Tutorial — Industrial performance
OEE (Overall Equipment Effectiveness)
Measure the real performance of your equipment: OEE isolates the 3 fundamental industrial losses — Availability, Performance, Quality. One indicator, three actionable levers. Python calculation with Pyodide in your browser: no data leaves your machine.
1. The 6 major time categories — measurement framework
Before calculating OEE, the day is split into nested time levels. Each level removes one category of loss from the previous level.
Tip: memorize the cascade TO → TR → TF → TN → TU. Each step isolates one
family of stoppages. OEE rests on this cascade — without it, you end up adding apples and oranges.
2. The 3 components of OEE
OEE = Availability × Performance × Quality
= (TF / TR) × (TN / TF) × (TU / TN) = TU / TR
| Component | Formula | Losses included |
|---|---|---|
| Availability | TF / TR | Breakdowns, changeovers, adjustments |
| Performance | TN / TF = (N × tc) / TF | Minor stops (< 5 min), slowdowns vs theoretical cycle |
| Quality | good / produced | Startup scrap, steady-state scrap, rework |
Rule of thumb: Availability answers “is the machine running?”, Performance answers “how fast?”, Quality answers “right the first time?”.
3. The 6 big losses (AFNOR NF E60-182 / Nakajima)
Nakajima (1988) formalized the classification of industrial losses into 6 categories, adopted by the AFNOR NF E60-182 standard. Each loss maps to one component of OEE.
| # | Loss | Component affected |
|---|---|---|
| 1 | Breakdowns | Availability |
| 2 | Changeovers / adjustments | Availability |
| 3 | Minor stops (< 5 min) | Performance |
| 4 | Slowdowns (reduced speed) | Performance |
| 5 | Startup / ramp-up scrap | Quality |
| 6 | Steady-state scrap / rework | Quality |
The < 5 minutes rule for classifying a stop as a minor stop comes from practice: it is the duration below which there is usually neither a work order nor a dedicated stop code. These short stops are therefore invisible in standard reports — and yet they bleed OEE.
4. OEE vs TRG vs TRE — choosing the right baseline
| Indicator | Formula | When to use it |
|---|---|---|
| OEE (standard) | TU / TR | Running production — the operational reference |
| TRG (overall yield rate) | TU / TO | Management view — includes planned stops (CAPEX / installation ROI) |
| TRE (economic yield rate) | TU / (TR − changeovers) | Isolating machine performance from the product mix (small batches) |
Always state the baseline in a report. Comparing an OEE (86%) with a TRG (72%) gives the impression of a decline when it is just a change of denominator.
5. Performance benchmarks
| OEE | Class | Interpretation |
|---|---|---|
| > 85% | World class | Reference performance (VDMA, JIPM World Class OEE). Diminishing returns. |
| 70 – 85% | High performing | Good, but ~15 points remain to be gained. Realistic target: 12-24 months. |
| 55 – 70% | Average | Median of French industry. Systematic Avail/Perf/Qual breakdown required. |
| < 55% | Needs improvement | Huge potential. Top priority: break it down before drawing conclusions. |
These thresholds apply to OEE (not to TRG). On continuous 24/7 processes, a 5-8 point gap between OEE and TRG is normal (unavoidable planned stops: sanitary cleaning, statutory breaks).
Structured input
Enter times in minutes and parts as whole numbers. An automotive assembly line preset (8 h, 3,600 parts) is preloaded.
Format: category;duration_min;cause. Categories: breakdown, changeover, micro, slowdown, planned.
Actions
How to measure on the shop floor?
- TO: shift boundaries (start of shift → end of shift).
- Planned stops: HR schedule + preventive maintenance.
- Breakdowns, changeovers: CMMS or work orders.
- Minor stops: impossible without an automatic counter → PLC, encoder, cycle sensor.
- Good / produced parts: final scan or line count.
