From fracture to fix in 40 minutes – a midnight success story for a cement roller press

Case Source: Online monitoring project of a large cement group in Henan Province, China – Cement Line #1 roller press, August 2, 2025.

Challenge

• Customer & Industry: A large cement group, core roller press grinding system.

• Equipment Overview: High-speed input shaft bearing of the roller press main reduction gearbox.

• Fault Characteristics – Cage Fracture: 

        ￮ Occurred at 00:50 – a critical period with minimal staffing and lowest manual inspection efficiency. 

        ￮ Once cage fracture happens, if not immediately intervened, it can cause the gearbox to seize and be scrapped within an extremely short time – 

        ￮ affecting normal production of the entire line, leading to greater economic loss or a serious safety accident.

• Limitations of Low-frequency Monitoring: 

        ￮ Usually collect data every one or two hours, even at longer intervals. 

        ￮ Late at night, if the system fails to capture fault characteristics during the rapid deterioration phase, a missed detection may occur, causing the best stopping opportunity to be lost and leading to extremely severe losses.

Solution

To address this pain point, we deployed a new-generation online vibration monitoring system that performs 24/7 online monitoring tasks comprehensively and efficiently. Built upon:

2.1 5-min high-density index data sampling – including RMS values, temperature, etc.

For sudden faults that deteriorate quickly (e.g., shutdown within a few hours), low-frequency collection may miss detections or provide insufficient data.

While high-density sampling can obtain multiple sets of detailed data (e.g., 8 sets) in a short time, reducing false alarm rates through cross-validation of multi-point features.

Even if a single data set shows suspected characteristics, the system can trigger an alarm in real time without waiting for all sampling cycles to complete, ensuring timely response.

At the same time, multiple data sets capture the fault evolution process more completely, providing a solid basis for diagnostic analysis.

2.2 Anomaly-triggered waveform collection process

The ronds vibration sensor system continuously monitors online, vibration trends are stable before 00:50.

From 00:50, the short‑term index increased from 18.053 m/s² to 42.892 m/s² in 5 minutes. The intelligent algorithm flagged an abnormal rate of change. Following this short-term index anomaly point, additional waveform collection was triggered, enabling further manual fault confirmation.

2.3 Comprehensive index system

We do not rely solely on RMS value indexes, we also have a comprehensive index system based on rotational speed calculations, such as cage-related indexes. Since there are many scenarios where RMS values are insensitive, a more comprehensive index system is needed as a second line of defense. This second line of defense uses precise indexes, which not only reduce false alarms but also serve as the final barrier against missed detections.

Diagnostic Process – Emergency Intervention at Midnight

• 00:50 The short-term index shows a significant vibration rise.

• 00:57 Level 3 high-severity alarm is triggered.

• 01:22 AI intelligent diagnosis + engineer confirmation, two-pronged rapid identification and fault review.

        ￮ The high-frequency acceleration RMS value of the reducer input shaft increased.

        ￮ From the high-frequency acceleration time-domain waveform, interval impacts from the input shaft bearing cage can be observed.

        ￮ Envelope demodulation of the noise energy in the high-frequency acceleration spectrum reveals the bearing cage fault frequency and its harmonics.

• 01:30 Based on the diagnostic results, the diagnostic engineer immediately notified the on-site maintenance team to stop the equipment as soon as possible and arrange an inspection of the gearbox high-speed input shaft bearing for damage.

• Validation: After receiving our alert, the customer arranged a safe shutdown. Inspection revealed a fractured cage on the reduction gearbox high-speed shaft bearing, which was replaced.

ROI

• Critical Intervention Time:  Through automated response at midnight, the system provided approximately 30 minutes of valuable intervention time, saving the equipment.

• Catastrophic Consequences Avoided: Successfully shut down safely before the high-speed bearing seized, avoiding:

        ￮ Gearbox scrapping and replacement cost.

        ￮ Emergency repair costs and non-continuous production losses caused by unplanned downtime.

• Total ROI: Successfully avoided multi-million-dollar catastrophic losses.

Conclusion

“In this critical case that occurred at 00:50 midnight, this comprehensive sampling strategy serves as the ‘nighttime guardian’ of us. It ensures that, at any time, the system’s diagnostic efficiency can match the fastest fault evolution. Facing the rapid deterioration of high‑speed bearing faults, this comprehensive sampling strategy + 24/7 non‑stop monitoring transforms ‘passive firefighting’ into ‘immediate intervention’ – a key factor in ensuring 24/7 continuous operation of our cement production line.” 

— Senior manager, cement plant