Edge control evolution
The future of industrial automation lies with true edge controllers, which combine programmable logic controller/programmable automation controller (PLC/PAC) functionality with general-purpose computing to deliver responsive edge analytics and real-time reaction to insights. See edge controller vs. PLCs, PACs, IPCs: 3 advantages.
- Edge controllers can do more than a traditional programmable logic controller or programmable automation controller (PLC/PAC).
- The right edge control/computing design should allow the deterministic and general-purpose aspects to evolve.
- Edge controllers can improve machinery health, supervisory applications and energy optimization.
Industrial automation users and designers have experienced significant interest in what is termed as the “edge.” Edge technologies are often considered to include hardware and software located near machines and sensors where data is sourced. They can perform some computation or even control optimization task, and also transmitting data to higher level and cloud-based systems. Accessing the expanding amount of available data and acting on the resulting information is necessary and valuable for improving operations. Edge devices that combine functionality of programmable logic controllers (PLCs) and programmable automation controllers (PACs) providing these functions.
From a high-level viewpoint, there may seem to be many satisfactory options for performing these types of tasks. However, achieving industrial-grade robustness while delivering advanced computational capabilities is a significant challenge, which narrowing the range of choices. This need is not only about enabling streamlined data flow up to the cloud. It also involves a downward flow of information technology (IT) and computing capacity into the operational technology (OT) space where control is being performed, and the ability to generate analytical results at the edge.
Consider a modern car, which uses automation hardware and software throughout. Drivers need drivetrain management systems to be reliable while secondary systems like in-dash navigation are considered important, but less critical, and more likely to need updates.
Is a comparable model possible for industrial automation applications, combining reliability control with the capability for performing advanced supporting computation (Figure 1)? Any designs built on inadequately industrialized technologies or implementations may compromise reliability, so significant challenges exist.
Optimized hardware and software options for delivering dependable, real-time automation combined with communication and computing capabilities at the edge are available. This article describes what designers should be looking for as they evaluate these types of future-ready edge solutions.
Controllers and computing for industrial use
Many controllers and computing components are advertised as suitable for the industrial edge, but it is important to understand some distinctions and underlying design details so users can ensure they are getting what they expect. Some products use a PC architecture with software virtualization and an emulated control runtime and may not be robust enough for the demands of industrial use. Other products use two separate processors to achieve the control runtime and general-purpose computing capabilities, which is expensive.
Two design terms often are associated with edge implementations: Hardware-independent and software-defined.
Hardware-independent edge implementations involve software intended to run on any industrial hardware platform. This flexibility can be a convenience, but it usually entails some level of sacrifice or risk. For instance, there may be esoteric incompatibilities or a lack of guarantees regarding determinism, compatibility, or performance, and downtime is usually necessary for certain upgrades. Hardware-independence is largely a consumer-grade model, but it is offered for some types of industrial applications.
Software-defined implementations are more rigorously tested to deliver the deterministic performance necessary for reliable, repeatable, and safe control and computing. This is crucial for industrial control applications, but often requires tailored hardware.
While general-purpose computing solutions may be suitable for non-control applications, most industrial control situations demand something more. For many years, industrial automation projects have used PLCs, and more recently PACs, to deliver deterministic control, with both delivering long lifecycles of 15 years or so.
However, PLCs/PACs have been rather limited for providing general-purpose edge computing. They tend to lack the processing power, memory, and storage required to run modern analytics or visualization applications typically available with Microsoft Windows and Linux operating systems. Industrial PCs (IPCs) can provide the desired general-purpose functionality and performance aspect but often lack the dependability required for real-time operations when loaded with third party software, and often have lifespans of five years or less.
A combined solution would be ideal, but a hardware-independent design can’t provide the necessary performance guarantees across the deterministic and non-deterministic applications. Only software-defined designs implemented on validated hardware can provide the performance required for mission critical operations, while enabling analytics and machine learning to work in parallel.
Edge computing reliability that can evolve
There are some hardware-independent edge computing solutions in the market, and they are suitable for certain forms of data gathering, analytics, and visualization. However, commercial-grade solutions can experience glitches that are not acceptable for industrial-grade use. When end users need near-real-time performance for systems which demand always-on operations and can’t accept even brief outages for security or other updates, a better solution is needed.
