It's 2002, and the widgets come rolling down the factory's conveyor. Then it happens: a misfortune nobody anticipated. An equipment breakdown halts production dead in its tracks. Technicians swarm over the machinery to make the fix as swiftly as possible. Every minute of the production shutdown means lost revenue.
Fast-forward to today's smart factory, where edge computing and solutions from the Industrial Internet of Things (IIoT) make predictive maintenance possible. This technology can closely monitor the conditions under which the equipment is apt to fail, and it alerts staff when a malfunction is impending. It also pinpoints the best time to perform proactive maintenance, when a lull in production minimizes any losses in time and money.
All of this is part of the smart factory revolution.
Edge computing transforms manufacturing
Sensors are everywhere in many of today's factories and fabrication plants, taking in reams of data. They measure any number of variables that affect the factory's performance or portend machinery breakdown. The processing of this data often occurs at the edge of the network, which affords a host of advantages over cloud-based systems. Information from the sensors is ingested by the software-driven smart gateways in real time for processing as near the device as possible. Edge computing cuts the latency that results when data goes to a remote cloud server and returns. The factory enjoys stronger performance, lower data costs, fewer network bottlenecks and greater security.
This technology is erasing much of the uncertainty in manufacturing and other processes. Imagine a piece of equipment that tends to fail when the temperature climbs too high. As IIoT sensors monitor that equipment and keep track of the operational temperature, edge computing processes and interprets the data via complex algorithms. When conditions reach statistically dangerous levels, alerts go out to staffers so they can deal with the impending failure proactively. This is an enormous benefit, because forecasting an equipment failure is always better than scrambling after it occurs.
Predictive maintenance, therefore, is one of the most transformative aspects of the smart factory.
But it's not as easy as it sounds. A factory owner does not walk into a local IIoT store to buy some one-size-fits-all computer products ready to plug in. That's because factories differ profoundly from one another. Part of the process is tailoring the predictive maintenance process to each facility's specific operations.
Among the daunting challenges is the fact that so much legacy equipment exists in factories today, and the disparate pieces usually don't talk to one another. There's no universal protocol. In addition, the IIoT solutions have to work seamlessly with both operational technology (OT) and informational technology (IT) – two modes of technology that are often worlds apart.
Stopping failures before they start
Mechanical failure is a costly issue that has bedeviled factories since the first Industrial Revolution. The related problem of product failure can be expensive, too. If conditions are inhibiting the proper manufacturing of a product, a factory ends up with a great deal of waste and lost productivity. Even worse, faulty products can end up in the hands of consumers. Apart from IIoT technology and the accompanying edge computing, all this might easily go undetected.
In a smart factory, IIoT sensors monitor all kinds of conditions – for example, whether the ambient temperature is ideal for spray paint adhering to metal. Data processing at the edge determines when conditions become detrimental to normal manufacturing, and alerts go out to staffers who can respond accordingly.
This technology also factors in history. Suppose there are certain times of day when temperatures tend to put product manufacturing out of compliance. Predictive maintenance can look at the history of manufacturing failures and educate employees on how to deal with an issue before it becomes a problem.
Trimming unneeded maintenance
Predictive maintenance also informs staff when costly, ongoing maintenance is unnecessary. It's a common scenario: Technicians shut down and service equipment at regular intervals, whether it actually requires that servicing or not. Sometimes they replace components that don't yet need replacing, or lubricate an idled machine that requires no such action. Predictive maintenance takes that guesswork out of the equation: The machine isn't lubricated until it needs to be, and the components aren't swapped out if they're still good. Production rolls on without interruption.
When equipment does need servicing, these advanced edge analytics can reveal the best times to perform such maintenance, the times that least affect the production schedule.
IIoT analysis has another benefit: Determining the conditions that have led historically to the best product yield. This is important in the case of products sensitive to their surrounding conditions, such as semiconductors.
In addition, IIoT sensors can determine when a piece of equipment is drawing too much current or operating in other ways that diminish energy efficiency. Staff can then recalibrate the equipment and save the factory even more money.
What to look for
There are a number of predictive maintenance solutions on the market today. ADLINK, Intel, IBM and PrismTech (an ADLINK company) have pooled their expertise to introduce predictive maintenance solutions. These address the challenges of connecting the unconnected and creating communication efficiency through fog architecture. Each of these companies has IIoT solutions designed to address the distinct challenges that arise on the factory floor.
These predictive maintenance solutions can analyze data from the cloud or from the network's edge, whichever is necessary in a given application. And because these solutions can operate at the edge of the network, results are more instantaneous. They are secure and integrate smoothly into the factory's OT and IT environments, with low IT management overhead. The solutions are also customizable and scalable, and offer predictive analytics and secure distribution of data, making the smart factory smarter. The result is less downtime, reduced costs, enhanced product quality and greater productivity.
With all the inherent positives – better yield, less downtime, better-informed decisions – these solutions truly are part of the revolution. It's a revolution that's bound to grow as people increasingly discover the benefits of turning their factories into smart factories.
(Jason Ng is a business development director for ADLINK in Singapore.)