Predictive maintenance: check the warning and plan the response

Separate predictive, preventive, and condition-based maintenance

Reactive maintenance follows failure. Preventive maintenance acts on a calendar or usage interval. Condition-based maintenance acts when a measured indicator crosses a rule. Predictive maintenance estimates the future development of the condition or the remaining window for action.

Use the least complex method that improves the decision. A stable pressure limit may outperform a machine-learning model. Predictive methods earn their place when several signals interact, degradation depends on load or product, or the team needs earlier notice than a simple alarm provides.

• Failure mode and consequence
• Detectable degradation signature
• Useful warning horizon
• Maintenance action available within that horizon

Build a decision-ready data pipeline

Start from the failure mode, then select signals. Vibration, temperature, current, pressure, acoustics, lubrication, controller alarms, cycle time, and process quality can all matter. Add operating context such as speed, load, recipe, product, ambient condition, shift, and recent maintenance so normal changes are not mistaken for faults.

Synchronize clocks and asset identifiers across PLC, sensor, historian, CMMS, MES, and manual records. Preserve raw readings, transformations, missing-data flags, maintenance timestamps, replaced components, and confirmed findings. A model trained on vague work-order text will reproduce vague labels.

• Sensor values and sampling context
• Machine state, load, recipe, and product
• Alarm, stop, repair, inspection, and replacement history
• Failure confirmation and “healthy after intervention” evidence

Evaluate the model as an operating policy

Accuracy alone is not enough. False negatives can allow a costly failure; false positives can waste parts and planned downtime. Measure precision, recall, warning time, calibration, drift, and performance by machine and operating regime. Then evaluate the maintenance policy created by the threshold.

NIST research highlights comparing condition-monitoring algorithms not only by algorithm metrics but by manufacturing outcomes such as production quantity or quality. A slightly less accurate model may be better if its alerts arrive within a usable planning window.

Follow an alert through the actual maintenance decision

Illustrative case: a bearing monitor issues eight alerts over a defined month. Competent inspections find actionable deterioration in five and no actionable finding in three. Record the three as non-actionable alerts under this inspection rule, not proof that the machine was healthy forever. Eight alerts alone cannot measure missed faults or establish model accuracy.

For one alert, the estimated warning window is 36 hours. Planning needs six hours to obtain the approved part and four hours for the approved intervention, leaving a nominal 26-hour margin. That subtraction does not establish a safe remaining life: uncertainty, inspection findings and site procedures govern the decision.

Assign inspection and intervention capacity before raising more alerts. Record when the signal arrived, who reviewed it, the inspection, the approved action and the condition found. Afterwards compare recurrence, quality and operating exposure with a suitable baseline. An avoided-failure saving needs a defensible counterfactual; closing an alert is not enough.

Test maintenance timing in system context

An equipment-health prediction does not automatically identify the best intervention time. Consider spare parts, labor, access, safety, changeovers, buffers, downstream demand, and whether the asset is the current system constraint. Use simulation to compare “run,” “inspect,” “derate,” and “repair” policies under uncertainty.

Set clear escalation rules and let people review the evidence. High-risk decisions may require a physical inspection or redundant signal before shutdown. Record why an alert was accepted, deferred, or rejected so the policy can improve.

1. Detect

   Identify a trustworthy deviation from the expected condition.

2. Diagnose

   Link the signal to plausible failure modes and evidence.

3. Forecast

   Estimate the warning horizon with uncertainty.

4. Plan

   Compare intervention options against production and risk.

5. Verify

   Inspect the asset and measure whether the action prevented recurrence.

Pilot narrowly, then scale by repeatable value

Choose one consequential, detectable failure mode on a well-understood asset. Establish baseline downtime, maintenance cost, false alarms, quality loss, and production impact before the pilot. Run alerts in shadow mode first so the team can evaluate them without changing operations.

Scale only after the data pipeline, alert ownership, response workflow, cybersecurity controls, and value calculation are repeatable. Monitor model and sensor drift. A predictive-maintenance system that produces unowned alerts is only another dashboard.

Maintenance strategies and monitoring techniques are different lists

Reactive, interval-based preventive, condition-based and predictive maintenance are four useful decision categories, not an exhaustive partition. Predictive methods often sit within a wider condition-based program. Other lists add reliability-centered maintenance, risk-based prioritization or proactive defect removal, which describe different levels of policy and analysis. There is no single seven-type list that every site must use.

Vibration, infrared thermography and oil analysis are three common monitoring techniques, not the only three types of predictive maintenance. The U.S. Department of Energy also describes ultrasound and motor analysis. NIST discusses model-based, data-driven and hybrid approaches to prognosis; those classify the prediction method rather than the sensor.

Start with the failure mode and a feasible approved response. A temperature image can reveal a hot connection, but temperature alone may not predict its remaining life. Keep detection, diagnosis, prediction and the decision to intervene separate.

The maintenance 10% rule is not permission to delay a task

One use of “10% rule” concerns the schedule-compliance window around a maintenance due date. Definitions differ: Allied Reliability describes a total window of 10% of the interval, or 1.5 days either side for a 30-day task. A policy allowing plus or minus 10% would instead give three days each side. Write down which convention your report uses. Neither is a universal safe operating limit.

Other writers use similar wording for repair-versus-replacement thresholds or a claimed prevention-to-failure cost ratio. These are different claims. Do not apply any of them without a stated basis. Approved site procedures, manufacturer requirements and any legal or inspection limits control the allowable timing. Evaluate cost from the asset’s actual risk, expected life and downtime consequences rather than assuming every repair has a tenfold penalty.