Condition-Monitoring Pilots: From Alert to Verified Work Order

Download the alert-to-work-order evidence sheet

Download the condition-monitoring pilot evidence sheet. Use one record for each reviewed event, with links to the original measurement, operating context, specialist finding, maintenance action and closure evidence. Blank fields and an unreviewed status identify work still required; they do not establish that the machine is healthy.

The asset monitoring service defines the project scope. The predictive versus preventive maintenance guide compares maintenance strategies, while this worksheet helps a manufacturing team decide whether a proposed monitoring workflow works on its actual assets.

Choose a decision that the pilot can exercise

Select assets using criticality, known issues, available access and the likelihood that the required operating conditions can be observed. The most expensive machine is not automatically the best first trial if no suitable reference or safe inspection opportunity exists during the period.

Write the intended decision in operational terms. The team may need to schedule a reference measurement, inspect a bearing during a planned stop or confirm whether a maintenance intervention changed a trend. Identify who has authority to interpret the evidence and who can arrange the action.

Agree the trial boundaries and success criteria before collecting data. Include installation quality, measurement availability, review timeliness and the ability to close an event with evidence. These are useful acceptance outcomes even when the machine does not develop a fault during the trial.

Establish the starting condition and baseline

Record the asset configuration, operating modes, recent work and known concerns. Preserve the actual measurement locations, settings and identity mapping. The baseline describes the observed starting point and should not be treated as an independent certificate of mechanical health.

Collect representative conditions and ask the responsible specialist to select valid comparison periods. A variable-speed asset or intermittent process may require several reference states. Keep startup, stable duty and unusual operation distinguishable instead of averaging them into one unexplained reference.

Document data limitations early. Missing speed context, an unstable mounting or an unsuitable acquisition interval can affect interpretation. Resolving those issues is part of the pilot's engineering work, and the record should show how the accepted measurement arrangement was established.

Separate detection from interpretation

An alert is a rule applied to a measurement or event. Record the rule, its source, effective version and the operating conditions in which it applies. The first review should check identity, data quality and duty before assigning a mechanical conclusion.

Keep the original alert and the reviewer finding as separate records. The reviewer may confirm a concerning change, request another measurement, explain the event through operating conditions or leave the cause unresolved. Each outcome is useful when its evidence and next action are retained.

Use precise labels for false or unhelpful alerts. A sensor fault, an incorrectly applied threshold and a valid process change are different findings. Lumping them together as false positives can hide whether the next improvement belongs in installation, configuration or operating-state interpretation.

Record the times that matter

Distinguish acquisition time, platform arrival, notification, acknowledgement, review and action. These timestamps describe different parts of the process. A promptly delivered alert may still wait for specialist interpretation or the next safe maintenance window.

Agree which intervals the team wants to assess and why. Review delay can reveal a missing responsibility, while the time to physical action may be governed by production access or parts availability. Keep those dependencies alongside the duration so the pilot does not judge the monitoring system for a constraint it never claimed to remove.

If a record arrives after a communication interruption, preserve its original acquisition time. The arrival of historical evidence should not be reported as an immediate live warning. The operating team needs to know when the condition was measured and when it became available for action.

Connect the event to an actionable work request

A useful work request identifies the asset, observed condition, relevant period and the check being requested. Attach the specialist's finding and any safe-access requirement. This gives the technician a focused task rather than a dashboard screenshot with no clear question.

Record the actual work-order identifier and responsible planner. If the request is deferred, retain the reason, authorised decision and next review point. A deferred action remains part of the evidence chain and should not disappear merely because the alert was acknowledged.

The technician's finding should describe what was inspected and observed. Keep photographs, reference measurements or part findings where the customer permits them. The link between an alert and a work order is useful, but the inspection evidence determines what the team can conclude about the condition.

Close the record with a comparable check

Compare post-work readings under suitable conditions and retain changes in configuration or duty. If several interventions occurred together, document them all. A changed trend can support a maintenance finding without proving how much each individual action contributed.

A closure can also state that the issue was not confirmed or that more evidence is required. The record should explain the basis for that outcome and any continuing observation. Completing a form is not a reason to force a successful repair narrative onto an inconclusive investigation.

Evaluate value without inventing avoided downtime

Measure the delivered workflow first: usable data, successful reference comparisons, completed reviews and traceable actions. Record the cost of devices, installation, communication, specialist interpretation and local maintenance effort. These quantities help the customer understand the operating commitment of an expanded system.

If financial value is claimed, state the attribution method and supporting production evidence. A repair performed after an alert does not by itself establish when a failure would otherwise have occurred or how much downtime it would have caused. Keep estimates, assumptions and measured outcomes clearly identified.

When no significant event occurs, report that fact and assess the agreed readiness criteria. A quiet trial may demonstrate stable measurement and a successful response exercise while leaving failure-detection performance unproven. The next-stage decision should reflect what the trial actually observed.

Decide what to expand

Review the complete record with maintenance, operations and the specialist. Identify whether the next improvement is another asset, better operating context, a clearer response role or a change to the measurement method. Expansion should follow the decision the evidence supports.

The maintenance framework in PNNL's Operations and Maintenance Best Practices guide provides background for combining measurement, inspection and maintenance. Apply the worksheet to the plant's own equipment and work process. Discuss a condition-monitoring pilot with the asset list, intended decision and people who will witness the evidence chain.

Frequently asked questions