Predictive methods go further by estimating deterioration or failure from models and trends. A filter-change threshold based on condition needs no prediction algorithm; define the decision before selecting the technique.
Combine methods where appropriate. Scheduled safety activities can remain in place while condition findings determine extra cleaning, adjustments or component changes.
Choose an asset and a useful signal
Identify the failure mechanism and whether it gives useful warning. Vibration, filter pressure drop, unusual temperature or visible wear may be relevant, but the measurement must match the asset and deterioration being assessed.
Interpret readings alongside operating conditions. Load changes can affect temperature or vibration without establishing a defect. Record context and compare equivalent states before treating an unexplained value as grounds for repair.
Monitoring needs sufficient warning for investigation and action. Failures without a detectable lead-up require other measures, such as appropriate planned tasks, standby provision or a replacement decision.
Build a complete response process
- Set a baseline using asset identity, operating conditions and available history.
- Specify the inspection or measurement method and competence needed.
- Define attention criteria from manufacturer information and suitable assessment.
- Allocate review, escalation and action for each significant finding.
- Record work completed and whether an acceptable condition was restored.
Assign responsibility for warnings and exceptions. Missing measurements, failed sensors and inaccessible inspections need defined responses; none provides evidence that equipment is healthy.
Retain the tasks that still apply
Do not cancel statutory inspections, manufacturer tasks or established duties simply because readings look acceptable. Assess any programme change against the requirements and risks of the particular asset.
HSE's maintenance guidance covers competent, safely planned work. Its maintenance procedures guidance concerns major-hazard plant and illustrates choosing an appropriate regime per item. Apply that principle in the actual equipment context.
What makes the approach worthwhile?
Compare monitoring effort with failure consequences and usable warning time. Periodic inspection may be sufficient without costly continuous equipment; conversely, occasional observations may miss deterioration progressing between checks.
At industrial premises, interpret findings against production and load. In offices, apply condition evidence to plant decisions without assuming every asset is operationally critical.
Common questions
Does condition-based maintenance require sensors?
No. Planned visual inspections and portable measurements may be adequate. Match the method to the asset, deterioration pattern and warning period needed for action.
Can it replace all planned maintenance?
No. Applicable scheduled tasks and inspections remain. Use condition findings to adjust only those programme decisions for which the evidence provides a sound basis.
What should happen when a threshold is exceeded?
Apply the agreed escalation and technical response: verify the reading, investigate or arrange work as appropriate. An unexplained reset or closure does not resolve the finding.
How do we know it is helping?
Review defects identified, warning time gained, completed actions and repeat failures. Quantity of data is not a measure of the quality of maintenance decisions.
Related services and next steps
Consider Condition-Based Maintenance, Asset Maintenance, Mechanical and Planned Replacement Programmes when defining the response to findings.
Review services, maintenance, projects, sectors and locations. Discuss condition-based maintenance with the assets, failure history and decisions the monitoring should support.
