I don’t need to sell you on the value of predictive maintenance.
If you can fix a machine before it breaks, the benefits speak for themselves: you don’t have a broken, unusable machine. Predictive maintenance, or using real-time information to predict when a piece of machinery might break, can drastically reduce downtime. For instance, a ConocoPhillips facility achieved a 91% plant availability rate, and dropped maintenance costs by 50% when they switched to predictive maintenance. Similarly, another facility used multivariate predictive maintenance to save roughly $650,000 a year, “running more generators at lower power.”
- Downtime costs $22,000 per minute, on average, according to Nielsen Research.
- Unplanned maintenance costs 2-5 times more than planned maintenance.
What I do need to sell you on is whether it’s the right solution for your facility. With the right computerized maintenance management tools and maintenance strategy, you’ll have a better idea of when problems are coming.
A predictive maintenance (sometimes abbreviated PdM) strategy is a journey, and a difficult one. Not all businesses will benefit equally from predictive maintenance. But, if you do need a predictive maintenance strategy and don’t invest in one, you could be looking at a lot of downtime, and lost money. If you don’t need a predictive strategy, however, the resources you’ll expend getting one up and running will be misspent. Predictive maintenance requires that you have a computerized maintenance management system, or some means of tracking your assets’ data, the right IoT sensors to feed data to the CMMS, and engineers who are able to keep the system running.
Check out these steps to see if you need a predictive maintenance strategy. They’ll tell you whether or not your business is the sort that would benefit from predictive maintenance, and whether or not your business would be able to keep up a predictive strategy.
Step 1: Know thyself.
“Know thyself” was engraved in the temple at Delphi for a reason: It’s the start of all other knowledge.
Some industries are a better fit for a predictive maintenance strategy. If your business is asset intensive, like utilities, healthcare, fleets, airports, factories or facilities, predictive maintenance could be a big help.
I call this problem the bearable/breakable issue. How bearable is it if an asset breaks (think an apartment complex’s AC going out in early August, or a cereal factory unable to fill its boxes)? If a break is unbearable, you should consider predictive maintenance. If you can manage without equipment for several hours, or even days, predictive maintenance will be too much of an investment. For instance, if you don’t have assets that run continuously, PdM is not as essential. “Predictive maintenance is good for assets that are in the process of breaking down,” says Bryan Christiansen. In other words, machinery like that in an assembly line, where every one is necessary, is a good choice for predictive maintenance.
If you do know that PdM is right for you, then you also need to know what software, sensors and technician knowledge you’ll need. You’ll need to know which assets you want, or need, to track.
“Before implementing a predictive maintenance program, identify the equipment most likely to fail and assess the costs, and urgency, of equipment breakdowns,” suggests Jeff Conner of Control Concepts. You’ll also want to review your assets in terms of importance: Are some assets more vital than others? How often have those assets broken down? How much money does that downtime cost you? If you know a piece of necessary machinery is prone to breakdown, a predictive strategy could extend its life. The costs of equipment failure can be extreme: in one study, a PdM strategy helped a facility to save $235,365. According to maintenance expert Keith Mobley, predictive maintenance can reduce your maintenance costs, whatever they are, by 50%. If you’re interested in checking on how much money you could save with PdM, consider running a failure avoidance cost matrix, like this one, that will compare the amount you’ll spend on predictive maintenance versus the amount you’d pay if an asset breaks.
Bryan Christiansen of Limble CMMS adds that you should make sure an asset is one that fits into a predictive strategy. “Before setting up a predictive maintenance program, get a good list of equipment you’re tracking, and find the failure characteristics and see if there’s one you can track,” he says.
Get a sense of what your informational resources, too, both computer and human.
If you’re already monitoring information like temperature and vibrations with a CMMS, you might have enough to start with. “In many of the cases we’ve worked on, we’ve talked to the facilities manager, and and they can start predictive maintenance with the data they have,” says Sumant Kawale of SparkCognition. “In most cases that I’ve seen, data is highly underused.” Out of the 10-20 variables some facilities collect, Kawale notes, many often monitor only one or two. “If you can bring your unused data to light, that’s the fist step” in determining if a predictive maintenance strategy is right for you. If you don’t have your data organized in a fashion where you can get to it, that organization will be the first step you want to take.
