How should a Thai factory deploy motor bearing ultrasound condition monitoring? Buying a handheld instrument or installing a wireless sensor is only one decision. The harder work is choosing the bearing, repeating a measurement under comparable conditions, interpreting a change, and assigning an action. This guide walks through device selection, baselines, lubrication decisions, an RFP, and FAT/SAT acceptance.
Ultrasound is one way to observe high-frequency friction and impacts associated with rolling bearings. It can help investigate lubrication changes. A displayed sound level alone cannot identify every cause or predict remaining life. Where temperature and vibration monitoring already exist, define the extra decision ultrasound will support: for example, whether lubrication is needed and whether the response to lubrication was consistent with that diagnosis.
1. Define the maintenance decision before selecting a device
“Abnormal sound” is not a useful procurement requirement. Name the asset, bearing position, possible failure mode, and action. A conveyor motor may need a repeatable check before and after greasing. A variable-speed fan may first need speed and load context because readings from different operating modes are not comparable. A sealed bearing cannot be given a grease instruction. A bearing with a suspected defect may need vibration analysis and an inspection plan rather than more grease.
| Question | Ultrasound evidence | Other evidence | Action |
|---|---|---|---|
| Is lubrication needed? | Change from this bearing’s baseline and sound quality | Last grease date, lubricant, speed and load | Verify; grease under an approved procedure if appropriate |
| When should greasing stop? | Trend during application and any reversal | OEM quantity limit, seal and temperature | Stop under the documented limit and stop rules |
| Is damage suspected? | Persistent impacts or changing sound | Vibration waveform/envelope, speed, inspection | Escalate to diagnosis and planned maintenance |
| Is the reading unreliable? | Position or contact-force sensitivity | Neighbouring machines, instrument setting, missing data | Flag data quality and remeasure |
Portable instruments suit inspection routes, listening, and supervised greasing. Permanent sensors suit assets where access is difficult or changes can occur between rounds. Continuous data still needs a named owner and an action deadline. An alarm is maintenance evidence; it is not a machine protection or emergency-stop function.
2. Give ultrasound, vibration and temperature distinct jobs
Temperature follows heat. Vibration captures movement associated with imbalance, misalignment, looseness and bearing damage. Ultrasound captures high-frequency friction or impact changes and can be particularly useful around lubrication work. An ultrasound anomaly does not automatically mean replacement. Fluke describes early ultrasound indications as a prompt for lubrication or further maintenance planning; SKF stresses the wider diagnostic role of vibration. The methods complement one another.
A bearing can change friction before a temperature alarm occurs, but a hot housing may also reflect load, cooling, ambient conditions or alignment. A raised vibration measure also has several possible causes. The value of adding ultrasound is to document what happened at the same point before and after an action, then read that observation alongside vibration, temperature, speed and maintenance history. Our existing guides to motor bearing temperature monitoring and vibration sensor diagnostics cover those methods in more detail. This article focuses on adding ultrasound to that workflow.
3. Select bearings for a useful pilot
Do not begin with every motor. Rank assets by downtime consequence, availability of a standby machine, spare-parts lead time, access, lubrication history and the ability to confirm an abnormal observation. A modest motor at the only transfer point can matter more than an expensive motor with a standby. Record the drive-end and non-drive-end positions separately; “motor M-12” is not an adequate point description.
Include a comparable healthy machine in the pilot, but do not substitute its reading for the target bearing’s baseline. Housing, mounting, load and transmission paths differ. Peer comparison can highlight a candidate for investigation; trending should centre on the same bearing and the same point. Prefer assets with speed, load, vibration, temperature and maintenance records so the team can explain a change.
For each candidate, photograph or draw the accessible contact point. Record bearing type, lubrication method, normal operating modes and safety conditions. If a probe requires removal of a guard or entry into an unsafe area, redesign the point or assess a permanently mounted sensor. Measurement never takes priority over the plant’s lockout and access rules.
4. Choose a handheld or fixed ultrasound arrangement
For a handheld device, test the contact probe, headphones, saved sound, time waveform or spectral function where needed, repeatable gain/frequency settings, export, glove use, calibration and training. Compare lifecycle costs: accessories, software, probes, calibration and staff time. Ask a second operator to repeat the route from the instructions; a device that only one specialist can use has a deployment risk.
