Defining and Implementing Predictive Maintenance With Examples

Where is the maintenance sweet spot? The point when you do just enough maintenance to keep assets in peak operating condition without needless interruptions in operations? 

It is a challenging question. One that can only be answered with the help of predictive maintenance. The ability to predict the remaining useful life of a part or an asset based on real-time data gives organizations unprecedented insights into managing maintenance resources.

To know if predictive maintenance is a strategy that could be useful to your organization, you first need to understand what it is, how it works, and how one goes about implementing a predictive maintenance program.

We’ll answer all those questions and more.

What is predictive maintenance (PdM)?

Predictive maintenance is a proactive maintenance strategy that uses condition monitoring tools to detect various deterioration signs, anomalies, and equipment performance issues. Based on those measurements, the organization can run pre-built predictive algorithms to estimate when a piece of equipment might fail so that maintenance work can be performed just before that happens. 

The goal of predictive maintenance is to optimize the usage of your maintenance resources. By knowing when a certain part will fail, maintenance managers can schedule maintenance work only when it is actually needed, simultaneously avoiding excessive maintenance and preventing unexpected equipment breakdown.

According to a predictive maintenance report from Market Research Future, the global predictive maintenance market is expected to grow to 23B by 2025. The highest number of implementations are happening in the manufacturing sector, but all businesses that have a lot of capital tied up in their equipment are very much interested in predictive maintenance.

When implemented successfully, predictive maintenance lowers operational costs, minimizes downtime issues, and improves overall asset health and performance.

FREE RESOURCES FOR PROFESSIONALS AND STUDENTS

[Template: Work Order Ticketing System]

Quickly set up a ticketing system using Google Sheets.

Download our Ticketing Guide & Templates

Includes everything you need to make a simple custom system that works for you.

Send it Over!

How does predictive maintenance work?

The main advantage of predictive maintenance is being able to schedule work based on the current condition of the asset. However, knowing the exact condition of complex assets is all but easy.

There are three main components that allow PdM to track asset condition and warn technicians about upcoming equipment failures:

1. installed condition-monitoring sensors send real-time performance data and machine health data 
2. IoT technology enables the communication between machines, software solutions, and cloud technology; essentially helping to collect and analyze huge amounts of data
3. predictive data models are fed with all of that processed data so they can spit out failure predictions 

Let’s take a closer look at all of the major elements mentioned above to get a clearer picture of how predictive maintenance works.

1) CM technology and predictive maintenance techniques

There are a variety of condition-monitoring sensors and equipment that can be installed/retrofitted. You can measure electrical currents, vibrations, temperature, pressure, oil, noise, corrosion levels, and more.

The sensors you go with depend on the type of assets that will be on your PdM and what you want to track.

An added benefit of using condition-monitoring sensors is that you can have an accurate representation of what’s happening inside the asset without any kind of productivity disruptions. In other words, you do not have to stop the asset and pull it apart to perform physical inspections.

Based on the sensors you use and tests you want to run, there is a variety of condition-monitoring/predictive maintenance techniques that can be applied like:

• oil/lubrication analysis
• vibration analysis/dynamic monitoring
• motor circuit analysis
• different variations of thermography
• ultrasonic and acoustic analysis
• radiography/radiation analysis
• laser interferometry
• electromagnetic measurements
• different performance measurements

For a more detailed breakdown and explanation on condition-monitoring and predictive maintenance techniques, check out this guide.

2) The role of IoT technology

It is one thing to gather data, but quite another to be able to analyze and use the data for its intended purpose. By using Internet of Things (IoT) technology, sensors mentioned earlier can collect and share data. PdM relies heavily on these sensors to connect the assets to a central system that stores the information coming in. These central hubs run using WLAN or LAN-based connectivity or cloud technology.

From there, the assets can communicate, work together, analyze data, recommend remedial action, or take action directly, based on how the system is set up.

3) Applying predictive algorithms

The most important part of predictive maintenance (and arguably the hardest one) is building predictive (a.k.a prognostic) algorithms. In essence, you have to build a model that will consider many different variables and how they interconnect and impact one another – with the ultimate goal being able to predict machine failures.

