Vibration Analysis for Induction Motors: From an Abnormal Trend to a Maintenance Decision

What an abnormal vibration trend can tell you

Rotating equipment produces vibration during operation. A repeatable change can give the maintenance team a reason to investigate, but a single overall value does not identify the cause. Imbalance, misalignment, looseness and bearing wear are among the conditions discussed in Fluke’s overview of common vibration faults.

Treat “abnormal” as a comparison that needs an explanation: abnormal relative to which baseline, under what operating conditions and with what measurement method? That framing helps prevent an alert from becoming an unsupported diagnosis.

A trend plot shows change over time. A spectrum, when available with appropriate acquisition settings, helps a trained analyst examine the frequency content of a measurement. Request the underlying measurement information and suitable analysis before assigning a specific defect label. Confirm what the selected monitoring configuration actually provides.

Step 1: validate the measurement before diagnosing the motor

Begin with the record itself. Confirm the asset tag, timestamp, measurement point, orientation, units and acquisition settings. Check whether mounting, sensor position, settings or firmware changed around the time of the alert.

Request comparable repeat measurements where appropriate. Do not merge readings from different directions or locations without preserving those distinctions. Fluke’s explanation of triaxial vibration data describes why directional information matters to fault interpretation.

• Is this a valid reading from the intended motor?
• Is the sensor securely mounted using the agreed method?
• Are there data gaps, clipping indications or configuration changes to investigate?
• Does a repeat reading under comparable conditions show the same change?

Missing data should trigger a data-quality investigation, not a “healthy” status. If the observation cannot be trusted, fix the measurement problem before using it to justify maintenance.

Step 2: compare equivalent operating conditions

Record the motor’s speed, load and operating state when the observation occurred. Note startup, shutdown, process transitions and recent maintenance. For a variable-speed asset, retain the speed information rather than assuming that readings collected at different speeds describe the same condition.

The driven machine matters. Ask the operator what changed in the pump, blower, crusher or compressor duty. Compare the electrical and temperature records where available, while recognising that their measurement locations and response characteristics differ.

A baseline assembled from mixed operating states may hide a useful change or create repeated alerts. Establish comparison groups that the plant team can explain, and record any operating state that has too little data for a confident comparison.

Step 3: build a short list of possible causes

Use the observation to choose checks, rather than to declare a failed part. The following table is an investigation aid, not a diagnostic lookup table. Similar observations can have different causes, and a motor can have more than one issue.

Ask the analyst to write down the leading hypothesis, the alternatives and the evidence that would separate them. If a bearing-related diagnosis is proposed, for example, ask what bearing information and measurement detail support it. If the needed data are unavailable, keep the conclusion provisional.

Avoid applying a universal severity number from an unrelated machine. The applicable evaluation criteria depend on the equipment and measurement method. Agree on the relevant OEM guidance, standards and plant procedures with a competent specialist.

Step 4: make a maintenance decision with an owner

Separate the technical finding from the operating decision. The analyst explains the evidence and uncertainty; the authorised plant team weighs the operating risk, available protection, redundancy, spares and access to a safe maintenance window.

This workflow does not prescribe a shutdown threshold. Existing protection and the plant’s safety and operating procedures take precedence. Do not approach moving components, remove guards or open electrical panels to investigate an alert without the required safe-work controls.

A worked example: a pump trend changes after service

Illustrative example, not a measured Augur Forge customer result. A maintenance team sees a sustained vibration change on an induction motor driving a cooling-water pump. The motor has recently returned from planned service.

First, the team checks the asset tag, sensor position and mounting record. It then compares periods with similar operating duty and confirms that the change persists. The operator records whether the process conditions changed, and the team reviews the service report.

The analyst proposes an inspection informed by those records. At this stage, “bearing failure” would be premature: the evidence has established a repeatable change, not its cause. The plant team assigns an owner, arranges the required access and records what the inspection should resolve.

If inspection identifies a correctable condition, the team records the finding and the work performed. If it does not, the investigation remains open with a revised evidence plan. Both outcomes are more useful than silently clearing the alert.

Step 5: verify the result and improve the next investigation

After corrective work, collect comparable measurements and preserve the before-and-after record. Document the operating conditions and any sensor changes so that the comparison remains meaningful. A lower overall value can support a review, but it should be interpreted alongside the completed work and other acceptance checks.

Close the work order with the confirmed finding, action, outcome and unresolved questions. If no defect was found, preserve that result too. Tracking inconclusive or unnecessary investigations helps the team assess monitoring quality and workload.

For a monitoring pilot, count confirmed findings separately from alerts and projected avoided failures. This makes the final review understandable to maintenance, production and finance. Our practical guide for plant maintenance teams explains the wider programme; the Augur Forge kit page describes the proposed sensing and gateway architecture.