For a fixed system, inspect the mounting surface and stiffness, cable routing, environmental protection, power or battery replacement, communications, timestamping, local data retention and asset tagging. UE Systems lists permanent ultrasonic sensing, trends, alarms and sound samples among available product functions. These are product examples, not a guarantee for another device or installation. Require the bidder to demonstrate the functions on the selected asset.
An airborne microphone beside a machine does not make an equivalent measurement to a contact probe on a bearing housing. Air leaks, air blowers and nearby machinery can dominate the airborne path. Even a contact point can transmit sound from an adjacent mechanism. During evaluation, vary the nearby machinery state and confirm that the signal follows the target bearing rather than the background. Decibel readings from two manufacturers may not be directly comparable; accept repeatability within the chosen configuration and a documented method for equipment replacement.
5. Build repeatable measurement points and routes
Keep the contact location, probe orientation, pressure, surface condition, device settings and operating state consistent. Save a point photograph, instrument and probe identifiers, sound-file format, reading, timestamp, speed, load, product and time since start-up. Mark greasing and repair events. For variable-speed drives, define the operating modes to compare; an unknown mode makes the observation questionable rather than “normal.”
Do not label a single first reading a healthy baseline. Confirm the machine’s condition from maintenance history, vibration, temperature and inspection, then collect repeat observations. A suspect initial bearing would otherwise teach the system that an abnormal state is normal. Reassess continuity after a bearing replacement, sensor move, grease change or significant machine modification. In some cases the new configuration needs a new baseline.
A route sheet should let a different technician reproduce the inspection. It should say where to stand safely, what to touch, what settings to use, which operating mode is valid, what to record and whom to call when a change appears. Keep the sheet in the language used by the team while preserving consistent asset tags across systems.
6. Interpret a trend against its own baseline
A baseline is a comparison point for one bearing under defined conditions; a trend is the history of measurements taken comparably. First ask whether the point, device settings, speed and load match. Then compare level and sound quality. Check whether vibration, temperature and current support the same story and whether a grease or repair event explains the timing.
UE Systems publishes examples of decibel changes relative to a baseline for lubrication and fault screening. These are manufacturer guidance, not universal ISO alarm limits. Its slow-speed guidance also says that very slow bearings may produce little ultrasonic energy and may require attention to recorded sound and time waveform rather than level alone. A pilot should therefore set asset-specific review levels and document exceptions. Do not insert a vendor example into every motor’s RFP as a guaranteed fault threshold.
Use separate statuses for invalid data, “remeasure,” “inspect lubrication or vibration,” and “escalate for a maintenance decision.” An alert volume that the team cannot process is not a useful design. Keep the reason for each setting and its change history. Link the plot to grease brand and amount, technician, bearing replacement and operating mode, so the maintenance team can see cause and sequence.
7. Separate lubrication response from damage suspicion
Do not automatically grease whenever an ultrasound level rises. Remeasure first. Check sound quality, vibration, temperature, speed, load and recent work. If the bearing can be greased and the evidence supports insufficient lubrication, use the approved grease, quantity limit and safe procedure. Apply in small increments while measuring the same point. A fall and stabilisation can support the lubrication hypothesis. No improvement, persistent impacts or a new rise calls for stopping and further assessment, not unlimited additional grease.
UE Systems describes monitoring during greasing to help avoid both insufficient and excessive lubrication. This does not override the motor or bearing manufacturer’s instructions. SKF’s documentation for one dedicated lubrication checker explicitly limits what that particular tool can diagnose: it does not establish overall bearing condition or prove over-greasing; unusual acoustic patterns call for vibration or lubricant analysis. Ask what the proposed instrument can actually capture and validate that scope during the pilot.
Improvement after grease is not proof of an undamaged bearing. If vibration and temperature remain unusual, investigate separately. Conversely, a high ultrasound reading under a different load may warrant a comparable repeat before intervention. A useful workflow asks whether there is enough evidence for a specific action, not whether one number is red.