The more variables you can use, the more accurate your models will be. This is why building predictive models is an iterative process. The initial models will have to be based on asset history stored in a CMMS or file cabinets, personal observations, FEMA analysis, already available internal sensors like accelerometers and flow meters, and similar sources. There might even be a need to first install condition-monitoring sensors and run them for a while to gather baseline data and finish the initial predictive models. 

Over time, the installed sensors will generate more and more data which can be used to improve the initial models and make near-perfect failure predictions.

Without getting overly technical, here is how the algorithms work. They follow a set of predetermined rules that compare the asset’s current behavior against its expected behavior. Deviations are an indication of gradual deterioration that will lead to asset failure. Based on deviations, current operating conditions, past failure data, and all the other variables built into the data model, the algorithms try to predict failure points. 

The end result is an automated system that:

• monitors operating conditions via installed sensors
• understands and predicts patterns created by data anomalies
• creates alerts when there is a deviation from established thresholds

Pros and cons of predictive maintenance

Like every other maintenance strategy, predictive maintenance comes with a clear set of pros and cons:

The main reason why predictive maintenance is not more widely adopted is because it has a relatively high barrier for entry. It can require a sizeable upfront investment to buy and install condition monitoring equipment, develop predictive models, and pair everything up with a CMMS or other specialized software. On top of that, creating models and algorithms takes specialized knowledge that often has to be outsourced. All of that means PdM can be complex to set up and run.

Despite those disadvantages, predictive maintenance is still expected to grow, which means that the benefits outweigh the challenges. We won’t repeat them again here as they are nicely summarized in the picture above. The bottom line is that predictive maintenance can provide a significant ROI in the long-term, making this a risk worth taking. 

Combining predictive maintenance with CMMS software

Over the years, CMMS has played an active role in the continuous shift from manual processes and reactive maintenance to more proactive strategies like preventive and predictive maintenance.

In fact, it’s safe to say that CMMS is at the front and center of PdM applications today. Here are a few reasons why.

CMMS provides the initial data to get PdM rolling

The information gathered over time regarding asset performance helps to form the starting point and the initial dataset before PdM implementation. Though there are other sources like hard copy maintenance records, CMMS will provide the most comprehensive and easy to access source of historical information.

CMMS integrates with PdM technology to generate automatic tasks 

A modern CMMS can automatically create an alert or generate a work order whenever sensors detect that an asset is operating outside predefined parameters. These alerts prompt the maintenance team to take preventive actions before the machine fails and makes a big mess.

In the picture below, you can see an example of an automatic task creation inside Limble CMMS. If you look closely, you can see how Limble automatically generated a high priority work order when a vibration rate on one of the assets exceeded defined parameters.

These automatic prompts may be for either a routine maintenance task or corrective work. Common examples of alerts and work orders include:

• Generate a “Fill Coolant” task when the level of coolant in the reservoir is getting too low.
• Create an inspection task to check bearings in an air-conditioner’s motor.
• Create an inspection task to check belt condition in a fan.      

In essence, even though PdM generates highly accurate asset data, that information will be limited in ease of application if it is not combined with a CMMS. On the other hand, CMMS on its own cannot measure or predict machine health. 

By combining both technologies, users get a powerful solution that is indispensable for a modern maintenance management team. Predictive maintenance lets you know when certain maintenance actions have to be taken and CMMS helps you manage your resources and incorporate those tasks into your maintenance schedule.

CMMS facilitates data interpretation

Although PdM tools provide valuable insights about asset condition through vibration, lubricant, heat, oil analysis, etc, the data generated is enormous and would be cumbersome for humans to manage manually. 

With the right CMMS, users get easy to understand “snapshot” of the data that’s coming in.

CMMS serves as a central organizational tool

CMMS pulls different information about assets together and presents them in a centralized platform that offers a more complete equipment “situation report.”

For example, PdM will give the raw data from the machines but CMMS goes further to include information from other modules like asset history, inventory, spare parts management, workforce management, repair schedules, and more – thereby helping maintenance departments make informed decisions.

Preventive vs predictive maintenance: which one is right for you?

The main difference between preventive maintenance and predictive maintenance is in the setup of the maintenance calendar. While predictive maintenance uses condition monitoring sensors and predictive algorithms to guide maintenance schedules, preventative maintenance calendar is based on equipment usage and time intervals (which are just educated guesses in the best case scenario).

