Until recently, the majority of companies followed a time-based maintenance strategy to prevent unplanned downtime. The only required input was calendar or usage time, which made it relatively easy and cheap to implement, since no additional sensors were required. Now that sensors have become cheaper, and machines are connected via the Internet of Things (read everything about monitoring your asset here), opportunities for predictive maintenance have emerged. Predictive analytics estimates future behavior of an asset by knowing when the probability of a breakdown is highest and generating service orders to prevent unnecessary and unplanned downtime.
Unplanned downtime has significant consequences for original equipment manufacturers (OEMs) and machine operators. It leads to unexpected revenue loss and maintenance costs, as well as a breach in trust between OEMs and their customers – the operators. For these reasons, more OEMs and operators are changing their maintenance strategy from reactive to proactive. To do so, condition information has to be combined with machine learning algorithms to implement predictive maintenance. Predictive maintenance gives insight in the future condition of their asset and allows them to act upon it to reduce unplanned downtime. Being able to indicate when and where a machine needs maintenance allows OEMs to guarantee a certain service level to their customers and machine operators to be better prepared for downtime.
For maintenance, a dilemma exists. If preventive maintenance is performed too early, good parts will go to waste, even when they had remaining useful life. In the case of corrective maintenance, costs can rise from the unexpected downtime, the sudden purchase of parts, and the unexpected loss of revenue. By using predictive maintenance, the optimal moment is selected to cause the least chance of downtime while optimising the components’ remaining useful lifetime.
Being able to pinpoint when maintenance needs to take place, aka the moment when certain parts are predicted to fail, has the additional advantage of allowing service engineers to manage spare parts and allocate service hours more efficiently. This type of maintenance is made possible by applying predictive analytics on condition information. To implement predictive maintenance, the condition of the asset has to be closely monitored to get insight into how its components work and why they fail. This process requires time and research, so an initial investment has to be done. However, the benefits are large. Not only does it give OEMs and operators a better understanding of their machines, they also save unnecessary maintenance costs and diminish unplanned downtime.
AI-driven algorithms continuously monitor machine conditions and suggest appropriate actions when necessary. They give the input for service planning: they can automatically generate service orders for service engineers indicating what components should be maintained and when those orders are critical. With machine learning, these predictions learn based on feedback from actuals to become more accurate over time. Algorithms bring speed, safety and accuracy to the maintenance strategy, allowing OEMs and operators to make decisions based on data and make the most out of their assets’ health. When machine reliability can be guaranteed, machine performance can be optimised to achieve higher operational efficiency, allowing OEMs and operators to further improve their machines.
Widget Brain can help OEMs and machine operators to realise predictive maintenance. We provide the platform (ALFA) and the algorithms to make machines smarter and maintenance more efficient. Want to know more about the Asset Health possibilities for your business and our intelligent algorithms? Contact us today at www.widgetbrain.com/getstarted/ and stay ahead of the pack.
Plug in intelligence with Widget Brain