Reactive versus predictive maintenance – which is the best strategy? Getting this right is key for any site to ensure operations run smoothly. Here we look at both these and examine the circumstances under which they should be used.
Reactive maintenance isn’t automatically wrong
Run-to-failure is a legitimate strategy and many plants probably use it more than they care to admit. It carries no inspection cost, no monitoring cost and no engineer time spent on a machine which still runs. For example, for a £40 proximity sensor with three spares on the shelf, run-to-failure could be the correct commercial call. Replacing it on a schedule would likely be a waste of money.
However, when this logic is applied to other equipment on the asset register, a range of different problems may arise. For example, an ELAU PacDrive M MC-4 drive or a PacDrive 3 LMC controller both belong in a different category. This is because fault diagnosis can take hours whilst sourcing parts even longer. All during this time the line has stopped and downtime costs begin to accrue.
So whilst reactive maintenance may often be considered a risky choice, this may be an incorrect judgement. In certain circumstances it could well be the correct choice, but determining what these are is crucial.
Understand the true cost of being reactive
The repair invoice is often the smallest cost for any incident and we have previously calculated the full cost of unplanned ELAU PacDrive downtime. In short the invoice covers parts and labour whilst the incident itself consumes production value, diagnostic hours, engineer time, emergency sourcing, call-out charges, restart/stabilisation and whatever repeat failures may follow.
Fluke’s 2025 Censuswide study found 68% of UK manufacturers were hit by unplanned downtime last year, with 45% reporting outages of up to 12 hours. These figures show the problem is not uncommon, however, they tell you nothing about your line. The hourly figure needs to be built up from your own production, finance and maintenance data and then applied to specific, named assets.
Reactive, planned and predictive are three different things
BS EN 13306, the European maintenance terminology standard, splits maintenance into “corrective” and “preventive”. It then further divides preventive into predetermined and condition-based.
- Reactive (corrective): action is taken only after failure. This can be the lowest cost to set-up and run, but potentially the most expensive in the end.
- Planned (predetermined): inspections or replacement of parts are scheduled. This leads to a predictable cost, but serviceable parts can potentially be scrapped. Also, faults between scheduled intervals can still occur.
- Predictive (condition-based): the equipment is monitored to assess how it is actually behaving before deciding when to intervene.
None of the above three wins outright. A well-run plant can combine all three grouped by asset criticality. The skill lies in allocating each asset to the correct category and revisiting when and if the asset changes.
Which assets justify predictive attention?
Answering these five questions can quickly identify which would benefit.
- Consequence: if the asset fails mid-shift, what stops and for how long?
- Availability: can you get a replacement this week or is the part obsolete?
- Diagnosis: can your team find root cause in twenty minutes or does it take a day?
- History: has the fault been intermittent and has it reset itself and cleared the alarm before anyone checked?
- Observability: is there any indication of or signs shown before failure?
If a guard switch scores low on all five then leave it as “reactive”. However, a PacDrive M controller on a single-line site which scores high on the first four would justify taking the predictive approach. As an additional note, question 4 really deserves more weight than it usually gets. Intermittent faults can punish reactive teams as the evidence has usually been and gone by the time an engineer arrives.
The tipping point: when predictive starts paying back
It usually pays to treat industry-wide payback figures with caution. Any calculation is local, site-specific can generally be completed very quickly once all the necessary inputs have been compiled.
Consider the asset calculation below (all values are illustrative only).
| What to work out | Illustrative value | Where this figure comes from |
| Production value through the line | £22,000/hour | Production + finance data |
| Incidents on this asset per year | 2 | Maintenance history |
| Average stop, fault to stable output | 6 hours | Maintenance + production history |
| Production value interrupted | £264,000 | 2 × 6 × £22,000 |
| Production recovered later | 55% | Planning / capacity review |
| Contribution margin | 20% | Finance data |
| Contribution at risk | £23,760 | £118,800 unrecovered at 20% |
| Direct incident costs, both events | £9,000 | Support, urgent parts, recovery overtime |
| Annual exposure on this asset | £32,760 |
To help arrive at the most appropriate decision, compare the annual cost of planned checks and condition visibility for this asset against its £32,760 exposure. Monitoring doesn’t have to entirely remove the risk to be worth implementing – in fact there is no complete guarantee it will. However, it will significantly reduce the chance of costly incidents occuring. Plus, if the asset or assets are obsolete such as PacDrive M, then monitoring takes on much greater significance.
Predictive doesn’t mean predicting the failure date
What a predictive strategy provides is earlier evidence. Drive loading creeping up over six weeks, a following error which appears twice a shift but clears itself, or a communication event which repeats before every stop. None of these are forecasts. Instead they are patterns and patterns give engineers something to act on before a hard stop, as well as guidance during diagnosis.
This evidence often goes missing, especially if engineers are unable to take PacDrive diagnostics. Here our Machine Analyser® can help significantly. Functioning as a read-only aid across PacDrive M and PacDrive 3, it retains the fault and event history a controller would otherwise overwrite, preventing data being lost.
Build a mixed maintenance strategy around risk
Following these six steps in order will help construct your plan.
- Classify: list production-critical assets such as ELAU PacDrive M and PacDrive 3 controllers, drives and critical axes first.
- Expose: run the calculations above on each using your own data.
- Decide: determine which assets should follow a reactive strategy and why.
- Schedule: define planned checks where a calendar interval genuinely helps.
- Observe: add condition visibility only where steps two and four justify it.
- Review: revisit at twelve months or sooner should any asset become obsolete.
So classify, allocate and review. The rest is detail.
Conclusion
Reactive maintenance can form part of any maintenance strategy. However, it should not be the default – especially on assets where failure is expensive to diagnose, time consuming to source replacement parts for and disruptive to production. Its inclusion should be based on a decision which has been properly evaluated and, critically, on an asset by asset basis.