While on the surface this might not seem like such a big difference, this does result in significant operational differences. It changes the way you manage your inventory, how you handle planned downtimes, and has a general impact on how you manage maintenance work. 

On paper, predictive maintenance is clearly a better strategy. However, implementing condition monitoring technology and developing predictive models can be challenging and expensive. As such, it is not cost-effective to place all of your assets on a PdM program. 

This is where preventative maintenance jumps in. Important assets that do not qualify for a PdM program can be put on a preventive maintenance plan.

Combining preventive and predictive maintenance is a recipe for success. If you’re interested in the differences between the three most popular maintenance strategies, check out our side-by-side comparison of reactive vs preventive vs predictive maintenance. 

How to establish a predictive maintenance program

Laying the groundwork for PdM is essential for creating a sustainable system. The key is to start small and scale up as the organization adjusts to this new way of doing things.

Here is a visual outline of the usual steps an organization should take when implementing a predictive maintenance program.

This is a general overview of predictive maintenance so we won’t go into too many details. We have a separate guide on how to set up a predictive maintenance program for those that want to learn more. Below is the outline of the implementation steps so you get a high-level understanding of the process.

Step #0: Secure the budget 

Implementing predictive maintenance is not something that happens overnight. Before making any plans, you need to get approval from top management and the commitment that this project will be properly funded. 

Step #1: Identify critical assets

Start by identifying critical assets to be included in the PdM program. Assets with high repair/replacement costs that are critical to production are often the best candidates. You can also run RCM analysis to see if an asset is worth being put on a predictive maintenance program.

Step #2: Establish a database

For the PdM program to be successful, another factor to consider is the presence of sufficient information that can offer actionable insights to machine behavior. Historical data  for each pilot equipment will be available from sources like CMMS, hard copy files, enterprise software from other departments, maintenance records and charts, technician’s personal experience working on the assets, etc. 

This data can be used to help establish failure modes and might even be useful when developing the first version of the predictive algorithms.

Step #3: Analyze and establish failures modes 

At this point, the organization will need to perform an analysis on the previously identified critical assets to establish their failure modes. This process will help them figure out which failures cause the most trouble and have the highest chance of actually happening. 

Step #4: Implement condition monitoring sensors and equipment

Knowing which failure modes they need to watch out for, the organization can buy appropriate sensors and technology to monitor parts that are most likely to fail. There are various condition monitoring techniques and equipment that can be applied, as we already discussed in the first part of this article.

Be wary of old assets as you might have trouble retrofitting it with modern sensors. You should double-check if this can be done before you spend money on condition monitoring technology.

Step #5: Develop predictive algorithms

With everything else in place, the next step is designing the right modeling approach that will form the basis for failure predictions. At this point, the organizations are likely to hire data scientists to develop predictive maintenance algorithms based on sensor measurements and other data the organization was able to provide.

Step #6: Deploy to pilot equipment

This is where the predictive modeling is put to test and validated by deploying the technology to a selected group of pilot equipment. 

If the process is executed properly, there will be significant improvements to the company’s operations. Noticeable impacts might take a few months to kick-in, depending on the size of your operations and how much machine downtime was experienced before the implementation.

Interested in implementing predictive maintenance into your organization? Learn how you can jumpstart predictive maintenance using Limble’s modular IoT Sensor Setup.

Application of PdM across different industries

A report by McKinsey Global Institute estimates that the current interest in linking physical assets to the digital world may actually still be understating its full potential. They project that this connectivity could generate up to $11 trillion annually in economic value by the year 2025 and that more than two-thirds of this value will be generated in a business-to-business setting.

The number of implementations and applications of predictive maintenance will increase as the cost of technology continues to drop, largely due to the switch from wire-based sensors to wireless ones.

How does all this affect maintenance specifically? Well, predictive maintenance is changing the age-old way of doing things. It’s no longer a case of “if it isn’t broken don’t touch it” but “keep it healthy so it doesn’t break.”

Here are a few industries that are at the forefront of using predictive maintenance.

