by 
The use of predictive maintenance in manufacturing is not new, but it’s evolving. Staying abreast of the trends is an excellent way to maximize returns on investment.
Here are five things that’ll likely shape predictive maintenance in the coming year.
- Improved efforts to pinpoint problems
Some of the earliest predictive maintenance solutions primarily alerted people to problems with particular machines. However, they did not offer specifics, such as which component had the issue that triggered a notification.
That’s starting to change. For example, Amazon Lookout for Equipment shows which specific sensors on a machine indicate issues. That makes it easier for a maintenance professional to start working on the problem efficiently. Additionally, Amazon’s product shows the estimated impact of an identified issue. That information could encourage decision-makers to get an issue assessed quickly before it causes a catastrophe.
Sensor data can enable better environmental monitoring, too. For example, one company uses smart sensors in a facility where people paint new cars coming off production lines. If the temperature in a room kept cool for manufacturing purposes starts creeping up, factory leaders could investigate before a problem halts production.
- Increased interest in remote assessments
Predictive maintenance strategies help manufacturing representatives identify issues sooner, but they don’t stop problems altogether. However, emerging technologies make it possible for a technician to have a clearer idea of what’s going on with a machine before arriving at a site to take a closer look.
Most people have probably called a technical support hotline and found it difficult to describe a matter with words alone. That’s why some agents ask customers to send images of error codes and similar accompanying media when possible. However, taking and sending those pictures wastes precious time.
That’s why some technicians use smart glasses that capture audio and video evidence of a machine’s behavior and broadcast it live to a support agent. The receiving party can then offer help in various forms, including sending relevant documents or drawings to the customer. This approach was particularly advantageous during the Covid-19 pandemic, but it’ll likely remain in demand well into the future.
- Enhanced efforts to take multidimensional approaches
Predictive maintenance offers numerous compelling advantages to manufacturers. For example, it increases the time between machine failures, which promotes productivity. When manufacturers began using predictive maintenance, it largely centered on condition monitoring. If a component’s value — whether temperature, vibration or something else — went outside a set parameter, it triggered an alert.
However, people are starting to harness next-generation predictive maintenance approaches. They give more context to a notification by using machine learning and multidimensional data to evaluate numerous factors of a part’s condition. This method of anomaly detection also assesses how an impending failure affects other components in a machine.
Additionally, applying predictive maintenance like this helps determine if a part is genuinely failing and needs replacement or is perhaps showing certain statistics due to higher-than-normal machine utilization. A related benefit is that people may get insights that help them determine why a part fails sooner than expected, rather than just knowing it’s no longer functional.
- Renewed realization of the ripple effects of neglected maintenance
The Covid-19 pandemic drastically reshaped the workforce, affecting things like the number of people on a shift at a given time, the layouts of assembly line stations and what workers had to do to verify they were healthy. The pandemic also disrupted maintenance schedules, often causing leaders to delay routine upkeep until things calmed down.
However, that decision had effects they probably didn’t expect, including an increase in factory fires. There was a 150 percent rise in those events in the first half of 2021 compared to 2020. Plus, 2021 is on track to have the most factory fires ever reported. Those sobering statistics may encourage leaders to see that delayed maintenance can cause costly disasters.
Hirra Akhtar is the director of supply chain risk consulting at Resilinc, a firm that released a report with those findings. The research indicates Covid-19-related staffing shortages were a major driver of the spike. “Why would a factory fire or chemical spill occur as a result of Covid? Because of less preventive maintenance, less attention to debris that collects near hot worksites, fewer safety audits taking place,” Akhtar explained.
- Combined digital twin and predictive maintenance strategies
It’s becoming more common for manufacturers to use digital twins to streamline their factory operations. Having a highly detailed virtual version of a factory makes it easier to see where bottlenecks occur, plan new equipment installations and test new processes before implementing them in the real world. However, a more recent trend concerns using digital twins to support predictive maintenance.
In one case, a digital twin saved a gas company $360,000 by predicting a plant outage. Not all digital twin-predictive maintenance pairings have such impressive results. However, they’re more likely to pay off when there is a maintenance issue with a target to predict and when a company has high-quality operational data for the machine in question.
Manufacturing leaders should also bear in mind there are numerous kinds of digital twins they might use for predictive maintenance. For example, some virtual models show single components, while others display the complete machine.
Predictive maintenance is evolving
Manufacturers have known for a while that predictive maintenance could help them enjoy more uptime and fewer unexpected costs. However, these five trends show that predictive maintenance, and how people use or consider it, is changing. Staying on top of those trends helps increase the chances of seeing a strong return on investment.
