Vibration analysis relates measured motion, operating regime, and machine dynamics to support a condition hypothesis. An overall value can flag change. Spectrum, waveform, phase, and context help distinguish possible sources. A maintenance decision becomes defensible when evidence is repeatable, compatible with the mechanism, and connected to asset risk.
High amplitude does not automatically identify misalignment, imbalance, or bearing damage. Point, direction, transducer mounting, bandwidth, resolution, speed, load, and structural transmission affect the signal. A mature route preserves these elements before comparing campaigns.
Audit the complete measurement chain
The signal begins with physical motion and passes through the transmission path, transducer, mounting, conditioning, and processing. Each step can introduce attenuation, saturation, noise, or phase differences. Changing a point or mounting method may break a trend even when the machine condition has not changed.
ISO 20816-1 provides general guidance for measuring and evaluating complete-machine vibration using magnitude and change criteria. Other parts address particular machine classes. Application requires the relevant part, current edition, measurement type, and internally approved limits.
What each representation contributes
- Overall value: summarizes energy in a defined band for screening and trending.
- Spectrum: distributes amplitude by frequency and exposes running orders and component frequencies.
- Waveform: preserves impacts, modulation, beating, and transient events in time.
- Phase: compares motion among points and supports hypotheses involving imbalance, misalignment, and resonance.
- Envelope: emphasizes high-frequency modulation in compatible applications with documented settings.
The method should match the mechanism. Overall velocity can track broad severity while losing detail. Acceleration may suit higher-frequency content. Displacement matters for other machine classes and measurement arrangements. Values converted without bandwidth and detection method are not necessarily comparable.
Minimum route data package
| Dimension | Required record |
|---|---|
| Asset | equipment, component, point, and direction |
| Operation | speed, load, flow, pressure, and process state |
| Acquisition | transducer, mounting, range, sampling, lines, and averaging |
| Processing | unit, band, filter, window, and detector |
| Reference | baseline, previous campaign, and applicable limit |
| Response | repeat, confirm, notify, plan, or escalate |
Safety and access belong in the task. A technician does not move a probe near rotating parts outside an approved point, remove an energized guard, or improvise a collection surface. An unsafe condition creates its own operational response.
Evidence dossier: motor-pump MP-12
All values are instructional and do not represent a customer, benchmark, or PM Run performance. MP-12 transfers process water at 72 m³/h. The train has a 75 kW motor, flexible coupling, and centrifugal pump at 1,780 rpm. Baseline data came after commissioning, accepted alignment, and a stable operating test.
Baseline and internal rule
The plant used two local triggers: an increase greater than 40% over baseline under equivalent operation, and an applicable approved limit for the machine class. The first calls for repetition and analysis. The second follows the response matrix. Neither trigger is a universal limit.
| Point | Baseline velocity RMS | Current campaign |
|---|---|---|
| Motor drive end horizontal | 2.1 mm/s | 5.8 mm/s |
| Motor drive end axial | 1.7 mm/s | 6.4 mm/s |
| Pump drive end horizontal | 2.3 mm/s | 5.5 mm/s |
| Pump drive end axial | 1.9 mm/s | 6.0 mm/s |
Collection occurred at 1,779 rpm, 71.6 m³/h, stable pressure, and bearing temperatures within the plant range. Magnetic mounting, points, and a 10 to 1,000 Hz band were unchanged. A repeat after 15 minutes varied by less than 5%.
Observed signature
The axial spectrum showed dominant 1× running speed and an elevated 2× component. Axial phase between motor and pump was compatible with relative motion across the coupling. The waveform lacked strong repetitive impacts. Envelope data did not show sustained bearing frequencies above baseline.
This combination favored misalignment as a hypothesis without confirming it. Imbalance, distorted base, pipe strain, looseness, and resonance could modify the pattern. History contained a seal replacement order with no final alignment evidence.
Classification and decision
The internal matrix combined trend, magnitude, criticality, redundancy, and available window. A tested standby train existed. Engineering authorized transfer and a planned inspection within 48 hours. The responsible authority and temporary operating condition were recorded.
The notification described values, regime, points, directions, spectra, waveform, and hypothesis. Cause remained open. The linked order covered base condition, soft foot, pipe strain, coupling, alignment, bearings, approved correction, and an operating test.
As-found measurements
The team measured 0.18 mm soft foot at one motor support. After correcting the base, cold alignment showed 0.21 mm offset and 0.35 mm per 100 mm angularity, outside the internal criteria for this train. A controlled pipe-strain check found no material movement.
Specified shims were installed, supports were tightened in the approved sequence, and shafts were aligned to the engineering thermal target. Final readings were 0.04 mm offset and 0.04 mm per 100 mm angularity. The coupling was inspected and retained.
Test and confirmation
At 1,781 rpm and 72.1 m³/h, the highest overall value was 2.3 mm/s. Axial 1× amplitude fell by 67%, and 2× returned to the baseline range. Phase, waveform, temperature, pressure, and seal condition met release criteria.
Verification included a reading after 24 hours, weekly readings for four weeks, and monthly readings for three months. The main outcome was recurrence of the signature. The team also tracked complete operating context, response time, as-found documentation, and final alignment evidence after coupling work.
Correct handoff to SAP PM
The notification preserves the observation, severity, technical object, date, and permitted attachments. The order contains operations, competency, tools, materials, and test criteria. Measurement points can store readings under company architecture. Closeout captures initial and final geometry without replacing evidence with a causal label.
The MP-12 plan gained the revised route and mandatory speed and flow. The coupling task list gained soft-foot inspection and final alignment evidence. History can now show whether the control reduces recurrence.
