What Is Predictive Maintenance? (Types, Examples & Benefits)

In November 2010, Qantas Flight 32 took off on a flight from Singapore to Sydney, Australia. An engine failure shortly after takeoff forced an emergency landing back in Singapore. Luckily, no one was hurt, but several delays and disruptions ensued. The lesson? With a more robust predictive maintenance solution in place, the airline industry could have avoided several headaches. Here’s how.

Predictive Maintenance, Defined

Predictive maintenance (PdM), sometimes referred to as condition-based maintenance, is a proactive maintenance strategy. It monitors asset condition and performance in real time to predict when an asset needs maintenance before it breaks down.

The strategy can uses a variety of sensors, IoT and machine-learning models. Working together, they spot early warning signs of equipment failure. This allows you to schedule maintenance only when your machinery actually needs it. If used frequently and correctly, predictive maintenance can help you avoid costly repairs and unplanned equipment downtime. In the case of having to ground an Airbus A380, we’re talking thousands of dollars.

Purpose of Predictive Maintenance

The goal of predictive maintenance is to increase efficiency, productivity and reduce equipment costs over time. It achieves this by using predictive analytics to determine when maintenance is required, keeping maintenance frequency as low as possible.

Another key goal is to reduce unexpected breakdowns and unplanned downtime compared to traditional, manual maintenance methods. In this sense, predictive maintenance might be viewed as a more advanced type of preventive maintenance. It simply detects faulty equipment conditions more quickly and before the equipment failure occurs.

Predictive Maintenance vs. Preventive Maintenance

However, predictive maintenance (PdM) should not be confused with preventive maintenance (PM), which is also a proactive maintenance strategy with the same end goal. Preventive maintenance differs in that it’s performed based on time (weekly, monthly or annually) or usage (every 1,000 miles used). Because preventive maintenance is done on a set maintenance schedule, it may lead to performing too much maintenance. Alternatively, predictive maintenance requires more advanced steps for accuracy and uses technology to optimize when equipment requires maintenance. Examples include:

• Condition-monitoring devices: Predictive maintenance uses sensors to collect real-time data on equipment health and performance. Depending on the type of predictive maintenance tool you use, information such as temperature, vibration, noise and pressure can be collected.
• Internet of Things:  IoT translates sensor data (think temperature or vibration) into digital signals for analysis. These can then be measured and analyzed over time to make predictions on when maintenance work is required.
• Machine learning: Machine learning uses the sensor data to learn the normal behavior of a piece of equipment, and when it has enough data, it can find anomalies in the performance of said equipment.
• Computerized maintenance management system (CMMS): A CMMS software is crucial for predictive maintenance. It offers both asset management and work order scheduling and tracking to ensure equipment runs as efficiently as possible throughout its lifecycle.

Types of Predictive Maintenance Technologies

To give you a better idea of your options for predictive maintenance, here are a few of the most commonly used technologies:

• Acoustic monitoring: Systems that can imitate the hearing abilities of experienced workers to diagnose malfunctions by sound. This method is commonly used in industrial environments, though it can be difficult to detect certain sounds above the background noise of a facility.
• Infrared technology: This thermography approach involves monitoring equipment temperature. For example, this method may include identifying “hot spots” in electronic equipment, identifying fuses that are nearing capacity and locating faulty terminations in electrical circuits.
• Vibration analysis: This method looks for and monitors significant changes from a machine’s standard vibration. To increase accuracy, typical vibrations should be recorded multiple times so that deviations are more easily noticed.
• Oil analysis: This type of analysis involves testing oil from a machine for viscosity, wear particles and the presence of water, to name a few. This method may be most effective and relevant to the transportation industry, where lubrication of spare parts is key.
• Motor circuit analysis: Used across a variety of industries — from automotive to the marine industry — motor circuit analysis measures a motor’s stator and rotor, in addition to detecting contamination and ground faults. Motor circuit analysis can test new motor inventory before installation of equipment as well as existing motors for system health.

How Much Does Predictive Maintenance Cost?

