be because they’ve never seen a PC with a serial port. This concept of knowledge dilution happens, not only because systems evolve and the folks who worked on them retire, but also because of increased specialization. The need for specialization happens because our systems spread into other domains and configurations, and also become increasingly complex.
Most organizations have specialists for each part of their system:
- Control and I/O predominantly at the field level;
- Servers and operator interfaces in the main control center;
- Software and applications that include coders to support engineers for each type of controller, and often a separate set of people for server-based tools;
- Network infrastructures consisting of field-level wired and wireless protocols, and controller-level communications; and
- Cybersecurity for all systems, including legacy devices with serial communications.
Because of the wide variety of skills required for each of these systems, it’s not surprising that a single person probably won’t understand it all. ICS architects may understand how all the pieces fit together and what part they play in the overall system, but that may not be enough for critical systems and infrastructures.
Let’s take a high-level look at one piece: wireless networks.
They can be installed everywhere, providing coverage for any part of a facility, which is why deployments continue to increase. However, automation systems and wireless sensor networks aren’t the only users vying to run on the 2.4 GHz frequency band. Bluetooth, Zigbee and the IEEE 802.15.4-based protocols are limited to 2.4 GHz. Some IEEE 802.15.4 variants can run in sub-GHz bands (868/915 MHz) for longer range.
Fortunately, most modern smartphones and Wi-Fi devices support both 2.4 GHz and 5 GHz Wi-Fi bands (often called dual-band Wi-Fi), and remove stress from the system. To ensure this happens, one facility where I worked created site-spectrum allocation maps, reserving process parts of the facility for 2.4 GHz, and forcing any non-process networks to run at 5 GHz.
Get your subscription to Control's tri-weekly newsletter.
Spectrum is one part of the challenge; the other is coverage, especially in process facilities with equipment that generates electromagnetic fields (EMF), reflections and shadows due to all the metal. There’s also the need for reliability and/or alternate routing if an access point gets temporarily blocked, but still has to maintain a required response time. Obviously, this isn’t a task for the weak at heart.
Models, including artificial intelligence (AI) tools, are one way to forecast demand, plan coverage and manage the intricacies of an integrated wireless system. Speaking of AI, there’s quite a bit of conversation about usinbg it for industrial automation (AI for IA). This is mostly about how to improve operations with targeted applications such as:
- Predictive maintenance and anomaly detection, which predicts equipment failures and detects anomalies;
- Quality control and computer vision that identifies microscopic defects with higher accuracy;
- Real-time tuning of process variables and feedback loops;
- Digital twins and edge AI to accelerate optimization cycles;
- Generative design and supply chain forecasting to optimize product flows;
- Collaborative robotics (cobots) that let robots and humans collaborate safely; and
- Energy management and sustainability to optimize energy use and manage emissions.
Perhaps using generative models to help us understand these systems and provide gap analyses and a common understanding of them would be an easy win because they’re suited to this type of work. Knowing how your system works will not only allow you to optimize its operation, but will likely increase reliability and reduce operating costs as well.
About the Author
