Predictive maintenance is one of the most common boxes that gets checked on the list of reasons to adopt the Internet of Things. But that’s only part of the story: using sensors, data, and analytics to predict equipment failure and direct preventative maintenance is a huge step forward, but it doesn’t really address the broader picture of reliability and end-to-end equipment maintenance.
That’s true, in part, because predictive maintenance doesn’t comprehensively address system reliability. In many cases, for example, predictive maintenance only applies to some equipment or a subset of failure modes. When machinery runs “on rails” and consequently the failure conditions are relatively well understood, or there are a limited number of ways in which equipment can fail, predictive maintenance is a powerful tool for managing those scenarios.
But that’s not always the case. If a piece of equipment is subject to a relatively unbounded number of failure modes, for example, it simply isn’t practical to use machine learning to train an algorithm housed in the IoT software to detect, predict, and take action on them all.
That’s not to say that predictive maintenance isn’t valuable – it is – but it’s simply a single tool in a much more expansive toolkit. The next logical step is to start thinking more broadly about digitally enabling systems reliability.
In other words, it’s important to capitalize on the potential of a fully-digitized maintenance framework. That means rather than simply capturing information from a small set of sensors in real time, there’s a much larger universe of data to work with.
And thankfully, that doesn’t necessarily mean adding a lot of new sensors or other technology to your existing system. You may already be capturing an enormous amount of information about your systems – just not in a form that’s immediately useful. Maintenance data might be captured in standalone databases, using non-standard coding, and free-form text fields entered by service technicians, engineers, or floor personnel who have no need or guidance with regard to storing data in a way that is forward-compatible with other digital systems.
It’s these building blocks that can help you step up to digitally-enabled reliability. It’s not a matter of implementing new technology or extending your IoT; it’s all about capitalizing the systems you have in place by converting manual and analog data collection to digital systems and standardizing the digital data so it’s suitable for analysis, machine learning, and analytics.
Of course, it’s also important to be able to make use of the data that you’re warehousing. The data needs to be accessible to diagnostic and fault detection tools, for example, and processes put in place to leverage that data on an on-going basis. That data can also be used in new ways that go beyond the tactical questions of maintenance and reliability. Armed with digital asset management solutions, the data can help inform strategic decision-making through processes like cost-benefit analysis, helping to steer the way towards upgrade, repair, replacement, and disposition decisions.
The bottom line is that the difference between the kind of predictive maintenance that comes from implementing smart sensors within an IoT and birthing digitally enabled reliability is one of holistic intent. As long as an organization thinks of predictive maintenance as a “feature” of the Internet of Things, they’ll be limited.
But when an organization proactively designs a plan for capturing data, standardizing it, and using it comprehensively to maintain equipment, the training wheels will have come off.