In this post, we make the point that evaluating predictive maintenance feasibility should focus on asset criticality and potential production impact rather than asset price.
Given that reliability leaders are focused (rightly so) on ensuring every dollar invested offers a meaningful payback, we’re writing a series of articles to help you understand how you can evaluate predictive maintenance software, such as ours at Factory AI.
In the past, we’ve discussed three simple steps to calculate whether smart, continuous condition monitoring software with intelligence could help. We've also shown how you can use the simple failure history method as a quick back-of-the-napkin trick to decide whether you should explore predictive maintenance further.
In this article, we want to address a common misconception about evaluating predictive maintenance.
We are all familiar with asset criticality models. To classify assets by criticality, we essentially ask, “What is the asset’s effect on production output?” You would then consider safety criticality, production impact, repair time/cost, and environmental impact to help you make your decision.
There are more layers to this, but in reality, most of the sites we work with end up using a three-tier approach where they classify assets as either:
When you look at a site’s asset criticality document, you will notice that criticality and asset price aren’t directly correlated.
You will likely find critical assets that are cheaper than some essential or general-purpose assets. This is typically true for assets like pumps that are needed to power nearly all the production lines and may not be as technically expensive as some cooling tower fans (if you consider the control unit and all bells and whistles).
Asset criticality is a better proxy for the potential value you would get from using predictive maintenance software, rather than simply the price of an asset.
This is because the real pain of unplanned downtime isn’t so much related to the value of the asset causing the downtime. The pain is more associated with the impact of production losses due to the downtime.
In their “Predictive Maintenance Market: 5 Highlights for 2024 and Beyond” report, IoT Analytics writes that “the median unplanned downtime cost across 11 industries is approximately $125,000 per hour.”
For many of our customers, this figure is overblown. Their true hourly downtime cost is more around the $10,000 mark. Still, if a small motor on a conveyor goes down and halts production for two hours, that’s $20,000 out the window. Spending $750 per year to monitor this asset (which is the type of asset we have a history of accurately predicting failures for) would make sense. This is true even if the value of such a motor is only $4,000.
With this simple example, you can see why the price of an asset is not a great proxy for determining whether it makes sense to include such an asset as part of your predictive maintenance program.
Moreover, companies that offer predictive maintenance software platforms, such as Factory AI, should be able to provide you with multiple smart monitoring strategies. For some sites, it makes sense to instrument highly critical assets with more, higher-quality sensors to better capture diagnostic information when parameters start to deviate from the normal healthy range. For more general-purpose assets that are smaller and have simple operating profiles, a very simple setup might work just fine.
If you’d like to stress test your predictive maintenance plan with someone who’s helped multiple food and beverage sites implement them successfully, even if simply for advice, we’re here to help.
Speak soon.