A fan starts vibrating more than usual. Its speed remains stable, production conditions haven’t changed significantly, and the equipment is still running. Is this a temporary fluctuation, or an early sign of mechanical trouble?
For reliability engineers, detecting abnormal vibration is often only the beginning. The more difficult task is understanding what caused it, whether other equipment is affected, and what maintenance action is justified.
In this simulated cement plant, TDengine’s AI examines a sustained vibration event on Fan 1 and identifies a bearing-temperature change that helps narrow the investigation. The engineer then checks the fan’s SAP maintenance records to assess the lubrication history and plan the next inspection.
The investigation follows the evidence from a comparison of two fans to the AI report and the maintenance review.
Why a Fan Vibration Alarm Needs Further Investigation
In a cement grinding circuit, the mill fan helps maintain airflow for material transport and separation. If the fan trips, the loss of airflow can interrupt the grinding process and potentially shut down the production line. In this simulated plant, the fan is rated at approximately 190 kW, compared with 3 MW for the mill drive. An abnormal vibration reading therefore deserves attention, but the alarm alone does not reveal which maintenance action is needed.
The 3D plant view highlights FAN01 in Grinding Line 01, showing elevated vibration, related equipment measurements, and a direct link to investigate the fan in TDengine IDMP
watch video clip
Compare Fan 1 with a Similar Operating Fan
The condition-monitoring dashboard shows Fan 1 (FAN01) developing higher vibration while its recorded fan speed and mill feed rate remain relatively steady.
Fan 2 does not show the same increase under comparable operating conditions. This comparison directs attention toward Fan 1’s local condition and provides a reference for the next stage of analysis.
Comparing measurements under similar operating conditions is an important part of industrial data contextualization. It helps engineers distinguish equipment-specific changes from variations associated with normal process operation.
Figure 1. Fan 1 shows increasing vibration while Fan 2 remains comparatively stable. Recorded fan speed and mill feed provide context for the comparison
watch video clip
Identify the Sustained Vibration Event
Two configured analyses monitor Fan 1’s vibration: Sustained Deviation and LOF Anomaly Detection.
The engineer selects the event “FAN01 DE Vibration H sustained high” and opens TDengine’s Root Cause Analysis feature. The event description records drive-end horizontal vibration above the operating band for more than two hours under steady mill feed.
Here, DE means drive end, while NDE means non-drive end. They refer to measurement locations on the same fan, not to Fan 1 and Fan 2.
Selecting a specific asset and event provides a defined starting point and a relevant period of operating data for the investigation.
Figure 2. The selected sustained high-vibration event identifies the affected FAN01 measurement and the event being investigated with Root Cause Analysis
watch video clip
Investigate the Possible Causes with AI
After selecting the sustained high-vibration event, the engineer opens TDengine’s Root Cause Analysis report to examine what changed before and during the event.
The AI report identifies a sequence worth investigating: drive-end bearing temperature begins rising approximately 90 minutes before the sustained increase in horizontal vibration. During this period, recorded fan speed and mill feed remain relatively stable.
Based on these observations, the report focuses attention on the drive-end bearing and proposes two possible explanations:
- Localized mechanical degradation
- Lubrication loss
The timing provides a useful clue. A developing bearing or lubrication problem could help explain why the temperature changed before the vibration became abnormal.
Other mechanical conditions, including imbalance, misalignment, looseness, and bearing damage, can also contribute to abnormal vibration. The engineer can distinguish between these possibilities by reviewing the operating evidence, maintenance history, and physical bearing condition.
In this example, AI helps narrow the investigation toward bearing condition and lubrication rather than treating every possible fault as equally likely. TDengine’s AI-powered analysis capabilities support this process by organizing equipment measurements and proposing explanations for engineering review.
Figure 3. The AI report describes a bearing-temperature increase preceding sustained vibration by approximately 90 minutes and identifies localized degradation or lubrication loss as possible causes
watch video clip
Check the Maintenance Records Before Deciding What to Do
The AI findings point toward bearing condition and lubrication. The next step is to examine the equipment’s maintenance history.
In this configured demonstration, SAP PM equipment and work-order records are displayed within TDengine IDMP, allowing the engineer to review maintenance information alongside the operating data.
The maintenance dashboard shows that 47 days have passed since the last drive-end regreasing, against a scheduled interval of 42 days. The lubrication task is therefore five days overdue.
Work Order 4005588 is associated with drive-end regreasing and has status CRTD (Created). Two other orders, 4005620 and 4005679, concern drive-end bearing inspection and a maintenance window. Both are marked REL (Released), which does not mean the work has been completed.
These records give the engineer a specific maintenance gap to investigate. The next step is to review the lubrication task, bearing-inspection work, parts availability, and maintenance window with the maintenance team before deciding on approved work.
The example also illustrates the value of making external maintenance information accessible in an industrial data environment. See TDengine IDMP Opens the Industrial Data Layer for more on using operational data alongside information from other systems.
Figure 4. FAN01’s maintenance records show a 42-day regreasing interval and 47 days since the last service, making the task five days overdue. The related work orders help guide the next inspection
watch video clip
How the Evidence Guides the Investigation
The investigation brings together operating trends, AI analysis, and maintenance records. Each provides useful evidence and points toward a specific next check.
| Evidence | What it suggests | Next verification |
|---|---|---|
| Fan 1 vibration rises while Fan 2 stays relatively steady | Focus on Fan 1’s local condition | Confirm comparable duty, sensor locations, measurement units, and time windows |
| Recorded fan speed and mill feed remain steady | The recorded changes in these parameters do not explain the vibration increase | Review other relevant operating changes before excluding a process contribution |
| AI reports bearing temperature rising before vibration | Bearing condition and lubrication are possible causes | Check the reported sequence and inspect the bearing and lubrication condition |
| Maintenance records show overdue regreasing | A specific maintenance gap requires review | Confirm task status, completion history, and the appropriate inspection or work plan |
Taken together, these observations give the maintenance team a more focused basis for its next action. Following any approved maintenance, the engineer should compare vibration and bearing-temperature trends under similar operating conditions to assess the result.
What This Investigation Shows
In this example, TDengine’s AI helps turn a sustained vibration event into a focused bearing investigation by organizing the operating evidence and proposing possible explanations. The engineer then reviews the maintenance history, including overdue regreasing, to decide which records and physical conditions to verify.
Connecting equipment measurements, event analysis, and maintenance information gives the team a clearer basis for its next action.
The workflow uses TDengine TSDB for industrial time-series data and TDengine IDMP for asset context, dashboards, and AI-assisted investigation.
Explore Industrial AI with TDengine to see how these capabilities support industrial troubleshooting and root cause analysis.
