Solid Strategies for Adopting Predictive Maintenance 

Let’s not beat around the bush: knowing in advance when something is (probably) going to happen (if not exactly when) is objectively better than knowing something will happen, but having no idea when. If you have a timeline, you can plan. If not, you can only try and game the system to stave off whatever it is you’re trying to avoid. And that timeline being helpful is the case for predictive maintenance (PdM).

We’ve written about it before, but as a refresher, PdM is a set of techniques used to determine the condition of in-service equipment to estimate when maintenance should be performed. By predicting when a fault will occur (called “time to failure”), you can make informed decisions, especially lining up important maintenance decisions with production needs, thus minimizing revenue loss and eliminating as much unexpected downtime as possible. Revenue is also saved because PdM allows you to discontinue replacing parts on an arbitrary schedule in favor of when the condition warrants it.

So, PdM sounds pretty great, right? We agree. Pretty much everyone agrees. However, being great does not make something easy to do. There are significant hurdles to effectively implementing PdM. But significant does not mean insurmountable. So let’s get into it!

No size fits all

Change is a contact sport. You’ve heard this before, and you’ve heard it because it’s true. Before you start emailing PDF SOPs to the global inbox, you’ll need to do the legwork. One of the only guarantees in implementing PdM is that if you don’t get buy-in, from the C-suite to the operator, you won’t be wildly successful—and being wildly successful is always helpful. So let’s take a look at a few of the things you’ll need to do to get everyone on board the transition train.

Executives: People tend to focus on what they’re judged on. This is human nature, so knowing (in general) what the C-suite talks about in their reviews (probably revenue & profitability) and showing how PdM will help is pretty important. You’re going to need an executive sponsor, so make a great case. We’ve already shown how PdM can increase revenue and decrease downtime, so build your business case around that. Don’t forget safety: ESG concerns are a powerful and ethical motivator, and PdM can do a lot to increase safety, especially when paired with non-destructive testing (but maybe we’re just partial to NDT).

Site Teams: No two sites are the same, and ignoring the wealth of experience and knowledge of your local teams is one of the biggest mistakes you can make. Before you implement a PdM plan at a site, or in a small part of the site, talk to the engineers and operators and managers that deal with them every day. Find out their pain points, needs and concerns. Figure out—together—how PdM can solve them. Discuss how PdM will affect operations, and if they have a valid concern that it will make things harder, listen to them and try to solve the problem. Eventually, these wrinkles will be ironed out and you’ll have partners in the process, not victims.

Maintenance Staff: While part of the site teams discussed above, maintenance staff deserve a special mention. These are the folks who are going to be asked to change the most, so make sure they have everything they need. Training is going to be especially important in this cohort. Everyone needs to understand their role and have a concrete plan to be successful. Don’t over-promise; there will be hiccups along the way, especially in the first few implementations, so make sure that’s communicated as well.

Clear communication is a big part of any change, so beginning with that in mind is important to the effort of moving to PdM.

Pieces of the PdM Puzzle

It’s impossible to say what all PdM strategies look like; they are as varied as operations because every operation needs a specific PdM plan to meet their needs. However, there are some things that they all possess, so let’s take a look at those.
Past is Present: Historical data is critical for PdM. It is the past that predicts the present. The more historical information, from any source that you can compile, the better understanding you have about the asset in question. Obviously, the time-to-failure curve varies by asset ; for some, failure events are so infrequent it may be difficult to gather information. OEM records may be helpful in these cases. 

Present is Future: The other side of the data coin is what is happening with the asset right now. Gathering this information is the bread and butter of inspectors; the more inspections you can do, the more information you have. Coupled with real-time information from assets, you have a solid indicator of where you stand in the present. 

Combining historical and current data to predict the future is the heart and soul of PdM. This analysis is typically performed by a combination of data scientists and machine learning working together. Current technology is quickly evolving, but neither the engineer nor the algorithm can do the entire job. The tedious work of compiling and collating all the information is best left to machines—we can recommend a good platform to ingest all kinds of data.  The scheduling is best left to humans. Data teams working with maintenance and planning staff can create an excellent roadmap to a safe, healthy plant.

Refine the Process: One thing about PdM: it can always be more efficient. The more historical data you have, the more accurate your predictions will become. PdM is not a “set it and forget it” kind of thing. Teams should always be striving for the best outcome, and that means continually refining the process. Remember all the talk about communicating above? That’s ongoing; the more everyone is talking, the better the results. Each success creates a better environment for the next phase, but only if you learn from what you’ve successfully implemented. The goalposts are always moving.

Eating the Elephant

So how do you get started? One piece at a time, of course—just like eating an elephant. Just like implementing any complicated program. Let’s break it down to the smallest piece: a single asset type.

This asset would serve as a pilot program. Assuming you have buy-in from above and below, start by achieving a successful PdM on a single asset type of your operation. In reality, chances are there’s already some level of PdM going on in almost every plant, so that might be a good place to start if you’re looking for a solid proof-of-concept. Use that success to build momentum. Results often speak for themselves, especially when you are practicing good communications with the teams you’re working with. 

In that spirit, it is helpful if you consider each new implementation as a new project, letting your past work inform your process. Even if you’re pretty sure that asset B is going to go mostly like asset A, or site 2 like site 1, give them the same level of attention. Sites have unique concerns, so it’s important to take them into consideration. This will let you build the largest knowledge base possible—and the greatest chance for success.

Conclusion

Implementing PdM requires a fairly wide variety of personnel and skills, both hard and soft. Understanding that you’ll be creating some disruption in the short term to benefit the future and working with site teams to address such are  the key takeaways, as the technological part and principle of PdM is pretty straightforward in the main. Putting all the pieces together takes time and effort, but leads to a more efficient, more profitable and safer world for everyone.