End-users require ever-more data from sensors, instruments and edge-computing devices, but they’d also appreciate scaling fewer networking and software-based layer to get it. Fortunately, several useful strategies are emerging and gaining ground.
“Initially, the desktop workstations of the computer revolution took hold on the manufacturing floor, and then moved to thin-client/server based architectures in the early 2000s. This was done to create a more manageable and cost-effective compute platform for industry. Recently, the desire to move large amounts of data combined with low-cost, edge-compute hardware allowed the addition of intelligent edge devices alongside on-prem and cloud servers,” says Ryan Gerken, principal engineer at Eosys Group Inc., a system integrator in Smyrna, Tenn., near Nashville, and a certified member of the Control System Integrators Association (CSIA). “Massive improvements to remote management and deployment technologies further enabled the viability of edge compute at scale.”
Get your subscription to Control's tri-weekly newsletter.
Gerken adds that a new class of DataOps software is available. “While this software is marketed to facilitate unified namespace (UNS) deployments, in practice, it usually adds another layer and additional cost to edge-data movements,” he explains. “While this software can be a good value to some, in our experience, many customers aren’t excited about the prospect of additional recurring subscription fees.”
Because these factors can make it hard to generate and promote savings, Gerken reports that Eosys uses products, such Rockwell Automation’s Embedded Edge Compute or Inductive Automation’s Ignition software on large projects. “Ignition software can serve as both a connectivity and DataOps platform. And, when used in a distributed fashion, it can help move the compute load closer to the edge, improving calculations times of real-time key performance indicators (KPI) and metrics.” says Gerken. “Traditionally, these calculations were done at the server, but they can get overloaded and out of scope. Consequently, they need to find better ways to push calculations down to the edge, where they can use software like Ignition Edge that runs on mini PCs or other small industrial personal computers (IPC).”
Different users, different needs
Gerken reports that two Eosys clients with multiple plants are using closer-to-the-edge computing and Ignition to improve their operations.
The first is a large construction materials manufacturer with numerous, geographically diverse, U.S. production sites, which maintains a light IT footprint in the field by running lots of monitoring and scheduling functions from its corporate headquarters, and describes itself as a “cloud-first company.” While many users think of the edge as their front-line of production with sensing devices in the field and on the plant floor, this organization considers each of its remote facilities as its edge, and deploys Dell Edge Gateway 5000 Series running Ignition software to manage them.
The second client is a large automotive manufacturer with fewer total plants that are also geographically distributed, though each is larger than the construction materials manufacturer’s sites, and has greater data density because they move more real-time KPIs and require more edge computing. Because they’re taking in data from millions of tags, these automotive plants generally report data by exception via MQTT publish-subscribe format.
“One lesson from these two types of plants and organizations is that it’s crucial to figure out how to organize and identify their data,” says Gerken. “If not, it will be very messy when they try to scale up.”
Gerken reports that the five-layer ISA-95 standard for developing automated interfaces between enterprises and control systems can help provide an initial model for organizing data from facilities, areas within them, and individual shops or work cells. However, he adds that relying on the ISA95 levels to identify the data won’t cover all cases in the UNS. For example, it’s possible to have operations nested within unit operations, or perhaps a sensor (like a weather station at a site), would be defined at ISA-95’s Site level.
Tags and templates
To solve this problem, Gerken states users can employ metadata tags and tag properties to classify and identify information. In turn, this can help users query the UNS without needing advance knowledge of the data. This is useful for dynamic analytics and reports.
Deploying edge compute at scale is possible using zero-configuration platforms to assist with provisioning, orchestrating and updating edge devices. This zero-configuration combined with application-level, remote provisioning allows quick commissioning.
“Users want more data from their equipment, and edge-compute, deployment technology helps them implement quick network connections, device templates and online, device-communication configurations to facilitate data collection, which saves on travel time to the field,” concludes Gerken. “These templates let users get data flowing in hours, where getting teams into the field previously took a week or two.”
