Predictive Maintenance, do you really need it? Maybe, but first you need to know what it is and how it works. Then you need to look at your own situation and decide what's going to work best for you.
Let's start with the basics: a solid understanding of predictive maintenance.
What is predictive maintenance?
When it's working well, it feels like magic-crystal-ball maintenance. With the power to see into the future, you can perform maintenance just before problems are about to happen, reducing unplanned downtime to basically zero. But it's not magic. It's math.
You start by putting sensors on assets to collect data. It depends on the asset, but you can monitor vibration, temperature, noise, pressure, oil and lubricant levels, and even electrical currents. Once the data starts pouring in, you start pushing it through complex algorithms. Eventually, reliable predictions comes out when your assets are going to have problems. And because you know when these problems are going to happen, you can schedule maintenance just before they do.
How is it different than condition-based maintenance?
Because both systems use sensors, it's easy to get them confused. The big difference is how the sensor data is used. With condition-based maintenance, you're always comparing the current sensor data against a predetermined "sweet spot." Once the sensor notices the asset is not performing the way it should, it sets off an alarm (which could be everything from sirens and flashing red lights to sending you a text or email through your CMMS software), and you perform maintenance. You have a fan that should be spinning at a certain rate. Once it's going too fast or too slow, the alarm sounds, and maintenance is performed. With predictive maintenance, the system picks up data that suggests the fan is going to spin at the wrong speed in the future and gives you a heads-up.
What are the pros and cons?
You can probably guess a lot of these because, in a very real way, predictive maintenance sounds like you're getting a super power. But there are some advantages you might not have considered. And there are, in fact, disadvantages.
There are a lot of good reasons to adopt predictive maintenance, including:
Maintenance gets done just before it's needed, so you could drastically reduce your unplanned downtime. Ideally, it would be very close to zero.
You're only doing the maintenance that actually needs to get done. If you're using preventive maintenance, you might be doing work a bit too early or a bit too often, wasting time, parts, and materials. "Better safe than sorry" has its practical limits.
Modern sensors are pretty tough, and can go in places where it's not safe for a technician to poke around too often. At the top of a stack or behind steam pipes? Sensors save technicians from time-consuming shutdown and lockouts.
Of course, it's not perfect.
There are some drawbacks to consider, including:
It's going to cost you money upfront. Sensors are not cheap, and you'll likely need someone to come in and set them up for you. Then you'll need to pay for your staff to get trained on how to use them.
There are ongoing costs, too. Every so often the sensors will need to be re-calibrated. And more importantly, you'll likely need to pay someone to take all that collected data and make sense of it.
Because you're only scheduling maintenance just before it's needed, it's trickier to set up a regular schedule for your technicians. Instead of saying, "We're going to replace this gasket every two months, so let's make sure we have the right parts and people on hand the first Monday of every second month," now you're saying, "We'll change the gasket when the sensor tells us to, whenever that is."
How do I know if it's right for me?
You need to look at unplanned downtime costs and predictive maintenance program costs and weigh the numbers against one another. Even if you do decide to go with predictive maintenance, it could be the case that you'll want a mixed maintenance program, where some assets are just on PMs and others are covered by predictive maintenance software.
Which leads us to the next question: With a mix maintenance program, which assets are worth covering with predictive maintenance? It'll vary from industry to industry, and the size of your operation will also be a factor. So, let's take an example that's not industry-specific and look at some of what goes into the decision-making process.
What can my car tell me about predictive maintenance?
If you could add predictive maintenance sensors and software to your car, where would you put them?
With the headlights, it's best to stick with run-to-failure maintenance. Keep using them until they die, which is fine because they're not especially expensive, and because you have two of them, when one dies you can still drive the car to the store to buy a replacement. For a relatively cheap part with a low criticality, it doesn't make sense to use predictive maintenance.
With the tires, it's fine to stick with condition-based maintenance. The old-school version is the visual inspection. Or, when you're on a straight stretch of road, you can let go of the steering wheel, checking to see if the car pulls to one side or the other. A lot of more recent models have built-in air pressure sensors, but it's still just condition-based maintenance. Once the air pressure gets too low, a warning light appears on the dash. But the sensor is not going to predict when the tire is going to lose air. If you do get a completely flat tire, unplanned downtime, you can swap in the spare by yourself at the side of the road. It's inconvenient, but not going to take you too long.
But with the engine and electrical systems, if you could have predictive maintenance, this is where you'd want it. The engine and the electrical systems have a very high criticality. Once they stop working, it's time to call the tow truck. And on the way to the garage, make sure you stop by the bank. Because these systems are critical to keeping the car running and because fixing them often takes a lot of time and money, it makes sense to look at swallowing the upfront and ongoing costs of predictive maintenance. You're going to save money in the long run.
If I don't need predictive maintenance now, does that mean I'll never need it?
Remember, you need to compare the cost of the system against the cost of breakdowns. So, as the cost of the hardware goes down, and this is something that always happens with new technology, predictive maintenance might start to make more and more sense for you. Consider the now-humble CD. The first player was introduced in 1982 and sold for $1000. By 1984, you could buy a portable player for just $300. Currently, you can buy them on Amazon for as low as $20. The predictive maintenance sensors and software will come down in price, and when they do, they might make more sense for your operation. But no matter how low prices fall, there are always going to be assets where it will almost never make sense to use predictive maintenance. Headlights, for example, are always going to be cheap enough and easy enough to change that you're never going to bother hooking them up to a sensor to be able to predict when they're about to burn out. Even if you did know roughly when a headlight was going to die, would you even bother to replace it early?
Before you can make any decisions about predictive maintenance, you need to know your current costs for downtime for different assets. And the best way to get a handle on these numbers is with a good CMMS. One that's user-friendly so your technicians will actually use it. One that's mobile and backed by cloud computing so you can drop in and pull out data from anywhere, in real-time. And one that's packed with auto generated reports so once the data is in, you can easily leverage it into actionable insights. If you don't have a CMMS or you have one but it's not what you'd hoped for, now's the time to reach out and start talking with providers.