Predictive maintenance is the initiative to predict (major) failures or breakdowns within specific equipment that will significantly impact the business. In this age of digital transformation, that business concept is now being powered by machine learning and AI. While just about every organization wants to begin or further their digital transformation journey so it may operate with greater intelligence, most are just not prepared to dive right in. There’s no better example of this than the industry that wishes to leverage predictive maintenance solutions.
Just about every company in the energy industry is old – some even breaking the century mark! It’s to be expected then, that the thousands of units of heavy machinery utilized on a daily basis break down many times during their lifetime. Sometimes the breakage needs only a quick fix; sometimes it causes work stoppage and costs tens or hundreds of thousands of dollars.
Understandably, every company wants to avoid accruing such costs. To do so, these energy companies employ predictive maintenance. But for companies invested in legacy machinery and hardware – such as major oil rigs or wind turbines – implementing a digital predictive maintenance solution can be cumbersome. So how do you know if your organization is prepared to adopt a predictive maintenance solution that will provide greatly improved forecasting and analytics, let alone make the transition a smooth one?
Pictured at the top is a basic road map our own experts at Neal Analytics uses to show to-be-adopters what they can expect. One major hurdle the majority of companies face is a mental one: expectations. Many think the process is simply for a consultant to create a custom solution and for it to be implemented. That’s it. If you live downtown and purchase a Tesla with nowhere to park it and no charging station within 200 miles, what’s the use? Likewise, for a solution to be accurate and effective – meaning not a big waste of money – there are prerequisites.
The most important one: it has to be fed clean, detailed datasets. The more the better! These datasets MUST contain failures as well. After all, that’s what we’re predicting with our solution. Companies often think their data is ready and we often have to be the messenger bearing bad news. And to maintain a minimum standard of excellence, our consultants recommend these failures to number in the thousands. This may sound extreme, but remember that there are typically dozens of machines each decades old.
The data points have to be as precise as possible – as in each machine needs to report a data point for each second of the day. Remember, the more accurate the data you input, the more accurate the forecasted output. If you do not have the capability, that’s where the next major hurdle comes in: sensors.
For accurate reporting, you need accurate sensors. Yours may very well do the job! But that’s one of the things that needs to be inspected. Once we make sure your datasets are satisfactory and you have sensors fulfilling all the needs of the predictive maintenance solution, you’re ready to move on in the road map! Of course, these are very general steps to get you started on your journey of digital predictive maintenance. If you have any questions or are having trouble implementing your solution, please don’t hesitate to reach out!