Sure, preventive maintenance only involves labor costs compared with the high initial capital investment of predictive maintenance. So, you might be wondering whether predictive maintenance is even cost-effective for your business — and what exactly the cost entails. Here’s a breakdown of what it may cost for your business:

• Sensors: Price for sensors depends on the type of sensors you need and the brand that you purchase from. They can range from $100 to thousands of dollars per sensor. For example, a temperature sensor can cost around $100 whereas a vibration sensor can cost around $1,000.
• Software: Depending on your setup, you may need more than one software. As mentioned, a CMMS is an effective tool that will help streamline the predictive maintenance process. It’s typically priced per user and starts around $400 per user a year. The other software you may need is a data analytics tool to collect and analyze the historical data from all your sensors, this can start from around $200 depending on the software you choose.
• Installation: If your technicians don’t have the maintenance techniques and skills required to install the sensors and connect them to the software themselves, then you will need to hire a specialist to install them for you. Depending on the number of sensors you need for data collection and the type of equipment, this can start from the low thousands and up to tens of thousands of dollars.
• Skilled maintenance expert: An experienced maintenance engineer is required to accurately interpret the data that is coming in from the condition-monitoring devices. The salary range for a maintenance engineer from Glassdoor is around $86,000 a year. However, a maintenance engineer may not be necessary for your predictive maintenance program if you have a clear understanding of the failure modes of your equipment.

Benefits of Predictive Maintenance

With the decision to use predictive maintenance for your business comes several benefits. A few of these might include:

• Spending funds only on necessary inspections and repairs, eliminating guesswork and waste
• Fewer or no lost-time incidents due to surprise malfunctions and reactive maintenance repairs
• Maximizes equipment lifespan through timely, necessary maintenance tasks and upkeep
• Increased revenue by protecting your most valuable and mission-critical equipment

Challenges of Predictive Maintenance

While there are many benefits to predictive maintenance work, there are also many disadvantages. A few of them are:

• Cost: With all the technology involved in setting up a predictive maintenance strategy, costs are definitely a factor. Even more than the cost of installing the different tools, predictive maintenance typically also requires skilled maintenance teams who can interpret the data accurately.
• Learning curve and downtime: Once you’ve started using a CMMS or predictive maintenance software and installed the necessary predictive maintenance tools, the relevant staff will have to understand how to use it effectively.
• Possibility of misinterpretation: For the amount you’ll be spending, it’s only fair that you’ll expect 100 percent accurate equipment readings. Unfortunately, this isn’t always the reality. Technology can fail to take essential factors into account when analyzing an asset, such as age of equipment, weather or other operating conditions.

The Return on Investment

Naturally, every business is different, but using predictive maintenance for your most valuable equipment may yield significant ROI rates. In fact, research provided by the U.S. Department of Energy shows that predictive maintenance could result in a 30 to 40 percent savings for the business that implements it.

Moreover, many businesses suffer from unexpected downtime due to equipment malfunctions each year. It’s not so much a question of if as it is when. Even worse (depending on the scale of your business), one hour of downtime can cost anywhere between $108,000 to $2.3 million, with the average amount of downtime lasting around 4 hours. Poof — a million bucks, lost. Sounds like a nightmare, right?

These are only a few key performance indicators (KPIs) that reveal just how valuable a predictive maintenance program is, should the installation costs cover the approximate expenses of your broken machinery. Predictive maintenance helps you plan for (and avoid) malfunctioning machinery, so you can avoid downtime altogether.

Examples of Predictive Maintenance

Predictive maintenance may seem like an abstract concept to some, so let’s break down a few use cases of how this strategy works in a variety of settings:

Refrigeration Sensor

In a restaurant, the health of any food storage or cooking utility is paramount to the business’ success and, therefore, profitability. If a restaurant’s refrigerator were beginning to malfunction or required any type of update, predictive maintenance technology could measure the functionality and alert restaurant staff to any issues. A sensor would be able to evaluate elements of the fridge such as temperature and vibration to determine asset performance. Having this advance notice would enable a maintenance technician to fix the potential problem before it resulted in restaurant equipment downtime.

