PREDICTION · MEASUREMENT · CORRECTIVE DECISION

Ship Torsional Vibration Correlation and Corrective-Action Record

A practical evidence record for comparing a propulsion shaftline torsional calculation with measured behaviour, testing competing noise and vibration causes, and deciding whether corrective work has been verified under comparable operating conditions.

Technical evidence resourceTorsional measurement + troubleshootingPublished 9 September 2026Vessel Doctor Technical Team
Quick answer

Correlation is not a single percentage. A defensible comparison keeps calculation revision, shaft location, measured quantity, speed and load, excitation order, amplitude convention, processing and uncertainty visible. Agreement at one speed does not validate the complete model, and a peak that falls after a repair does not prove the assumed cause unless the test conditions and competing explanations were controlled.

01 · DECISION QUESTION

State what the comparison must decide before selecting a plot.

A calculation-to-measurement exercise can be required to confirm a predicted resonance, investigate an unexpected shaft response, review a barred-speed range, support commissioning, check a changed engine or propeller configuration, or distinguish torsional behaviour from casing, foundation and hull response. These are different decisions. Record the contractual or technical question, the authorised decision owner and the evidence that would change the next action.

Use the torsional, machinery and structural vibration comparison to select the correct measurement family. A shaft twisting response cannot be established from a bearing-housing accelerometer alone. Likewise, a torsional peak does not automatically explain a noisy cabin or vibrating deck. Multi-source symptoms should remain within a broader noise and vibration troubleshooting scope until the source, path and receiver evidence converges.

Blank decision and acceptance-owner register
Decision IDQuestion to decideRequired evidenceOperating rangeAcceptance ownerExcluded conclusionNext gate
DEC-01__________________
DEC-02__________________
02 · CALCULATION BASELINE

Freeze the exact prediction revision used for comparison.

Identify the shaftline calculation, revision, issue date and author or responsible organisation. Record the represented engine, coupling, gearbox, shaft sections, propeller or driven equipment, inertias, torsional stiffness values, damping assumptions, firing or excitation orders, load cases and speed range. If a maker supplied predicted critical speeds or stress limits, preserve the units, amplitude basis, location and operating restriction exactly as issued.

Model updates made after the test must not silently replace the pre-test baseline. Keep both revisions and list every changed input. A revised damping value may improve numerical agreement without independently proving that the physical system had that damping. The record should distinguish measured evidence, stated design input, assumed property and fitted parameter.

Blank torsional calculation baseline
Model / revisionShaft locationQuantity / unitsSpeed / load caseOrder / modeInput statusLimit / source
CAL-01__________________
CAL-02__________________
CAL-03__________________
03 · MEASURED QUANTITY

Keep angular motion, torque, strain and derived stress separate.

A torsional system may be observed through angular displacement, twist between two stations, instantaneous angular speed fluctuation, torque, strain or a derived stress. The sensor principle and shaft location determine what is actually measured. Record whether the channel comes from an encoder, toothed-wheel pickup, optical system, telemetry strain arrangement or another method, and state the transfer or calculation used to derive the reported quantity.

Do not compare unlike amplitudes without conversion and disclosure. Peak, peak-to-peak, zero-to-peak and RMS values are not interchangeable. Neither are degrees, radians, angular velocity, torque and stress. Document filtering, resampling, order extraction, averaging, windowing, reference pulses and any removed trend. A clean graph without this chain is not a reproducible measurement.

Blank measured-channel and quantity register
Channel / locationSensor principleDirect quantityDerived quantityAmplitude conventionProcessing / bandwidthCalibration / check
TV-01__________________
TV-02__________________
REF-01__________________
04 · CORRELATION ALIGNMENT

Align speed, load, order, location and amplitude before judging agreement.

Build the comparison on common axes and comparable states. Confirm actual shaft speed rather than relying only on bridge or engine-room indications. Where gear ratios or multiple shafts are involved, state which rotational reference defines each order. Match engine load, cylinder condition, propeller state, clutch or gearbox condition and generator or power-take-off status to the calculation case as far as practicable.

Compare predicted and measured resonance speed, response shape across the sweep, dominant order, relative phase or mode evidence where available, and amplitude at the same physical or analytically equivalent location. Use tolerances agreed for the decision rather than declaring a universal correlation threshold. When the calculated quantity must be transformed to match the measurement, retain the formula, constants, units and reviewer.

Blank prediction-to-measurement correlation sheet
Comparison itemCalculated valueMeasured valueCommon basisDifferenceTolerance / rationaleFinding
Critical speed__________________
Dominant order__________________
Amplitude / stress__________________
Phase / mode evidence__________________
05 · DISCREPANCY REVIEW

Treat disagreement as diagnostic evidence, not automatic model failure.