8. Plan for slow speed, variable speed and harsh conditions
At very low speed, events are sparse and a short sample can miss an impact. Require recording duration, saved sound or waveform and a speed tag appropriate to the asset. Under variable speed or changing process load, build mode-specific comparisons. Ask whether adjacent air blowers, leaks or machinery change the reading. Test that influence in the actual plant rather than assuming that frequency filtering removes it.
Humidity, dust, heat, cleaning chemicals, damaged cables and poor wireless reception affect installation and ownership cost. Treat missing measurements as missing data, never as zero or a healthy reading. If data arrive late after reconnection, preserve original timestamps. Record sensor changes, calibration and firmware updates as quality events that may break trend continuity.
9. Connect observations to maintenance work
Use a minimum event record: asset ID, bearing position, point, sensor ID, time, operating mode, value, saved sound, quality flag and action. For a permanent system, verify the availability of raw sound or waveform when needed, export through CSV or API, retention and access rights. UE Systems lists routes, graphs, alerts and exports in its product software, but the scope depends on the selected system.
Assign a first reviewer, a lubrication technician, a machine owner and a person authorised to decide on a stop. Notifications should include point, change, comparison conditions and a direct link to the evidence. When a CMMS work order is created, provide a way to close a false alert and enter the amount of grease and the post-work result. A one-way alarm feed cannot teach the organisation whether the action helped.
The data integration questions overlap with our wireless vibration retrofit guide. For ultrasound, explicitly pair the before and after lubrication observations as one maintenance event. Decide retention, export format and data ownership before a supplier change makes them urgent.
10. Put testable requirements in the RFP
Ask bidders for a point-design proposal, data flow, sample records, training, maintenance terms and FAT/SAT plan, not only a product catalogue. If a vendor uses a proprietary health score, ask which inputs it needs, how missing data are shown, what evidence the technician can inspect and which failure modes are outside its scope. “AI predicts bearing failure” is not an acceptance criterion.
| RFP area | Required answer |
|---|---|
| Assets and actions | IDs, bearing positions, lubrication capability, target failure modes and response |
| Measurement | Contact or fixed mounting, settings, sound/waveform storage and repeat method |
| Context | Speed, load, product, warm-up and post-grease state |
| Data | Timestamp, tags, quality flag, retention, CSV/API, ownership and exit |
| Alert | Review levels, delay, acknowledgement, closure and audit trail |
| Security and service | Network, credentials, updates, calibration, replacements and training |
| Acceptance | FAT/SAT steps, pass criteria, evidence and open-issue handling |
Compare bids on the same cost perimeter: sensors, mounting, network, subscriptions, calibration, replacement, training, support and time required to review alerts. A capability claimed in a brochure should have an agreed site test. A system that generates more alerts than the team can evaluate may cost more than its hardware price suggests.
11. Run a pilot, FAT and SAT for different purposes
The pilot asks whether ultrasound adds a useful decision on the selected bearings. Repeat measurements on a healthy bearing, assess a suspect bearing where available, change safe operating conditions, and observe before/after greasing under an approved job. If no fault occurs, evaluate repeatability, missing data, route effort and linkage to maintenance history. Do not create a hazardous fault merely to demonstrate a detector.
FAT verifies configured tags, screens, file storage, exports, access rights, simulated notifications and behaviour during communications loss before delivery. A simulated alert proves the data path, not actual bearing diagnostic performance. Match instrument IDs, calibration documents, software versions, spares and manuals.
SAT verifies the selected site points: safe mounting, repeatable reading, correct speed/load tags, accessible sound and history, protection of wiring or wireless links, power and operator workflow. A period with no real failure cannot prove detection of every future defect. Agree on a follow-up review period and how the baseline or alert settings can change. Numerical criteria should specify the method, denominator and handling of missing samples before a target number is written into a contract.
12. Give the programme an owner after installation
Training must be demonstrated, not only attended. Ask another technician to follow the route sheet and explain the same result. ISO 18436-8:2013 covers qualification and competence assessment for personnel performing ultrasound condition monitoring; it does not supply an asset-independent dB threshold. Use it to inform training requirements, while verifying the plant’s own procedures.
