Given the relationship between edge computing and IoT, it’s not surprising that the industrial sector — which has spawned its own IoT subcategory, also known as Industrial IoT (IIoT) — is ripe for edge computing use cases.
The industrial sector—which we use here as a broad term for companies like manufacturing and energy (think heavy engineering and power plants, for example)—actually had a head start on the edge concept: industrial SCADA systems. In short, they are local, isolated control systems responsible for all types of critical industrial and other processes on site. You could think of them as the forerunners of the modern edge architecture.
“Industrial SCADA is a form of edge and in some ways has been around for more than 30 years,” said Andrew Nelson, lead architect at Insight. “Most facilities today will have a siled control system.” In fact, they often have multiple such systems and processes — and edge computing deployments are increasingly likely to either complement or even replace them.
Industrial environments themselves are fundamentally edge locations – meaning they are typically far from a centralized data center or cloud – and are therefore ideally suited to growing edge adoption. An oil and gas platform in the middle of the ocean? That seems to fit everyone’s definition of “the edge”.
In this sense, the industrial sector is addressing inherently challenging locations: Nelson points out that edge computing use cases overlap with other contexts, such as warehousing or logistics, but often involve more hostile environments.
All of this makes the industrial sector a good location for edge use cases. How are industrial CIOs and other IT and business leaders actually thinking about and implementing edge infrastructure and applications?
[ Building an edge strategy? Also read Beat these common edge computing challenges. ]
First, an overview of the big picture, courtesy of Red Hat Technology Evangelist Gordon Haff: There are two main directions in industrial edge computing.
“On the one hand, sensor data – often filtered and aggregated – flows from the operational/shop floor edge level towards the core,” says Haff. “On the other hand, code, configurations, master data, and machine learning models flow from the core — where development and testing happens — toward the factory.”
This has great importance for edge strategy in various industries. With edge-to-core streaming, IT leaders must decide what actually needs to live at the edge and what can or should be kept in a centralized cloud or data center.
“The idea is that you often want to centralize when possible, but remain decentralized when needed,” says Haff. “For example, sensitive production data may not be allowed to leave the premises, or you may need to protect your ongoing industrial processes from outages due to network issues outside the factory.” (The latter is a big part of the SCADA connection – in many industrial environments, unexpected downtime is not an option. )
The core-to-edge stream is primarily about operational reliability and efficiency. As with edge architecture in general, don’t expect to dispatch a human IT professional every time you need to update a configuration or patch a system at an edge location. In the industrial sector, says Haff, there may be hundreds of plants with thousands of processes running: “Automation and consistency are key,” says Haff.
Ishu Verma, technical evangelist at Red Hat, adds that with core-to-edge stream, organizations can extend the same practices and technologies they use in the cloud or on-premises to their edge nodes, even in the the harshest industrial environments.
“This approach allows organizations to push the best practices to the edge for emerging technologies — microservices, GitOps, security, etc.,” says Verma. “This enables edge systems to be managed and operated using the same processes, tools and resources as with centralized locations or in the cloud.”
Edge computing in manufacturing and energy
Within these two-way streams, here are four examples of how industrial companies are using edge computing.
1. Optimization of processes in real time
These traditional SCADA and other control systems are like the monolithic or legacy applications in many other sectors: important but not particularly easy or flexible to use in the modern environment.
“Traditional SCADA and control system infrastructure is typically closed and proprietary,” says Nelson. “An IoT/Edge deployment can help with real-time operations in a single pane of glass, rather than jumping back and forth between systems.”
“Many industrial plants will have multiple control systems that may or may not be integrated. The IoT/Edge use case can pull data across systems and correlate events and predict failures.”
Monitoring and preventative maintenance are good examples in this category: the sensors and instruments in a plant can be used for real-time operations, helping industrial operators to better plan when critical maintenance and other work is required. This has been more difficult in the past due to data silos – a well-known challenge for CIOs in many companies.