As for how much data you should check before getting started, Mike Johnston at Plant Engineering suggests you check at two or three years of data to get a full picture.
Establishing a PdM strategy is one of many situations where having the right CMMS software can help make your plant more efficient. Vendors like eMaint, Bigfoot, and mPulse all offer predictive maintenance capabilities. If you’ve got software in place that already organizes your assets and schedules, determining what you need is even easier.
Step 2: Make Sure You’ve Got The Right Team
Predictive maintenance is great, but successful PdM implementation needs the right team.
You’ll need people who know the machinery and its failure codes to provide the insight you need for a predictive strategy.
Your maintenance team will also need trust between the people who will be working with the technological side of the plan, as well as the techs who’ll enact it in the field. At utility company Salt River Project in Arizona, supervisor Andy Johnson understood the importance of understanding between roles. “Sometimes you worry, are the plants going to trust you?” asked Johnson. “One thing we’ve been very conscious about is building that trust.” In Johnson and Salt River Project’s case, that meant building their maintenance models in-house, among a familiar staff.
Whether you choose to do that or not, knowing your workers are assured of the overall worth of the plan is important. Bryan Christiansen also adds that a strategy shouldn’t replace the people working with it. “Predictive maintenance is not a replacement for a good technician,” he notes. “Data’s great, but if you don’t have someone to make those decisions, the maintenance strategy won’t be as successful.”
Step 3: Make Sure To Have The Right Sensors
While more recent equipment may have sensors already installed, don’t assume your machinery does. For a predictive maintenance program to work, you’re going to need the right sensors to track the information that makes predictive maintenance possible.
You’ll want to figure out how your predictive maintenance strategy would work, small-scale, before you embark on a larger strategy. Kawale suggests, “It’s best to do a pilot project and only sensor one piece of equipment,” so you can get an initial idea of how predictive maintenance will look on a larger scale.
If you’re interested in what kinds of sensors your business needs, firms like National Instruments can help you determine what would work best for your situation.
Once you’ve got the right sensors in place, a CMMS like Limble can help you structure when, and how, you repair. “As long as you’ve got an IoT sensor that can send data to Limble, it can update and track that asset over time,” says Christiansen. Moreover, Limble can be set up to send alerts, or even work orders, to technicians when a sensor reading hits a certain point. Those notifications can be set up with further instructions, whether you want your tech to fix, monitor, or diagnose a problem.
Data analytics companies can also help you get benefits from your data. For example, SparkCognition practices what they call cognitive analytics with plants and facilities’ data, helping them determine money-saving strategies. One of their customers, a small beverage bottling plant, demonstrated the success that comes with predictive maintenance. The bottling plant had a breakdown-prone piece of equipment. “A bottling’s plant’s like an assembly line, with a lot of things happening—if anything in that chain comes down, the entire plant comes down,” said Kawale. In other words, downtime for one asset is downtime for the entire business.
With SparkCognition’s help, the plant installed an array of sensors that showed them the frequent breakdowns were the result of a mixture of temperature and humidity. Those same sensors, and the data they provided, then became the basis of a predictive maintenance strategy. “When they saw the facility was entering those temperature or humidity zones, they’d know to change things to avoid further downtime,” added Kawale. Thanks to predictive maintenance, conditions never got in the way of business.
If you’re further interested in setting up a predictive maintenance plan, check out this predictive maintenance template from Microsoft. It’ll give you a basic idea of what you’ll need to start your predictive maintenance journey.
Your predictive maintenance journey
Have you used, or are you using, a predictive maintenance strategy? If so, let me know in the comments below, and let other readers know how your journey from reactive to predictive maintenance went!
Looking for Fleet Management software? Check out Capterra's list of the best Fleet Management software solutions.