Guided exercises
Each exercise has a ready-to-load dataset (“Load a preset” menu above) and a worked interpretation. Work through the analysis before clicking the solution.
easy EX1 — Automotive assembly line — high availability, degraded quality
Context. Automotive assembly line, 8-hour shift (480 min). Reliable machine (few breakdowns), but quality is struggling on one campaign: high startup scrap after a part-number changeover. Theoretical cycle time: 6 s/part. Question: which component is dragging OEE down?
See the solution
Expected result: Availability ≈ ~ 94% · Performance ≈ ~ 95% · Quality ≈ ~ 90% · OEE ≈ ~ 80% · Verdict high performing
Interpretation. Strong availability (94%) and decent performance (95%), but quality only 90% → the limiting component is quality. Priority improvement lever: Pareto analysis of scrap, shorter ramp-up time after changeovers (SMED + startup standard with first-part validation).
Pitfall to avoid. Pitfall: focusing on breakdowns (8 min, marginal) while the 360 scrapped parts cost 6 × 360 = 2,160 s = 36 min of lost useful time. Always compare losses in the same unit (minutes) before prioritizing.
intermediate EX2 — Plastic injection press — performance degraded by minor stops
Context. 250 t injection molding press producing a technical part. 8-hour shift. The operator reports that 'it's running badly' with no outright breakdown. Theoretical cycle 45 s/part. Question: is the press reliable?
See the solution
Expected result: Availability ≈ ~ 82% · Performance ≈ ~ 78% · Quality ≈ ~ 97% · OEE ≈ ~ 62% · Verdict average
Interpretation. Excellent quality (97%), but performance only 78%: the 55 min of minor stops + 25 min of slowdowns are the real problem. Typical of injection molding: clogged nozzle, ejection faults, pick-up robot hitting its end stop. These losses are invisible because they are short → install an automatic counter (PLC or cycle sensor) before continuing to work 'by feel'.
Pitfall to avoid. Pitfall: performance is the hardest component to measure without instrumentation. Without a counter, operators always underestimate minor stops. Rule: if availability is good but OEE is low, suspect performance before blaming the machine.
easy EX3 — Bottling line — close to world class
Context. Mineral water bottling line, nominal rate 800 bottles/min (cycle 0.075 s). Continuous 8-hour production. Experienced team, preventive maintenance up to date. Question: what level is this line at, and what is left to gain?
See the solution
Expected result: Availability ≈ ~ 93% · Performance ≈ ~ 97% · Quality ≈ ~ 99% · OEE ≈ ~ 89% · Verdict world class
Interpretation. OEE ~89% → world class (> 85%). Remaining margins: cut the 12 min of breakdowns through predictive maintenance (pump vibration, motor thermography). Beyond that, any optimization costs more than it gains. Focus the effort elsewhere in the plant.
Pitfall to avoid. Pitfall: going from 89 to 95% costs more than going from 60 to 75% elsewhere in the plant. The law of diminishing returns hits OEE hard. Benchmark the whole plant, not a single line.
advanced EX4 — CNC machine — frequent changeovers, OEE vs TRG
Context. 5-axis machining center, small-batch production (10-30 parts). Frequent changeovers (6 per shift). Cycle 180 s/part. Question: OEE is low (50%) — is the machine to blame, or the product mix?
See the solution
Expected result: Availability ≈ ~ 74% · Performance ≈ ~ 92% · Quality ≈ ~ 97% · OEE ≈ ~ 66% · Verdict average
Interpretation. Availability of 74%, dragged down by 90 min of changeovers (19% of TR). OEE = 66%, but calculating TRE (OEE with changeovers excluded) gives ~82%. Here the machine is healthy — it is the product mix that weighs. Two levers: (1) SMED to go from 15 to 5 min per changeover (an easy 30%), (2) group work orders by tooling family.
Pitfall to avoid. Pitfall: comparing the OEE of a small-batch CNC machine with the OEE of a mass-production line is absurd. Always compare TRE (which excludes changeovers), or compare with a benchmark for the same type of machining.
intermediate EX5 — Extruder — viscous slowdowns
Context. PVC profile extruder, nominal rate 25 m/min (i.e. 1 m every 2.4 s). Continuous production, no changeovers. Problem: actual output averages 18 m/min. Question: calculate the performance component.
