Three foundational stepping stones for predictive maintenance

To successfully implement a predictive maintenance (PdM) strategy, it is essential to have the right tools, methods of data collection, and an overall framework in place. Although some manufacturers may want to immediately implement PdM, it is critical to take a gradual approach to guarantee long-term success.

Before diving into predictive maintenance, take the time to assess your current maintenance strategy by conducting an audit to identify areas of improvement. If your manufacturing process lacks access to data flows, historical databases, or sensors to monitor asset performance effectively, it may be too early to implement predictive maintenance efforts.

 Even if you’re in the early stages of developing a predictive maintenance strategy, you can build upon existing workflows and methodologies to work your way toward a data-driven predictive approach. 

Below, we’ve highlighted three “stepping stone” stages working up to a predictive maintenance strategy. Keep reading to learn what each stage entails and how you can advance toward a true predictive maintenance strategy.  

1. Reactive maintenance

Reactive maintenance is the act of repairing assets only after a break event occurs. Primarily, teams are on call for repairs instead of planning ahead to prevent breaks. While some planned work can be done (e.g. lubrication, changing filters, etc.), it centers on the concept of ‘if it ain’t broke, don’t fix it.’ This is the most basic maintenance strategy because it does not require as much investment or data.

As the manufacturing industry has evolved, reactive maintenance has become less and less common in manufacturing. While some breaks, jams, or other events are unavoidable, most manufacturers try to avoid reactive maintenance when possible. 

On the surface, the low startup cost might make reactive maintenance seem like a sound investment. You don’t have to take your experts away from day-to-day tasks to focus on building future predictions, you get by as needed, and you can sometimes keep a leaner staff. 

However, reactive maintenance can have a negative long-term impact on the life of your assets, the morale of your workforce, and your overall efficiency. With reactive maintenance, your assets are decaying and you’re stuck in a “waiting for it to break” mode (sometimes waiting for breaks is unavoidable, especially in volatile markets). 

By only completing maintenance when an asset is taken offline after an event, your company may experience more frequent maintenance, longer delays, and higher costs overall. You also must hope that you get it right during the first unplanned downtime pass so that you can get up and running without a ton of trial and error.
Many manufacturers feel the production quota impact of unplanned downtime, but it can sometimes be difficult to calculate the financial impact of these events. Reducing unplanned downtime helped one of Braincube’s building materials customers save $250k by identifying an ideal replacement part’s schedule.

2. Preventative maintenance

Preventative maintenance builds upon the shortfalls of reactive maintenance. With preventative maintenance, organizations determine a maintenance schedule based on past events. 

It’s important to note that this doesn’t necessarily mean teams use historical data to schedule planned maintenance. Instead, it’s about sticking to a regular maintenance cadence that has always been in existence. 

Often, this is where SMEs and process experts are highly valued. These employees are intimately familiar with their processes, equipment, and what needs to be done. Realistically, this type of maintenance is built off the way it’s always been done: changing the filter every month, the oil every quarter, and resetting the machine during a long weekend.

Maintenance is then pre-scheduled to occur following a predetermined number of days, runs, batches, etc. These maintenance tasks are performed whether a maintenance event has occurred or not. 

Preventative maintenance certainly builds upon the limitations of reactive maintenance. By performing routine, preventative maintenance, companies are extending the life of their equipment. Performing maintenance in this way may save money, too, as a series of smaller “upkeep” tasks are often cheaper than repairing major damage caused by breaks and outages.

In the long run, though, you may end up with higher maintenance costs. If you are replacing parts before they are truly at their end of use, you may dispose of parts that are still usable. You may end up spending more money on maintenance labor and new parts than necessary. 

Another benefit of preventative maintenance, as opposed to reactive maintenance, is that teams work on removing unknowns. With reactive maintenance, teams never really know when a break will occur. This is both risky and a potential blindspot in terms of operational efficiency. Since maintenance is more organized and planned under a preventative maintenance strategy, teams can better plan around downtimes. Ideally, this will also lead to fewer unplanned breaks.     

Preventative maintenance can be easy to improve, too. As long as teams are tracking events and documenting solutions—even if this is done manually or on paper—they can start to discover trends and repair cycles. In turn, these trends can inform maintenance calendars and help teams work toward a predictive maintenance strategy.

With preventative maintenance, though, it’s hard to know if there is a better way. Maybe you have found the secret sauce and your OEE indicates your success. However, most manufacturers find that merging expert, intrinsic knowledge with data helps them to test their assumptions. Combining these two powerful resources enables them to build more reliable ways to measure their success.