Make a one-page abnormal-event flow. Verify the asset and point, check settings and data quality, remeasure under comparable conditions, inspect sound and other signals, then choose lubrication, detailed diagnosis, planned maintenance or monitoring. Record a “no action” reason separately from a normal condition. A load mismatch, sensor fault and existing replacement plan should not all become a closed alert with no explanation.
At monthly review, examine alerts reversed by remeasurement, improvements after lubrication, confirmations by vibration or physical inspection, missing-data time and open cases. Retain unsuccessful or ambiguous cases: they are needed to tune the method. Before changing a threshold, find out whether the cause is an operating-mode mismatch, loose mounting or a temporary post-grease state. Record the reason and version for each change.
Receive photographs of points, the tag dictionary, baseline conditions, alarm rationale, calibration records, network map, permissions, export instructions, escalation flow and maintenance contacts as one handover set. Test export and playback yourself. After a bearing, sensor, grease or firmware change, decide whether the old trend is comparable and when to establish a new baseline. Budget the continuing work, not only additional sensors.
13. Make procurement and deployment workable in a Thai plant
Production, maintenance, IT, purchasing and an external service partner may use different languages and priorities. Translate route sheets and stop/grease rules into the language technicians actually use, while keeping equipment tags identical to the English design documents. Name the person who authorises a shutdown or additional diagnosis. A translation of the display alone is insufficient if the handover of responsibility is unclear.
For a cloud service, check the plant network route, data buffering on disconnection, storage location, cyber review and the scope of remote vendor changes. An on-premise design still needs backup, updates and portable-device controls. Test connectivity in the actual equipment area at SAT. Include point design, mounting, safety provisions, support hours, multilingual training, calibration and consumables in the cost comparison. Expand from a small route after the team has documented decisions made with the observations.
14. FAQ
Can ultrasound replace a motor bearing temperature sensor?
No. Temperature shows heating and cooling conditions; ultrasound observes high-frequency friction or impact change. Ultrasound may help investigate a change before a temperature alarm, but it cannot guarantee early detection of every fault. Keep the temperature record and define the additional decision ultrasound will enable.
Is ultrasound needed when vibration monitoring already exists?
It depends on whether lubrication decisions remain uncertain. Vibration is important for mechanical diagnosis; ultrasound can add evidence before and after greasing. If the current process already makes reliable decisions, test whether the new method changes an action before expanding it.
How many dB above baseline means grease is required?
Vendor examples exist, but no universal value applies to every bearing, speed and instrument. Establish a baseline on the target asset. Review comparable conditions, sound quality, other signals and maintenance history. At very low speed, recorded sound and waveform may matter more than a single level.
Should a technician keep adding grease if the reading does not fall?
No. Stop at the approved limit and investigate the measurement, grease path, lubricant compatibility, bearing condition and other friction sources. Use vibration or physical inspection where needed. Unlimited greasing can itself be harmful.
What is the difference between pilot and SAT acceptance?
A pilot investigates usefulness for the selected assets. SAT checks that the contracted points, safety, tags, data path, alarms and work process function at the site. A fault-free trial period does not prove detection of every possible fault; agree on continued review separately.
Conclusion
Motor bearing ultrasound condition monitoring becomes useful when the team can compare the same point under known conditions and connect a change to a defined maintenance action. Record the response before and after lubrication, set a stopping rule, and use vibration, temperature and inspection to evaluate possible damage. Put repeatability, data access, owner and acceptance tests in the procurement package before committing to a plant-wide rollout.
If you are deciding which motors to pilot, whether handheld or fixed sensing fits, or how to connect ultrasound with existing vibration and temperature records, contact TOMAS TECH. We can help define measurement points and acceptance criteria from your asset register and inspection history.
Sources
- ISO 17359:2018, general condition monitoring guidelines.
- ISO 18436-8:2013, ultrasound personnel qualification.
- UE Systems, slow-speed bearing monitoring.
- UE Systems, condition-based lubrication.
- UE Systems, remote monitoring products.
- Fluke, vibration monitoring setup and complementary methods.
- SKF, inspection and monitoring.
- SKF, TLGU 10 lubrication checker manual and scope limitations.
