Back

How Factory AI Detected a Hidden Failure on a Slow-Moving Asset at One of Australia’s Largest Biscuit Manufacturers

Aug 1, 2025

Predictive Maintenance Success Stories

In manufacturing, the slowest-moving parts often cause the biggest problems—especially when they fail without warning. While many predictive maintenance systems struggle to deliver value on slow-speed equipment, this case study proves that, with the right technology, it's possible to detect early-stage failures even on the quietest assets.

At Factory AI, we understand that food manufacturers run a complex mix of fast and slow equipment. This article tells the story of how our system picked up a failure on a slow-moving rotary cutter bearing at one of Australia’s largest biscuit manufacturing facilities—before it became a costly unplanned breakdown.

Why Slow-Moving Equipment Poses a Unique Challenge in Predictive Maintenance

Not all assets are created equal. In food and beverage manufacturing, many machines run at high speeds—think conveyors, mixers, and packaging equipment. These assets generate strong vibration signatures and heat patterns, making them relatively easier to monitor with traditional condition monitoring tools.

But on the other end of the spectrum are slow-speed assets: gear-driven feeders, rotary cutters, spiral conveyors, and cooling tunnels. These assets often rotate at just a few RPM, generating low-amplitude vibration signals that are much harder to detect.

The stakes, however, remain high. When a slow-speed machine fails, it can bring an entire line to a halt. The failure may only be in a $300 bearing, but the downtime cost could stretch into tens of thousands of dollars—especially in a plant with high throughput and tight schedules.

Common challenges of monitoring slow-speed equipment:

  • Low signal amplitude: Vibration signatures are smaller and harder to distinguish from background noise.
  • Long fault progression cycles: Failures build slowly, often escaping detection until they become serious.
  • Limited monitoring options: Many traditional sensors aren't sensitive enough for low-RPM detection.
  • Underestimated risk: Maintenance teams may deprioritise these assets because they seem less active.

That’s why leading manufacturers are choosing to include slow-speed equipment in their predictive maintenance programs—with the help of always-on, AI-powered monitoring from Factory AI.

Taking a Proactive Approach to Monitoring Low-Speed Assets

One of Australia’s largest biscuit manufacturers runs highly automated production lines with a mix of high- and low-speed assets. While most predictive maintenance efforts tend to focus on the fast-moving machinery, this manufacturer recognised that certain slow-speed machines—such as rotary cutters—were just as critical to uptime.

These cutters play a key role in the shaping and portioning of product before baking. Any issue with a cutter can affect quality, product flow, and downstream equipment. And because cutters are often enclosed and operate slowly, it’s difficult to visually inspect or manually monitor them without taking the line offline.

The company decided to include these slow-speed assets in the scope of their Factory AI rollout. Our sensors—designed for sensitivity and long-term trend analysis—were mounted directly to critical components, including bearings and gear housings.

One of those sensors proved its value just a few weeks later.

Case Study: Detecting a Fault on a Rotary Cutter Bearing

Asset Profile

  • Asset: Rotary Cutter
  • Component: Non-operator side bearing
  • Speed: Low RPM (slow-moving asset)

Detection

The Factory AI system detected a consistent increase in vibration and temperature data on the rotary cutter’s non-operator side bearing. Because our sensors run 24/7 and automatically establish a baseline for each asset, even small deviations were noticeable.

The system generated a real-time alert. The asset wasn’t failing yet—but it was no longer healthy.

Our vibration and temperature anomaly detection algorithms are designed to pick up subtle changes, even on low-speed assets. In this case, both data streams indicated that something was off.

Action Taken

Based on the alert, a technician from the maintenance team submitted a work order to inspect the asset. During the inspection, he found early-stage bearing degradation—not yet visible to the naked eye, but clearly at the beginning of a failure cycle.

He applied Open Gear Lube to two sets of spur gears and provided the following work order feedback:

“Correctly identified impending fault, applied Open Gear Lube to two sets of spur gears. Suggest PM be raised for this weekly.”

This proactive maintenance action restored the asset to a healthy condition and prompted a change in the site’s ongoing preventive maintenance plan.