For example, a CPG company struggled with frequent machine downtime and unidentified root causes. This forced them to implement overtime weekend shifts with additional staff and pay. They had data and were using it every day, but it wasn’t harmonized in a way that made it easily applicable to their maintenance strategy. Essentially, they were operating a preventative maintenance strategy. 

Using the Braincube IIoT platform, they could utilize data more effectively and improve their OEE by more than 20%—far beyond their expectations given the complexity of their process, which produced more than 1 million parts per year.

Better access to centralized and structured data helped them build a better maintenance plan, resulting in substantial savings for the company and improved morale for employees who were no longer required to work extra hours.

Despite its shortfalls, preventative maintenance is an important stepping stone to PdM. While tracking events can help lead to some improvements, there is still a lot to do if your teams are to start predicting maintenance needs before they occur. 

Don’t be discouraged, though. Remember: you have to walk before you can run. As your organization tackles PdM as a series of stages, preventative maintenance can help create steady wins and keep your staff engaged in your long-term strategy. 

However, by only determining maintenance from an anecdotal experience, you lack visibility into a potentially better way. Without data to validate when and how often to conduct repairs, teams may over or under-service equipment both of which come at a cost (whether it’s monetary, wasted time, or reduced productivity). 

3. Condition-based maintenance

With condition-based maintenance, teams strive to repair or service an asset before an unexpected maintenance event occurs. At this stage, maintenance departments are moving closer to deploying a comprehensive predictive maintenance strategy by monitoring real-time conditions. 

Utilizing real-time data means that maintenance can occur when there is an actual need as opposed to the kinds of estimates used in a preventative strategy. Think of it as adding a data stepping stone toward a more comprehensive preventative maintenance strategy. 

In this stage of your strategic maintenance journey, front-line workers use real-time data to understand when critical conditions arise and indicate a pending break. These event indicators may include a change in air pressure indicating a valve needs replacing, a shift in basis weight indicating the need for a new filter, or grinding vibrations requiring more lubrication. Ideally, critical condition thresholds are tied to alerts that notify teams before a failure occurs. 

These sensors and data-gathering technologies have come a long way. Tools such as acoustic analysis monitoring can detect problems in equipment based on emitted sonic or ultrasonic sounds. Infrared thermography images leverage cameras to detect high temperatures that lead to equipment rust or wear. Vibration analysis uses built-in real-time sensors or handheld analyzers to monitor a machine’s vibration. Changes or inconsistencies in these vibrations could indicate a worn component.

Tools like Braincube’s FFT pre-configured node makes frequency monitoring (e.g. vibrations, sound, etc.) and alerting straightforward and simple. The FFT node pulls in data from different sensors, systems, or devices as part of a self-built data flow. Being able to use a drag-and-drop, pre-created node saves IT teams time and resources. In addition, it allows ongoing customization and changes as new instances are identified and/or use cases are expanded.

This type of maintenance strategy can be honed and improved over time. For example, after gathering data for some time, alerts can be automatically triggered when key signals are detected. In other words, teams don’t need to wait until critical conditions arise: they can see drifts or trends occurring in production prior to reaching critical states. 

These data-driven advancements improve service scheduling as teams have a better “heads-up,” enabling them to plan maintenance more effectively. As a result, maintenance teams are no longer over-servicing assets (saving both time and money) and operations are less impacted by downtime.

Frontline workers have many competing priorities. Tools that automate formerly manual tasks (e.g., reporting temperatures, detecting vibration changes, or other mishaps that indicate failure) can help catch these problems without requiring employees to be in front of each machine. Additionally, applications like Braincube’s Dashboarding and Live Plant Floor apps multiply manufacturers’ abilities by visually focusing efforts based on the most pressing needs. 

Implementing IIoT into the shop floor architecture goes beyond improving processes via better, more usable data. These technologies make your teams more capable and efficient. They can rest assured that they will be alerted when they are needed. Additionally, data can be fed into root cause analysis (RCA) and predictive model building, gearing your company up toward the next stage in maintenance maturity. 

Summary

As your teams gain proficiency in these foundational predictive maintenance approaches, they move closer and closer to adopting a true predictive strategy. Though it’s tempting to get started right away and push a predictive strategy forward, taking the time to get employees and workflows working in sync will have valuable payoffs down the road. 

When you’re ready to make the move to a more predictive approach, make sure to check out our key considerations for building a predictive strategy.