“Many industrial plants have multiple control systems that may or may not be integrated,” says Nelson. “The IoT/Edge use case can pull data across systems, correlate events and predict failures.”
2. Running AI/ML workloads on industrial sites
Latency – as in reducing or eliminating it – is one of the key drivers of edge computing strategy. This is especially true for AI and machine learning applications, as well as other forms of automation that need data — and lots of it — to be effective.
Industrial IoT offers huge potential for AI/ML and automation, but also huge impact on data and latency.
“For intelligent machines to function smoothly at the edge, a lot of data is required,” says Brian Sathianathan, CTO at Iterate.ai. “Good AI needs data. Great AI requires a a lot of of data, and it demands it immediately.”
This can become problematic in connection with Red Hat’s First Stream Haff described above: sensor data flows from the edge towards the core.
“I’ve seen situations in manufacturing plants where there’s ‘too much’ data to go from a robot on the ground, through the local network, and then all the way to the cloud and back,” says Sathianathan. “That’s not good because, as manufacturing CIOs know, decisions need to be made immediately to be effective.”
If latency is an issue, then actual downtime is an absolute killer—especially in industrial settings (where a data outage or network problem could shut down a gas pipeline, for example) and related segments like manufacturing.
While some downtime is typically acceptable in standard IT environments, this is simply not the case in manufacturing. The cost of stopping production lines because edge applications stall can be in the hundreds of thousands of dollars per minute—there’s simply no room for error.
Storing necessary data at the edge will be a game changer, pairing edge computing with AI/ML use cases and minimizing the “too much data” scenario described by Sathianathan.
Having edge applications that can automatically monitor and optimize energy usage at industrial sites isn’t just good corporate citizenship — it’s potentially a big win for the bottom line.
3. Improvement of energy management
Having edge applications that can automatically monitor and optimize energy usage at industrial sites isn’t just good corporate citizenship — it’s potentially a big win for the bottom line.
There is a big push for monitoring energy consumption and controlling the load in both manufacturing and industrial applications,” says Insight’s Nelson. “There are huge savings in industry by simply turning off or metering the electrical load during peak periods.”
In fact, rising energy consumption and costs in industrial companies are such a big problem that it was the subject of a conference presentation and a 2021 paper: An edge-computing energy management system for industrial plants.
It’s not exactly a brisk read on the beach, but CIOs and other IT leaders can certainly appreciate its gist: designing an edge application that can automatically adjust and optimize energy consumption based on fluctuating prices could be a use case that really moves the needle.
[ Related read ESG strategy: 3 ways CIOs can play a pivotal role ]
“Reducing the cost of electricity has become an urgent problem to be solved,” write the authors of the report. “Meanwhile, remote monitoring of connected devices and intelligence pushed to the fringes of monitoring devices is becoming critical in the Industrial IoT.”
4. Improving employee and site safety
Here’s a pattern: Industrial Edge/IoT use cases feed off the massive number of sensors and other machines in these environments. But it’s not just about machines – it’s also about people. Nelson says the industrial edge also offers significant opportunities for employee and site safety.
“Tracking employees and contractors and alerting them when they’re not where they’re supposed to be working is a big deal,” says Nelson.
Like many edge applications, this is a category that typically includes or is integrated with other technologies (like AI/ML). It’s also one where seemingly non-technical devices — like the ubiquitous employee ID card — can get a modern makeover.
“Computer vision, RFID, and BLE can all be used in this use case,” says Nelson. “The integration with building security badge readers and security cameras is a useful integration.”
Or try on another well-recognized security item for the greatness, one that predates edge, cloud, and, well, digital computing as we know it: the hard hat.
“They make hard hats with built-in sensors that can be tracked via WiFi access points for this use case,” says Nelson.
[ Discover how priorities are changing. Get the Harvard Business Review Analytic Services report: Maintaining momentum on digital transformation. ]