Power Outage Prevention

Power outages can cause major inconveniences for those they affect. With predictive maintenance technology, you can detect power outages early and avoid them completely. Sensors would again come into play in this scenario to provide insight into assets using artificial intelligence. This real-time data informs companies within the energy industry when equipment is likely to fail.

Oil & Gas Industry

The oil and gas industry often lacks visibility into the condition of equipment in remote, offshore places. Maintenance technicians would visit these sites based on a time interval to inspect the condition of the equipment even when it’s not necessary. With predictive maintenance, oil and gas companies can evaluate the health and performance of their equipment and only schedule maintenance upon detection of an equipment health issue.

Building Management

Building managers can use predictive maintenance technology to monitor buildings and equipment remotely. This smart system would enable owners and managers to keep the building environment within a certain temperature range, track humidity and more. This monitoring can help improve overall energy cost savings for the building.

Manufacturing Monitoring

Since a manufacturing plant tends to have many costly assets and valuable equipment, they might invest in infrared imagers to monitor certain elements of assets, such as temperature, to prevent overheating. This system of predictive maintenance helps plants avoid overusing essential equipment, pushing machinery to the point of disruptive breakdowns. In fact, one study showed manufacturers using predictive maintenance saw an 87 percent reduction in equipment defects.

Aircraft Maintenance

As shown in the Qantas Flight 32 example, if an airplane has an unexpected engine failure, it can lead to delays and incur big operational costs. Usually aircraft maintenance relies on a set schedule of maintenance for inspections and repairs. However, using sensors and flight data for predictive maintenance allows maintenance teams to determine when irregularities may occur and schedule planned maintenance ahead of time.

Is Predictive Maintenance Worth It?

Predictive maintenance requires a high capital investment in technology and labor to implement. If a business does not have much expensive equipment or strict health and safety standards, then they should consider a preventive maintenance program rather than a predictive maintenance one.

Here are some questions your business should consider before you decide to implement a predictive maintenance program:

• What is the approximate value of our equipment/machinery?
• Is it necessary for our machinery?
• What has the equipment history been like thus far?
• What do our records show regarding equipment downtime, defects, losses and safety threats?
• What sort of predictive maintenance tools would be best for our business’ needs?
• Is my business financially able to invest in these kinds of data-driven technologies or experts at this time?

How to Implement a Plan for Your Business

Implementing a predictive maintenance plan in your business requires that you first identify a list of critical assets for the program’s implementation. Afterward, establish a baseline of all identified equipment, and monitor each piece of machinery regularly to assess condition. This will help you determine if machinery is malfunctioning.

After identifying your most valuable assets and equipment, here are additional steps to help you begin applying predictive maintenance to your business:

• Step 1: Use pre-existing data to analyze your equipment. Collect the data from sources such as maintenance records, a CMMS or data that your machine already produces for future incident logging or historical analysis. This data will give you actionable information about the machine behavior and will set a baseline for the program.
• Step 2: Identify failure modes. Perform an analysis of the critical assets that you’ve chosen to establish failure modes. You should focus on the frequency of failures, severity of machine failure and the difficulty of identifying failure.
• Step 3: Install IoT sensors on your equipment. The failure modes you identified for a piece of equipment will help you determine the type of sensors to install. For instance, a rotating equipment will need a vibration meter to detect the common faults a rotating equipment is prone to.
• Step 4: Work with one series of algorithms, and watch for improvement. Apply one standard algorithm to all of your equipment rather than to each one. Over time, these algorithms will automatically improve, optimizing performance as they accumulate more data.
• Step 5: Analyze data in bulk. Cloud-based technology enables efficient data readings for hundreds of machines at a time. Rather than bombarding or retraining your engineers to crunch these numbers, let artificial intelligence get the job done faster for you.
• Step 6: Set up alerts for detections. Connect your condition-monitoring devices to a CMMS to alert your technician of an anomaly in equipment.