Maintenance Strategies: Reactive vs. Preventive vs. Predictive

What is maintenance in manufacturing?

Maintenance is the combination of all technical, administrative and managerial actions intended to retain a machine or production asset in — or restore it to — a state in which it can perform its required function. In the DIN 31051 framework (the German standard that defines maintenance terminology), this covers four activities: servicing (Wartung), inspection (Inspektion), repair (Instandsetzung) and improvement (Verbesserung). Every unplanned minute of downtime feeds directly into the availability factor of OEE. In a typical mid-market manufacturing plant, maintenance-related losses account for 15–25 % of total downtime — and the majority of it is preventable.

What are the 4 maintenance activities per DIN 31051?

DIN 31051 is the reference standard used across DACH and most of continental Europe. It splits maintenance (Instandhaltung) into four distinct activities — each with a different purpose and trigger:

Most plants invest 80 % of their maintenance effort in servicing and repair — the routine and the reactive. Inspection and improvement are underfunded. That is exactly where the biggest OEE gains hide: catching a degrading bearing at inspection before it fails catastrophically during a Friday night shift, when the spare part is not in stock and the repair takes 6 hours instead of 1.

What are the 4 maintenance strategies and when should you use each?

The choice of strategy determines whether maintenance is a cost centre or a competitive advantage. Here is the honest comparison:

The pragmatic approach for most mid-market manufacturers: preventive for 80 % of assets, condition-based or predictive for the 20 % that are bottlenecks. Reactive maintenance on critical machines is not a strategy — it is a lack of one.

How does maintenance impact OEE availability?

OEE availability is calculated as: Operating Time / Planned Production Time. Every minute of unplanned downtime caused by a maintenance failure reduces availability directly. But the impact goes further than the OEE formula shows:

• Unplanned stops destroy the production schedule. A 2-hour press breakdown does not cost 2 hours — it costs the cascading delays on every downstream operation. At Schmiedetechnik Plettenberg, the integration of MES real-time data with finite capacity scheduling made exactly this visible: a maintenance event on one forging press shifted 14 downstream orders.
• Planned maintenance is not an OEE loss if scheduled correctly. DIN 31051 servicing during planned non-production time (weekends, shift breaks) does not reduce OEE availability. The same servicing during production time does. The difference is not maintenance strategy — it is scheduling discipline.
• Short, frequent micro-stops are often maintenance-related but never classified as such. A sensor that drifts and causes a 30-second alarm 20 times per shift is a maintenance problem that shows up in the performance factor, not the availability factor. Without automatic alarm capture, it is invisible. At Neoperl, SPS-based alarm capture correlated alarm patterns with downtime events and identified that 4 alarm codes — all maintenance-related — accounted for 80 % of total losses. That insight reduced downtime by 10 % and scrap by 15 %.

How does an MES transform maintenance from reactive to data-driven?

Without an MES, the maintenance department operates on work orders, calendar schedules and the shift leader's phone call when something breaks. With an MES, maintenance gets three things it never had:

• Automatic downtime logging with reason codes: Every stop is captured with timestamp, duration, machine, order and — if connected to the PLC — the alarm code that triggered it. No manual paper log. No end-of-shift guessing. The SYMESTIC alarms module captures PLC alarms in real time and correlates them with downtime events automatically.
• Failure pattern recognition: The MES shows which machines fail most, which alarm codes precede failures, which shifts have more breakdowns and whether failure frequency increases after specific material lots or process parameter changes. This is the data foundation for condition-based and predictive maintenance. At Neoperl, the correlation of SPS alarms with quality defects made it possible to identify degrading components before they caused scrap — not after.
• Maintenance KPIs in real time: MTBF (Mean Time Between Failures), MTTR (Mean Time To Repair), maintenance-related OEE availability loss — calculated automatically per machine, per line, per plant. When MTBF on press 7 drops from 48 hours to 12 hours over 3 weeks, the trend is visible on the dashboard before the catastrophic failure happens.
• Transition from calendar-based to usage-based intervals: Instead of "replace belt every 3 months," the MES tracks actual cycle counts and running hours. The belt gets replaced after 5,000 hours of actual operation, not after 3 calendar months (which may include 4 weeks of plant shutdown where zero wear occurred).

At Meleghy Automotive — 6 plants, bidirectional SAP integration — the SYMESTIC MES feeds maintenance-relevant data (alarm patterns, downtime trends, availability KPIs) back into the ERP. The maintenance planner in SAP sees real machine data, not estimated intervals. Result: 10 % reduction in downtime, 5 % improvement in availability.

What maintenance KPIs should every plant track?

The most important insight: MTBF and MTTR calculated from manual logs are unreliable. Operators round times, forget to log short stops and classify reasons inconsistently. Automatic MES capture eliminates this — the timestamps come from the machine, not from memory.

FAQ

What is the difference between maintenance and TPM?
Maintenance (DIN 31051) defines the activities: servicing, inspection, repair, improvement. TPM (Total Productive Maintenance) is the management philosophy that involves operators in routine maintenance tasks (autonomous maintenance) and aims for zero breakdowns, zero defects, zero accidents. Maintenance is what you do. TPM is how you organise who does it. An MES supports both: it captures the data that maintenance needs for planning and the data that TPM needs for operator-level visibility.

Should maintenance be planned during production time or outside it?
Outside production time whenever possible — this preserves OEE availability. But the real answer depends on the asset. A bottleneck machine that runs 24/7 has no "outside production time" — planned maintenance must be scheduled as a deliberate availability loss. The MES quantifies the trade-off: is it cheaper to stop press 7 for 2 hours on Saturday (planned, controlled, spare parts ready) or risk a 6-hour unplanned breakdown on Tuesday (uncontrolled, cascade delays, overtime for repair crew)?

How much should a plant spend on maintenance?
Industry benchmarks: 2–5 % of asset replacement value per year for discrete manufacturing. But the absolute spend matters less than the ratio of planned to unplanned. A plant spending 3 % with 90 % planned ratio will outperform a plant spending 5 % with 50 % planned ratio every time. The MES provides the data to shift that ratio — because you cannot plan what you cannot see.

What is the connection between maintenance and predictive maintenance?
Predictive maintenance is an advanced maintenance strategy that uses data (vibration, temperature, alarm trends, process parameters) to predict when a failure will occur and schedule repair before it happens. It sits at the top of the maintenance maturity ladder: reactive → preventive → condition-based → predictive. Each step requires more data. An MES provides the data foundation: alarm history, downtime patterns, cycle time trends and process parameter logs are the inputs that predictive algorithms need.

Related: TPM · Predictive Maintenance · OEE Explained · SYMESTIC Alarms Module · MES: Definition & Functions · MTBF · MTTR

About the author

Martin Brandel

MES Consultant at SYMESTIC. 30+ years in industrial automation — from Simatic S5 to Cloud MES. Dipl.-Ing. Communications Engineering. Specialises in brownfield machine connectivity and maintenance data integration. · LinkedIn