Blog

Unplanned downtime represents one of the most significant challenges facing production facilities. According to the International Society of Automation, between 5% and 20% of production time is lost due to breakdowns, translating to approximately 15 hours of weekly production loss and around 20,000 interruptions annually for the average plant.

"The main point for the big loss is maybe not the equipment itself. It's more or less the production stop," explains Olivier Giertz. "If you have reactive maintenance and suddenly production stops, that could have a big impact because you cannot finish production. It maybe goes further that you cannot deliver on time."

Manufacturing facilities typically employ one of three maintenance approaches:

1. Reactive Maintenance
This approach involves running equipment until failure occurs, then repairing it - similar to driving a car until it breaks down on the motorway. While requiring minimal planning, this strategy leads to the highest costs and most significant production disruptions.

2. Preventive Maintenance
The most common approach today involves scheduled maintenance based on time or usage metrics - comparable to changing a car's oil every 10,000 kilometres regardless of driving conditions. While better than reactive maintenance, this method often results in unnecessary part replacements or missed failures.

3. Predictive Maintenance
The most sophisticated approach analyses how equipment is actually being used, not just how long it has been operating. This strategy can detect subtle changes in equipment performance that indicate potential future failures.
"Predictive maintenance is giving you the advantage. It plans your maintenance schedule, and then you can do it in a controlled way," notes Olivier Giertz.

Modern manufacturing equipment now comes with built-in intelligence that can transform maintenance operations. Our company has developed AI-based predictive maintenance functions integrated directly into their robot controllers and servo systems.

The key innovation is that these systems perform complex data analysis within the components themselves, eliminating the need for external sensors or powerful computing infrastructure.

The AI systems create a model of normal operation for each specific piece of equipment, then continuously monitor for deviations:

For robots: The system analyses movement patterns, load distribution across axes, and subtle vibration changes that might indicate motor problems, bearing wear, or belt loosening.

For servo systems: The technology can detect issues with connected mechanical parts like ball screws, gears, or belts by identifying changes in vibration patterns or current draw that human operators would never notice.

The return on investment for predictive maintenance can be remarkable. Mitsubishi HiTec Paper Europe GmbH in Germany implemented a monitoring system for production fans that paid for itself in just one hour of prevented downtime. With daily production of 300,000 kilograms of specialty paper, even brief interruptions would cost significantly more than the entire predictive maintenance implementation. This demonstrates how critical continuous operation is for manufacturers of specialty products, where production stoppages directly impact both output and quality.

For manufacturers looking to adopt predictive maintenance, experts recommend a gradual approach aligned with Smart Manufacturing Kaizen Level (SMKL) philosophy - a core principle at Mitsubishi Electric that emphasizes implementing changes incrementally through small steps.

- Start with critical production equipment where failures would be most costly
- Implement predictive maintenance on these high-priority assets
- Gradually scale the approach across the entire production facility

"The main point is that the strategy needs to be implemented for predictive maintenance in the factory. So there is a change of thinking. They need to switch from reactive actions to proactive thinking," advises Olivier Giertz.
The technology is accessible; the challenge lies in changing operational mindsets to embrace a proactive approach that promises substantial returns while following the SMKL methodology of continuous, incremental improvement.

As manufacturing continues to evolve, the shift from reactive to predictive maintenance represents a fundamental change in operational philosophy. By leveraging AI-powered insights built directly into production equipment, manufacturers can dramatically reduce downtime, lower maintenance costs, and ensure more reliable supply chains.

The technology to enable this transformation is already available and surprisingly accessible. The real challenge now lies in changing mindsets and embracing a more proactive approach to equipment maintenance - a shift that promises substantial returns for those willing to make the leap.

If you would like to deepen your knowledge on this topic, you can watch the complete webinar here:

From reaction to prediction: AI eliminates downtime I Mitsubishi Electric - YouTube