PM Run supports planning, mobility, execution, and feedback over SAP PM. It does not provide a transducer, collector, automatic diagnosis, prognosis algorithm, or predictive AI. Qualified professionals and plant governance own acquisition, analysis, and technical decisions.
Alarms, trend, and operating changes
An absolute alarm protects against high magnitude under an applicable criterion. A relative alarm detects change from a stable machine. Trend indicates change rate, but extending a straight line to failure ignores acceleration, intervention, and regime variation.
Variable-speed machines may need order tracking, speed maps, synchronous analysis, and stable windows. Comparing 1,200 rpm with 1,800 rpm as equivalent can mistake a normal dynamic response for degradation.
Measurement uncertainty
Repeatability should be measured at the route level. Position error, surface preparation, mounting force, cable movement, sensor range, and analyst settings can all affect results. A route audit repeats selected points with another qualified analyst and compares the deviation.
Calibration status does not remove field uncertainty. The record should identify the instrument and transducer, while the procedure controls mounting and configuration. An unexplained step change after equipment replacement may belong to the measurement system.
Technique boundaries
- The structural path can attenuate or amplify a source.
- Resonance can raise response without a similar rise in excitation force.
- A normal overall value can hide a narrow relevant component.
- Calculated frequencies require correct speed and geometry.
- A compatible pattern supports a hypothesis but not a unique cause.
- Transients and reciprocating equipment can require specialized methods.
- No measurable change does not prove integrity against every failure mode.
Program indicators
Governance follows critical-asset coverage, repeatability, operating-context completeness, hypothesis confirmation, post-maintenance recurrence, and usable P-F time. Alarms without decisions and orders without verification remain visible as separate queues.
The teaching target was 95% of collections with valid speed and load, every high deviation decided within 24 hours, and no axial-pattern recurrence over three months. The target does not turn 2.3 mm/s into a corporate threshold.
Combining the methods
Predictive maintenance integrates techniques into condition decisions. The P-F concept helps define available response time. Shaft alignment in maintenance corrects confirmed geometry. Vibration analysis measures and interprets dynamic response.
Resolution, window, and acquisition time
Frequency resolution follows from span and line count. Two nearby components may merge when resolution is inadequate. More lines extend acquisition time and can capture operating variation. The analyst selects a compromise that matches running speed and the frequencies that must be separated.
Windowing reduces temporal truncation effects for compatible signals while changing peak amplitude and width. The route records the selected window and correction. An unmarked setting change breaks comparison. Averaging can stabilize random noise but may hide a transient event that matters to the mechanism.
Testing the MP-12 hypothesis
The misalignment hypothesis produced three predictions: an axial running-speed component, consistent phase across the coupling, and reduction after geometric correction. Soft-foot inspection explained part of the base distortion. No sustained bearing feature in envelope data lowered, without eliminating, the bearing-damage hypothesis.
Contradictory evidence remained visible. Horizontal response had also increased, which could arise from structural transmission or another source. The order therefore required checks of base, piping, coupling, and bearings before finalizing the scope rather than prescribing alignment alone.
Closeout criterion
Closure required four pieces of evidence: accepted geometry, reduced axial signature, stable process, and no additional damage. If vibration remained high after alignment, the team would perform an impact test and resonance assessment before replacing another component.
Route audit
Every quarter, an analyst repeats 10% of points collected by another qualified person. Differences are stratified by direction, mounting, and band. The purpose is to identify ambiguous instructions, poor surfaces, or configuration drift, not to rank technician speed.
During the pilot, two points varied by more than 12% because the physical mark permitted different positions. Engineering installed permanent identification and revised the task photograph. Variation fell to 4% across three later repeats.
Decision quality review
The program samples notifications and asks whether the recorded evidence could support the assigned priority. A spectrum without speed, a phase value without reference, or a global without bandwidth fails the review. The item returns for correction before being used in strategy analysis.
Baseline acceptance record
A baseline enters service only after the machine passes functional testing, alignment review, and stable operation. MP-12 baseline files include speed, flow, temperature, point photographs, sensor identification, spectra, waveforms, and phase references. A measurement taken during unresolved commissioning work remains a diagnostic record rather than the normal reference.
After a major overhaul, engineering decides whether the previous baseline remains applicable. Changed foundations, rotor, coupling, or process duty can justify a new series while preserving the old one for audit.
Managing an inconclusive reading
A signature remains inconclusive when operating regime, repeatability, or connection with a mechanism is inadequate. It receives neither a normal label nor a preferred cause. The response calls for collection under a specified condition, a complementary technique, or a dynamic test according to risk and available time.
An earlier MP-12 axial record lacked flow information. The team preserved it as an observation but excluded it from trend growth. This prevented a rate calculation across operationally different points.
Competency and technical review
The collector needs command of safety, point location, and settings. The analyst needs machinery dynamics, failure modes, and method limits. Maintenance authority converts a hypothesis into action. For critical machines, an independent review examines evidence and consequence before a shutdown or major teardown is recommended.
Technical references
- ISO 20816-1:2016, official page, accessed August 23, 2026.
- NASA Reliability-Centered Maintenance Guide, vibration section, accessed August 23, 2026.
- SAP Help Portal, Process Maintenance Notification, accessed August 23, 2026.
Frequently asked questions
Does a high overall value identify the fault?
It indicates severity in the selected band. The hypothesis depends on spectrum, waveform, phase, operating state, and machine knowledge.
Does the standard provide one threshold?
The series contains general guidance and machine-specific parts. The company selects the applicable document and approves its criteria.
What returns after repair?
As-found condition, geometry, correction, operating test, and a new signature under comparable operation. This cycle tests effectiveness.
To execute routes and orders connected to the corporate system, review PM Run's operational layer for SAP PM.