Enhanced data and analytical features operate on dynamic data, so these computations are most effective when performed close to the source, such as within a PLC. High-speed control combined with edge computation is a higher-level value proposition than edge computation by itself. This is because low-latency data can be collected and analyzed in real time. The resulting insights can be put into action without tying up operators or unreliable multi-system interfaces. Advanced tasks like machine learning (ML) depend on edge-located data access and computation.
The right edge control/computing design should allow the deterministic and general-purpose aspects of edge control and computing to each evolve in their own space and at their own pace. True edge control can meet end user needs when it is properly architected.
Four edge controller characteristics
To correctly merge deterministic control and analytic computing at the edge, a new class of hardware is needed (Figure 2), particularly when it has the right characteristics:
- A deterministic real-time operating system (RTOS) for control, which infrequently requires updates.
- A general-purpose operating system (GPOS) for computations, which can be upgraded at will to add features such as new apps or machine learning algorithms, or to provide security updates.
- Hardware virtualization to ensure interdependence of the RTOS and GPOS, so each OS runs, and can even be rebooted independent of the other.
- The ability for the two OS’s to securely interact, so the GPOS can obtain data from the RTOS, and so the GPOS can inform the RTOS of optimal settings.
A true edge controller can’t be fashioned by running just any type of software on any generic hardware. Instead, purpose-built hardware managed with a hypervised system organizes one part of the hardware dedicated to run a RTOS, and another to run the GPOS.
While the RTOS is demanding in some ways with regards to timing, the deterministic functions are performed by modern hardware in a carefully-architected edge controller. The key is ensuring the GPOS functions do not interfere at all with the RTOS, beyond specifically-designated secure communications.
To the edge and beyond with industrial controllers
The next question is once one has an industrial-grade edge controller, what can be done with it?
In the simplest case, an edge controller acts just like a PLC/PAC for control applications. It operates with standard industrial I/O and supports standard, open communication protocols. The edge controller also can be used for collecting and storing data from PLCs/PACs and other OT sensors and data sources, processing and analyzing this data and then visualizing or sharing to client applications and/or higher-level IT systems like a PC (Figure 3). However, most users would not choose to implement such a capable edge controller for just one of these dedicated goals, when a standard PLC/PAC or an IPC would serve.
A true edge controller can fulfill these roles. This is what the future of industrial control systems (ICS) needs to move towards—edge controller hardware/software capable of projecting IT functionality into the OT environment in a reliable manner.
Edge controller vs. PLCs, PACs, IPCs: 3 advantages
Consider these three examples where a true edge controller shines in comparison with using traditional PLCs/PACs and IPCs:
- Supervisory applications: An edge controller integrated in conjunction with multiple PLC-controlled machines provides the advantage of coordinating/synchronizing the individual assets while accessing the data and logic so users can also optimize line performance. For example, if a downstream machine is encountering issues, the rate of an upstream machine can be slowed to prevent overloading. Or, if a product is trending negatively on a quality parameter, an upstream operation can be adjusted in real-time to compensate.
- Machinery health: An edge controller allows the machine to proactively track cycles, alarms, quality rates, and more. By knowing the status of various components, the machine can actively compensate for wear or other issues without human intervention. It also allows the operator to directly access information and even view it on the local HMI along with the standard machine operating functions.
- Energy optimization: An edge controller can actively track energy usage for increases, anomalies or other issues, and send alerts or even proactively make changes. With access to energy prices or other relevant data, the machine can be programmed to make adjustments (lower rates, lower temps, etc.), or even idle itself at times when it might be most costly to keep operating.
Future of controllers and industrial automation
End users have used reliable real-time control platforms for many years. Joining that requirement is the growing need for data accessibility and cybersecurity. Traditional PLCs/PACs/PCs, and some newer edge solutions, can provide portions of what end users require.
The real future of PLCs/PACs, and indeed industrial automation, is widespread adoption of modern edge-enabled control, made possible by specifically-designed edge controllers. Solutions built on generic hardware or relying on consumer-grade software may satisfy in limited cases.
However, the intrinsic reliability and edge-located performance delivered by modern edge solutions is the only complete answer to meet current and future industrial control system needs.
Derek Thomas is the vice president of marketing and discrete sales for Emerson’s machine automation solutions business. Edited by Chris Vavra, web content manager, Control Engineering, CFE Media and Technology, email@example.com.
Keywords: edge computing, edge controllers
Where did you successfully use edge controllers, and what were the results?