Estimated Financial Impact

According to the maintenance team, failures on this rotary cutter typically result in 2–3 hours of unplanned downtime. With lost production and labour costs considered, each incident carries an average cost of $25,000.

By catching the issue early, the Factory AI system helped avoid a full-blown failure and its associated financial impact.

What Made Detection Possible?

Detecting faults on slow-moving assets isn’t just about having sensors—it’s about having the right sensors and the right software.

Key enablers in this case:

  • Always-on vibration and temperature sensors: Many condition monitoring programs rely on periodic handheld readings. These miss gradual changes. Factory AI’s sensors collect data every few seconds, building a rich, contextual trendline.
  • Adaptive baselining: Our AI builds a unique normal operating profile for each asset. Deviations from this baseline—no matter how small—trigger alerts.
  • Noise reduction techniques: Our signal processing algorithms are tuned to distinguish true fault signatures from ambient mechanical noise.
  • Low-speed sensitivity: Our sensors are specifically designed to detect anomalies on assets running at <100 RPM.

In short: it wasn’t luck. It was the result of having a system purpose-built to handle challenging monitoring environments.

Why This Matters for Your Maintenance Strategy

This story isn’t just a win for one site. It’s a practical example of what’s possible when slow-speed assets are taken seriously in a predictive maintenance program.

For manufacturers still relying on reactive or calendar-based maintenance for slow-moving equipment, there are three key takeaways:

1. You Can Detect Problems Earlier Than You Think

Even on slow-speed bearings, fault progression leaves a signature. With the right hardware and software in place, these signals can be picked up before they result in downtime.

2. Small Wins Deliver Big Value

The bearing in this case was likely worth a few hundred dollars. But the cost avoidance from preventing a $25,000 breakdown makes this a clear ROI case. In fact, most customers using Factory AI see full return on investment in less than 6 months.

3. Slow-Speed Doesn’t Mean Low-Risk

Just because an asset turns slowly doesn’t mean it should be deprioritised. Quite the opposite—many of these components are linchpins in the production line. If they fail, everything stops.

How Factory AI Supports Low-Speed Asset Monitoring

At Factory AI, we’ve built our solution with slow-moving assets in mind. Here’s how we support reliability teams in food, beverage, and manufacturing environments where slow-speed failures are common:

  • Real-time monitoring of bearings, shafts, and gearboxes
  • Combined vibration and temperature analysis
  • Support for assets under 100 RPM
  • Anomaly detection tuned for low signal environments
  • Clear, human-readable alerts with actionable context
  • Integration with existing maintenance workflows (e.g., work orders, CMMS)

Conclusion: We Won’t Catch Every Failure—But We’re Catching the Right Ones

No predictive maintenance system can guarantee 100% detection—especially on slow-moving, enclosed, or highly variable machinery. And we don’t claim to.

But what we can say with confidence is that our system is already catching failures that would have otherwise gone unnoticed.

In the case of one of Australia’s largest biscuit manufacturers, our solution picked up an early-stage fault that was invisible to manual inspection. The issue was resolved before it turned into downtime. That single event delivered more financial value than months of monitoring costs.

This isn’t theory. It’s a real-world result from a real production site—on an asset just like the one you may be concerned about today.

If your plant includes slow-speed rotary equipment and you’re not yet monitoring it with high-sensitivity, always-on sensors, you’re likely flying blind.

Want to See What We’d Find on Your Site?

If you’re interested in evaluating whether Factory AI could help detect hidden failures on your own slow-speed assets, book a short demo with our team. We’ll show you how it works, walk you through similar use cases, and help you identify where the biggest opportunities for risk reduction may be.

JP Picard

Jean-Philippe Picard

Jean-Philippe Picard is the CEO and Co-Founder of Factory AI. As a positive, transparent, and confident business development leader, he is passionate about helping industrial sites achieve tangible results by focusing on clean, accurate data and prioritizing quick wins. Jean-Philippe has a keen interest in how maintenance strategies evolve and believes in the importance of aligning current practices with a site's future needs, especially with the increasing accessibility of predictive maintenance and AI. He understands the challenges of implementing new technologies, including addressing potential skills and culture gaps within organizations.