PdM in the manufacturing industry

The benefits of the increased reliability that PdM offers is quite enticing and manufacturers are taking note. The same report mentioned above indicates that manufacturers’ savings from using predictive maintenance could reach between $240 and $630bn globally by 2025.a

The everyday applications of predictive maintenance in manufacturing are numerous. Specific examples include:

• using vibration sensors to identify patterns of fragile spindles in milling machines

• measuring the temperature differential upstream and downstream of the heat exchanger to identify the first signs of clogging

• monitoring robot CPU and housing temperatures, as well as positioning and overload errors, and using this data to estimate the health of a robot 

PdM in the transportation industry 

Connected transportation devices are on the increase as brands produce vehicles equipped with sensors that gather data and relay information about the machine’s performance. Fleet managers can use this information to proactively plan for vehicle maintenance. 

These practices have resulted in a measurable decrease in on-road breakdowns and spending and have reduced technician diagnostic time. Fewer unplanned downtime, lower fuel and maintenance costs, and shorter servicing time are some of the reasons why fleets are moving towards predictive maintenance.

PdM in the aviation industry

Interestingly, the origin of predictive maintenance can be traced to aircraft maintenance. Back in 1943, a British scientist named C.H. Waddington commented that “inspections tend to increase breakdowns” after he observed the maintenance activities of the Royal Air Force Coastal Command 502 Squadron. Instead, he recommended a process that was more “in-tune with the actual condition of the equipment.”

Since then, PdM has evolved and is still used by aviation companies to reduce the costly effects of maintenance-related flight delays and cancellations. Predictive maintenance is also used in the industry to help airlines monitor engine health and performance during flights by measuring various temperature and vibration levels.

Future trends for industrial application of predictive maintenance

As the cost of technology decreases and lowers the barriers to implementation, we will start seeing an exponential increase in the number of use cases across different industries. 

Early adopters of PdM are sectors that can realize the highest ROI. They are easily identifiable by the following traits:

• high capital and operational expenditure
• sensitive and very expensive equipment
• business processes that are severely affected by equipment downtime (or have a need for very high equipment uptime)
• high risks to human safety

Alongside those we already mentioned, the ones that match these criteria are industries like healthcare, mining, oil and gas, electric power, and construction. 

Businesses operating in the energy sector have a very high risk of exposure, which is why it’s estimated that they will grow the most in terms of investing into predictive maintenance technology. 

Now, before we wrap this up, let’s take a quick look at the future of predictive maintenance and how it can be improved.

Prescriptive maintenance: a step beyond predictive maintenance 

If you are researching predictive maintenance, you have probably come across the term prescriptive maintenance. Prescriptive maintenance is a maintenance strategy that looks to build upon predictive maintenance as PdM improves upon CBM and preventive maintenance.

Prescriptive maintenance does not only let you know when something needs to be corrected, but uses artificial intelligence and prescriptive analytics to suggest a few scenarios of how you can deal with the predicted problem. 

As Dan Miklovic from LNS research explains in his post:

No longer will you need an ensemble of experts to tell you how and when to maintain your assets, as the assets themselves will tell you what they need if they are unable to fix themselves.

He goes further to give an example:

Let’s say a piece of equipment is showing increasing bearing temperature. Predictive analytics looks at the temperature profile and tells you it is likely to fail in X amount of time. On the other hand, prescriptive analytics tells you that if you slow the equipment down by Y%, the time to failure can be doubled, putting you within the already scheduled maintenance window and revealing whether you can still meet planned production requirements.

Lastly, Dan suggests that the acronym for prescriptive maintenance could be RxM and while we like it, it remains to be seen if it will catch on.

Even though the applicability of prescriptive maintenance seems to be tied to advancements in AI and machine learning, how things look today, we probably will not have to wait long until prescriptive maintenance becomes the next best thing in the maintenance industry.  

Parting thoughts

And that wraps up our outlook on predictive maintenance. It is a strategy that carries major advantages, but is certainly not the one that is easiest to implement and run. PdM is great for asset-heavy industries that are not afraid of modern technology and have the ability to see the big picture. While upfront costs can be high, it is a great long-term solution that can provide a significant ROI through cost savings and improved machine performance.

If you’re interested in our modular IoT sensor setup or need a deeper clarification of how Limble CMMS integrates with predictive maintenance, don’t hesitate to reach out.

Predictive Maintenance Guides

Predictive Maintenance Tools

Predictive Maintenance Resources