See the solution
Expected result: Availability ≈ ~ 97% · Performance ≈ ~ 74% · Quality ≈ ~ 99% · OEE ≈ ~ 71% · Verdict average
Interpretation. Excellent availability, excellent quality — the 74% performance is what kills OEE. The 85 min of slowdowns = chronic under-speed. Classic extrusion causes: material viscosity not optimized, heating-zone temperatures set too low (fear of defects), downstream haul-off limited by cooling. Work with the team leader + the materials lab.
Pitfall to avoid. Pitfall: in a continuous process, slowdowns stay invisible until you measure real-time throughput. An encoder on the haul-off costs €200 and reveals the problem immediately. Never estimate performance 'by eye'.
intermediate EX6 — Packaging line — startup scrap
Context. Primary food packaging line, 3 format changes during the day. Cycle 0.5 s/sachet. The 3 startups generate significant scrap. Question: what is the impact on quality?
See the solution
Expected result: Availability ≈ ~ 81% · Performance ≈ ~ 92% · Quality ≈ ~ 95% · OEE ≈ ~ 70% · Verdict average
Interpretation. Quality 95%, but of the 2,200 scrapped units, 1,800 come from startups (82% of the total). Clear lever: first-part standard, gradual speed ramp-up, or keeping the line hot between similar formats. Potential gain: +3 OEE points by eliminating half of the startup scrap.
Pitfall to avoid. Pitfall: lumping quality together without separating startup scrap, steady-state scrap and end-of-run rejects. These are 3 problems with 3 different solutions (SMED, process control, compliant line clearance).
advanced EX7 — Welding robot — long breakdown vs minor stops
Context. Car-body welding robot cell, cycle 35 s/part. A major breakdown (torch replacement) stopped the cell for 45 min. After restart, everything looks normal. Question: does a one-off breakdown hurt OEE as much as a pattern of minor stops?
See the solution
Expected result: Availability ≈ ~ 85% · Performance ≈ ~ 95% · Quality ≈ ~ 99% · OEE ≈ ~ 80% · Verdict high performing
Interpretation. OEE of 80%, high performing, despite 45 min of breakdown. A documented long breakdown does less structural damage than a chronic pattern of minor stops — it is visible, measured and can be analyzed (5 whys, spare part). Minor stops, on the other hand, erode OEE silently. Priority: the breakdown has already been dealt with; focus resources on the 8 remaining minutes of minor stops.
Pitfall to avoid. Classic pitfall: fixating on the visible (emotionally striking) breakdown and ignoring minor stops (8 min/day × 250 days = 33 h/year lost). Annual minor stops often add up to more than the annual breakdowns.
advanced EX8 — Ceramic kiln — TO 24/7, TR/TO ratio is key
Context. Ceramic tunnel kiln running 24 hours a day, 7 days a week. Theoretical cycle: 1 part every 30 s. The night shift has less scheduled maintenance but more unplanned events. Question: how should OEE be interpreted on such a continuous installation?
See the solution
Expected result: Availability ≈ ~ 93% · Performance ≈ ~ 95% · Quality ≈ ~ 98% · OEE ≈ ~ 86% · Verdict world class
Interpretation. OEE of 86%, world class on a continuous 24/7 process. Here TR = TO − planned stops = 1,350 min, and TRG (relative to TO) = ~82%, different from OEE (relative to TR, ~86%). Both indicators coexist: TRG for management (installation view), OEE for the production manager (view of the machine as something that can be steered).
Pitfall to avoid. Pitfall: confusing OEE and TRG on a 24/7 installation. TRG = TU / TO includes the (unavoidable) planned stops — an artificial penalty. OEE = TU / TR is the reference for running production. Always state the baseline in a report.
Trainer or production manager? See the full teaching guide (2-hour session plan, FAQ, quiz, AFNOR NF E60-182 normative references).