“You've got to be closer to the edge than ever to win.” Dale Earnhardt
That’s true in auto racing. And it’s becoming increasingly true in manufacturing – at least with regard to information technology.
It’s more critical than ever that smart factories perform data processing at the edge of the network, closest to the devices that generate that data. In many ways, this is enormously preferable to computing at a centralized source, such as a server or the cloud. When everything flows to and from a centralized source, there are issues of bandwidth, latency, connectivity, security and cost. The failure to address such negatives can eat away at a manufacturer’s competitive advantage.
To thrive in today’s manufacturing, it’s necessary to have a coherent edge computing strategy.
The edge: What is it?
With edge computing, processing occurs at the network’s outermost periphery, not at the aforementioned centralized points. Sensors and cameras, for example, can function as smart devices without being internet-connected. They can now make decisions, such as shutting down a conveyor for maintenance.
Reducing data flow is more important than ever. In addition to being unwieldy, the vast amounts of data used in manufacturing are also expensive to move to and from centralized sources. This plodding movement of information also results in high-latency issues. Latency is something with which we’re all familiar: We watch a video buffer and wait impatiently for the content to resume. While we’re merely inconvenienced by latency in that setting, the consequences in manufacturing are more serious. Delays in sending and receiving can slow down important decision-making, diminish efficiency and hurt the bottom line.
Sensors and other gateway devices that process data at the edge are changing all of this. With an effective edge strategy, a smart factory can reduce constraints on the system, lower costs, and enhance security and privacy.
Predictive maintenance and other benefits
One boon the smart factory enjoys as a result of edge computing is predictive maintenance. Instead of waiting for equipment to fail, computing near the device can help determine when a maintenance check is necessary. Not only can such a strategy prevent costly breakdowns, but it can also eliminate the superfluous maintenance checks that hold up production unnecessarily.
Devices that once had no digital footprint and were dependent on external processing can now monitor conditions in the factory and make decisions. There are a host of applications here, including the benefit of preventing defective products from proceeding down the line and out the door.
This strategy also informs staff in the smart factory when energy usage has become unnecessarily high. The ensuing recalibration of equipment further reduces costs in the manufacturing process.
Security and privacy
Edge computing addresses the ever-present need for safety, security, privacy and regulatory compliance. That’s because traditional cloud and server strategies create undue risk. When such a glut of data passes from the factory’s multiple devices to the cloud or to a remote server, the potential for a data breach can be considerable. Data flowing offsite becomes exposed, vulnerable. But when processing takes place at the network’s edge, it stays in house and is not nearly as susceptible to nefarious outside forces.
To make edge computing a reality at any given manufacturing plant, the first step is to determine the business case for doing so. Or in other words, what’s the performance and cost benefit to be gained? Once that is determined, the edge implementation is constructed to support that business goal.
In general, an initial requirement is the ability to extract data from the machines. This is often a challenge because the pieces of legacy equipment in many factories are not set up to communicate with one another. And in some cases, the mere opening of the “box” is problematic, possibly voiding the warranty.
Once you determine how to extract the data, the next question is this: What will that data look like and where does it need to go? It may be necessary to convert analog data into digital or to work with disparate protocols. Some data will need to head north to the cloud, other data to the east and west to other machines. And so, the first step is installing the proper hardware and software solutions that will make the data usable.
In a typical platform, a smart gateway gathers data from sensors in the factory and performs real-time edge analytics. Optimized for operational technology environments, rugged gateways should use standards-based architecture to ensure compatibility with I/O and data protocols, as well as built-in security features to provide the reliability and up-time that smart factories need. Add to that a data distribution service layer that makes it easy to connect and integrate new devices into the solution, and business analytics software to provide predictive modeling for edge analytics.
On the server side, an industrial server can be preconfigured to support data capture, visualization, scoring model updates and results deployment. This type of solution can predict asset failure or quality issues, as well as enable data integration into back-end systems for further analysis and business improvements. Once the infrastructure is in place, the factory is able to operate in a new realm, a smarter and more efficient one.
Getting close to the edge makes a world of difference.
About the Author: Jason Ng is a business development director for ADLINK in Singapore. Ng worked for six years as a software control engineer before joining ADLINK Singapore as a general manager, a position he held for 15 years. Currently, he is a business development director at ADLINK Technology, responsible for the smart manufacturing vertical market.
Edited by Ken Briodagh