A shifted resonance can follow from inertia, shaft stiffness, coupling condition, fluid or propeller interaction, boundary assumptions, gear representation, temperature or an incorrect speed reference. An amplitude difference may reflect damping, excitation magnitude, load, combustion balance, instrument placement, conversion basis or processing. Missing peaks can also result from an incomplete speed sweep, insufficient hold time or a sensor location with weak sensitivity to the mode.

Rank each discrepancy by its effect on the actual decision. Record the evidence for and against each explanation, the smallest safe discriminating test and who authorises it. Do not tune several model parameters simultaneously and present the resulting match as independent validation.

Blank calculation-measurement discrepancy register
DiscrepancyPossible explanationEvidence forEvidence againstDiscriminating checkSafety / authorityStatus
DSP-01__________________
DSP-02__________________
DSP-03__________________
06 · TROUBLESHOOTING HYPOTHESIS

Use controlled changes to test source-path-receiver explanations.

For a vessel that is noisy or shakes at a particular RPM, list credible torsional, rotating-machinery, structural, hydrodynamic and acoustic hypotheses separately. Predict what each hypothesis should change: shaft order amplitude, casing response, deck or foundation motion, sound spectrum, phase relationship, spatial pattern or response during run-up and run-down. A useful hypothesis risks being disproved.

Change one safe variable at a time where the authorised vessel team permits it. Examples include a controlled speed increment, load redistribution, machinery lineup change or repeat in a documented operating state. Never instruct unsafe propulsion operation, cross a barred-speed restriction without authority, defeat protection or alter machinery merely to obtain a cleaner result.

Blank hypothesis and falsification matrix
HypothesisPredicted signatureChannels / evidenceControlled changeObserved resultSupported / weakenedRemaining alternative
HYP-01__________________
HYP-02__________________
HYP-03__________________
07 · CORRECTIVE-ACTION DECISION

Connect every proposed change to the evidence it is intended to address.

Corrective options may involve operating restrictions, tuning or balancing work, coupling or damper review, alignment investigation, mounting or foundation work, structural modification, isolation, acoustic treatment or further measurement. The responsible designer, maker, owner, yard, class or flag route depends on the vessel and change; a measurement consultant should not imply authority that has not been granted.

Before work begins, record the target signature, unintended risks, drawings or calculations required, approval route, installation evidence and verification plan. A reduction in airborne noise may improve the receiver condition without resolving a torsional concern. Conversely, a shaftline intervention may change stress while leaving a separate structural or acoustic path. Keep the decision boundaries visible.

Blank corrective-action evidence gate
Action IDEvidence addressedResponsible partyReview / approval routeTarget signatureUnintended riskVerification plan
ACT-01__________________
ACT-02__________________
No fabricated authority.

This page contains no completed vessel result, customer identity, maker conclusion, class or flag approval, external review, acceptance statement, backlink or testimonial.

08 · MATCHED VERIFICATION

Repeat the decision-relevant conditions and report what remains untested.

Use the same or technically equivalent shaft location, sensor principle, reference channel, quantity, units, amplitude convention, processing, speed sweep or steady holds, load and machinery configuration. Record changes in draught, trim, propeller condition, sea state, fuel or cylinder state, temperature and other variables that could affect the comparison. If a perfect match is impossible, state which conclusion remains defensible.

Report the before and after traces together with run validity, uncertainty or repeatability relevant to the decision. Separate measured change from causal interpretation. A successful result at one operating point does not verify the full speed and load envelope unless that envelope was retested. Release restrictions only through the authorised route identified before the work.

Blank matched-retest and release record
ItemBaseline conditionRetest conditionMeasured changeComparability limitationDecision retainedRelease owner
RET-01__________________
RET-02__________________
RET-03__________________
09 · CURRENT PRIMARY SOURCES

Confirm the governing edition and contractual acceptance route.

  • ISO 20283-4:2012, with its published 2014 amendment, addresses measurement and evaluation of vibration of ship propulsion machinery; the official record shows the base edition remains published and was confirmed in 2023.
  • ISO 3046-5:2001 addresses torsional vibrations for reciprocating internal-combustion-engine performance and was confirmed in 2025.
  • ISO 13373-1:2002 provides general machinery-vibration measurement and data-collection procedures and was confirmed in 2024.

These records define broad document scope; they do not create a universal shaftline acceptance limit. The contract, calculation, maker requirements, applicable class rules, flag instructions and authorised technical review control the actual decision. IACS does not approve suppliers.

NEXT STEP

Define the prediction, measurement and decision chain before attendance.

Send the shaftline arrangement, calculation revision, operating range, symptoms, previous data, restrictions and required acceptance route so the correlation and troubleshooting scope can be prepared.

Request a propulsion-vibration scopeReview torsional measurementReview troubleshooting