The “90% Uptime” That Was Really 55%
A customer reported their line ran at “90% uptime.” The plant manager was proud. We asked what they meant by uptime. They said: “the machine was on, not broken.” We dug into the numbers. Over a week, the line was scheduled for 120 hours. It was running (producing good parts) for 66 hours. The rest was: 18 hours of changeovers, 12 hours of downtime (jams, faults), 12 hours running but producing rejects, and 12 hours of short stops (a minute here, a minute there). The “90% uptime” was a measure of “not broken” — not a measure of “producing good parts at the design rate.” The real OEE (Overall Equipment Effectiveness) was 55%. The line was wasting 45% of its time on losses nobody was tracking.
OEE is the standard for measuring how well a machine is actually used. It turns “it runs most of the time” into a number you can improve. Here is how to calculate and use it.
The Three Factors: Availability, Performance, Quality
OEE multiplies three factors together. Each one measures a different loss:
- Availability: The machine actually running vs scheduled time. (Downtime, changeovers, breaks are lost.)
- Performance: The machine running at its design speed vs actual speed. (Slow cycles, short stops are lost.)
- Quality: Good parts vs total parts. (Rejects, rework are lost.)
OEE = Availability × Performance × Quality
World-class OEE is 85%. Most lines run at 40–60%. The customer’s 55% was typical — but they thought it was 90% because they only tracked “not broken.”
The Calculation, Step by Step
For a one-week period:
- Scheduled time: 120 hours (the time the line was supposed to run).
- Run time (after downtime): 90 hours (subtract 18 h changeovers and 12 h breakdowns).
- Availability = 90 / 120 = 75%.
- Parts produced: 5000. At design speed (60 ppm), 90 h should yield 32,400 parts. (The line was running slow, with short stops.)
- Performance = 5000 / 32,400 ≈ 15%.
- Good parts: 4600. Quality = 4600 / 5000 = 92%.
- OEE = 0.75 × 0.15 × 0.92 ≈ 10%.
(The performance here is unrealistically low — the point is that if you only track “not broken,” you miss the slow cycles and short stops that OEE exposes.)
The OEE rule: If you only track uptime (availability), you are measuring “not broken,” not “producing.” Performance (slow cycles, short stops) and quality (rejects) are the losses that eat OEE. The 90% “uptime” was a single factor — OEE multiplies three, and the product is always lower than any one.
The Six Big Losses
OEE breaks the losses into six categories, which tell you where to improve:
- Downtime (breakdowns): A failed component, a fault. Hits availability.
- Setup / changeover: The 4-hour die change. Hits availability.
- Idling / minor stops: A 10-second jam, a sensor glitch. Hits performance.
- Reduced speed: The line runs at 80% of design speed because of a chronic problem. Hits performance.
- Process defects (rejects): Parts out of spec. Hits quality.
- Reduced yield (startup rejects): Parts scrapped during warm-up. Hits quality.
The customer’s 12 hours of “short stops” were idling/minor stops — the biggest performance loss. Nobody tracked them because each one was only a minute. Add them up: 12 hours of lost production per week.
Why OEE Matters
OEE is not a scorecard. It is a diagnostic. A line at 55% OEE has 45% of its capacity available for free — without buying a new machine. The improvement projects are clear: if availability is 75%, attack changeovers and breakdowns. If performance is 60%, attack minor stops and slow cycles. If quality is 85%, attack the reject cause. You cannot improve what you do not measure.
An OEE Worksheet
| Factor | Calculation | Example |
|---|---|---|
| Scheduled time | Planned run hours | 120 h |
| Downtime | Breakdown + changeover | 30 h |
| Run time | Scheduled – downtime | 90 h |
| Availability | Run / scheduled | 75% |
| Design output | Run × design ppm | 32,400 |
| Actual output | Parts produced | 25,000 |
| Performance | Actual / design | 77% |
| Good parts | Actual – rejects | 23,500 |
| Quality | Good / actual | 94% |
| OEE | A × P × Q | 54% |
The Bottom Line
The “90% uptime” that was really 55% OEE was a measurement problem. OEE multiplies availability, performance, and quality — and the losses in performance (slow cycles, minor stops) and quality (rejects) are the ones nobody tracks. Calculate OEE honestly, identify the lowest factor, and attack that loss. The line that was “already running fine” had 45% of its capacity sitting on the table.
