By nature, the manufacturing industry works to maintain a tightly run organization. Floor managers know what can and should be accomplished in a shift based on the orders and product lines needed to run. Included in this is knowing when and how frequently a line can run, who can keep it running and how to prevent unexpected downtime to the best of everyone’s ability.
How long and effectively a component will continue to run remains a question when it comes to keeping things on track, leaving maintenance of components an unknown. Over the years, the industry has tried several approaches from reactive maintenance, which promotes a “if it’s not broken, don’t fix it” mentality to preventative maintenance that applies service at pre-determined intervals based on known failure rates.
- Works until it no longer works
- Potential for unknown durations of downtime
- Risk of collateral damage to surrounding components
- Based on known failure rates
- Applies historical knowledge, statistical modeling and simulations
- Possibility to plan maintenance
- Potential to lose full use of components
As the manufacturing industry strives to become lean, more companies are applying predictive maintenance, which anticipates upkeep to avoid costs with unscheduled downtime. Unlike reactive maintenance that waits until something breaks and preventative maintenance that may remove components before they reach their full use, predictive maintenance uses data and monitoring to predict and reduce the potential of failure and associated downtime.
The ability to use predictive maintenance increases as manufacturing become more technologically advanced. The ability to monitor the data produced by the devices translates into the need for manufacturers to spend less money replacing parts, maximizing up-time and optimizing the lifetime of components and equipment.
A healthy reliance on data is necessary to successfully implement a predictive maintenance solution. In a best-case scenario, a manufacturer would have the benefit of data prior to, during and post-failure. In additional to data collected from sensors, the information would also include notes taken by staff operating the equipment, environmental data, specifications from that and surrounding machines and run info. Effective data mining and predictive algorithms can also help identify preventable failures.
Questions to Ask:
- Will a failure occur within X hours?
- How many hours remain in the life of the asset?
- Is the asset behaving normally?
- Which asset needs service first?
Over the last couple of years, advances in cloud computing, data analytics, the Internet of Things (interconnection via the Internet of computing devices in everyday objects), and Machine Learning have made predictive maintenance more available. Successful implementation of predictive maintenance also relies on understanding the larger goals of the organization, time frame and desired ROI. Once determined, it will be important to monitor, review and update the key performance indicators and ROI within the predictive maintenance plan to ensure projects remain on task.
This article is brought to you by the edge prep experts at Mutschler Edge Tech/MET. To learn more about our custom edge prep work, contact MET today and speak with our experts about creating an edge-prep solution for you.