7 Common Causes of Unplanned Downtime (and How to Fix Them)

Quick answer: The 7 most common causes of unplanned downtime in manufacturing are mechanical wear, lubrication failure, operator error, changeover overrun, software glitches, spare parts shortages, and sensor failure. In most plants a few of these causes carry most of the lost hours, and your own stop log shows which ones.

Related deep-dives: 6 root causes deep-dive · iceberg cost effect · Pareto analysis · why PM fails (82% rule).

Cause 1: Mechanical wear. Bearings, belts, seals, chains degrading over usage cycles. Predictable from sensor data once you measure it.

• Cost: shows up as repeat breakdowns on the same components
• Fix: condition-based PM triggered by sensor thresholds (vibration, temperature, acoustic), not the calendar. How condition-based PM works.
• Time to result: 3-6 months for MTBF to start rising (the time to result figures on this page are typical estimates, not survey data)

Cause 2: Lubrication failure. Missed re-greasing, contamination, wrong viscosity. Easy to prevent, often overlooked because nobody owns it.

• Cost: hard to see in the downtime log, but disproportionately destructive, because each skipped relubrication eats the safety margin built into the interval and repeated misses end in bearing failure
• Fix: route-based lubrication PM with mobile checklists. Lock the route, name the technician, set each interval from the bearing maker's relubrication calculation (speed, size, load, temperature and contamination), not a blanket calendar.
• Time to result: 4-8 weeks. The fastest-return cause to fix.

See how these causes map to the 6 OEE losses.

Cause 3: Operator error. Wrong setup, missed alarm, parameter typo. Almost never the operator's fault, it is a process gap.

• Cost: often logged as a machine fault, so check how your plant records it
• Fix: standardized setup procedures with visual confirmations + Computer Vision recording each stop on video. Operators do not memorize, they follow.
• Time to result: 8-12 weeks. Heavily depends on operator buy-in.

Cause 4: Changeover overrun. The planned 45-minute changeover took 78 minutes. That extra 33 minutes is unplanned downtime hidden as planned.

• Cost: hidden inside planned stops unless you measure every changeover against its standard
• Fix: SMED methodology (Single-Minute Exchange of Die). Break the changeover into internal (machine stopped) and external (machine running) steps, move as much as possible to external.
• Time to result: First-pass SMED on one line takes 6-10 weeks. Subsequent lines go 3-4x faster.

See the broader 6 root causes framework.

Cause 5: Software glitch. PLC freezes, HMI hangs, scheduling conflict locks the line. Modern plants see this more than they should.

• Cost: usually a small share of unplanned downtime, but trending upward as plants digitize
• Fix: watchdog timers plus scripted recovery for known failure modes, designed so that restarting a PLC or HMI never restarts machine motion: the line comes back in a safe stopped state and an operator starts it deliberately. Manual recovery becomes the exception, not the rule.
• Time to result: 2-4 weeks per failure mode.

How to actually use this list. Do NOT try to fix all 7 at once. The order matters:

1. Run a Pareto analysis on your last 12 weeks of downtime data, classified by cause
2. Pick the top 2 causes for your specific plant
3. Apply the matched fix from this list, one cause at a time
4. Measure for 4 weeks, then move to the next cause

A modern OEE solution with native CMMS records every stop and its duration as it happens, so you walk into the weekly review with the data already collected, not a half-day spreadsheet exercise. That is the difference between Fabrico and a generic "things broke" dashboard.

Cause 6: Spare parts unavailable. The part you need is the one you do not have. Single-day events become multi-day waits.

• Cost: measured in waiting time, often the longest single events
• Fix: failure-mode-driven spare parts policy with min/max levels tied to MTBF data. The 80/20 rule applies: a small share of critical-path SKUs usually covers most failures, so find yours from work order history.
• Time to result: 3-4 weeks to set up min/max + reorder triggers. Effects show within 2-3 months.

Cause 7: Sensor failure. The sensor says the machine is down when it is running, or vice versa. Triggers false alarms, missed real ones.

• Cost: modest as direct downtime, but inflates other cost categories because trust in the data breaks
• Fix: redundant sensors on critical paths + cross-check with Computer Vision as second source. CV catches sensor drift before it becomes a failure.
• Time to result: 4-6 weeks for installation + tuning.

See data collection methods that catch sensor drift.

What counts as unplanned downtime, and how to measure it

Unplanned downtime is any stop during scheduled production time that was not on the plan: breakdowns, jams, waiting for parts or people, and changeovers that overrun their standard. Measure it as unplanned stop minutes divided by planned production minutes. In OEE terms, equipment failures are an Availability loss, while short stops the operator clears in a minute or two count as a Performance loss, so set a threshold, for example two minutes, and require a reason for every stop longer than that.

What unplanned downtime costs

The Siemens True Cost of Downtime 2024 report estimates that unplanned downtime costs the world's 500 largest companies almost $1.4 trillion a year, 11 percent of their revenue. An hour of downtime in a large plant now costs about $36,000 in consumer goods and $2.3 million in automotive. Plants average 25 downtime incidents and 27 lost hours a month, down from 42 incidents and 39 hours in 2019, so the events are rarer, but in most sectors each hour costs more than it did five years earlier.