OEE never walks alone. 2 billion units show who it walks with.
We measured 25 production lines over 285 days and crossed OEE with safety, maintenance, quality and raw material data. Some results confirm what every manager suspects. Others contradict what many people still believe.
Machine data, not spreadsheet data
No surveys or opinions: the data came straight from the lines' PLCs, collected automatically by the ProMOS® software, plus records entered by the operators themselves. Everything was grouped by day and tested statistically.
The biggest loss wasn't downtime. It was speed.
Adding up all the data, OEE came to 69.4%. Break it into its three factors and you see what almost no red light ever shows: the lines lost more time running slowly than standing still.
Out of every 100 planned minutes, where did the time go?
What about TEEP? Of all calendar hours, the lines were used for production in 56.9% of them. Multiply by OEE and TEEP lands at 39.5%. Both numbers are right — they just answer different questions. More on that below.
Eight suspects. Five with a proven link.
For each indicator the question was the same: on days when it went up, did OEE go up or down with it? The coefficient r ranges from −1 to +1. The farther from zero, the stronger the link. The sign tells whether they move together (+) or in opposite directions (−).
Six things the data told us
A stable process and an efficient line are the same conversation
Process capability explained 87% of the variation in OEE (R² = 0.87). It is the strongest link in the study, now confirmed with real production data rather than simulation only. A process that varies less stops less, runs faster and scraps less.
Recovering fast mattered more than stopping rarely
Mean time to repair was more strongly associated with OEE than mean time between failures (MTBF, r = 0.56). On high-speed lines every stop charges twice: in downtime and in the ramp-up curve until the line is back to rate and quality.
More safety observations, higher OEE
Contrary to what some managers fear, days with more safety observations had higher OEE, not lower. Caring about safety did not cause more stops, lower speed or more scrap.
Logging defects didn't hurt production
Days with many maintenance tags opened did not have lower OEE. When logging is quick and easy, operators flag the problem without losing focus on the line — and the defect goes to the right queue.
Measuring only physical loss is misleading
Raw material loss showed no association with OEE. Whoever measures productivity only by what ends up in the bin can't see time: at an average rate of 522 units per minute, every minute stopped meant 522 units that never existed.
TEEP is not OEE by another name
Virtually no association. TEEP includes time when the line wasn't even scheduled; OEE measures how well it ran when it was supposed to run. Using one in place of the other creates unfair comparisons between shifts and lines.
Correlation shows that two indicators move together; on its own it doesn't prove which one causes the other. That's why every finding comes with its shop-floor explanation.
How much does a stop cost after the line restarts?
On average in the study, lines ran 84 minutes between failures and took 15 minutes to recover. What few people count is the ramp: after restart, the machine takes a while to reach full rate. That slice lands in performance, not availability.
Ramp assumed linear, from initial speed to ideal rate. Illustrative values.
Every link, in one chart
After the eight hypotheses, we crossed every indicator with every other: 36 pairs. Tap or hover over a square to read what it says.
Safety, for example, was linked not only to OEE but also to process capability, MTBF and TEEP. And MTTR was linked to the rejection rate: after a stop, the line usually needs adjusting — and that's where scrap is born.
How each indicator misleads when used alone
The research also compared the indicators plants commonly use to measure productivity. All of them have value. None of them, alone, tells the whole story.
Quantity produced
The shift that did the changeover always looks worse than the next one, which got the line ready to run. It compares volume, not efficiency.
Rejection rate
It can be improved by stopping the line at every suspicion or running slower. Scrap goes down; productivity goes down with it.
Stops (count or time)
A line can go a whole shift without stopping — while running below rate or producing defects.
TEEP as efficiency
A plant-wide meeting or scheduled preventive maintenance drags the shift's TEEP down, even if the operation was flawless.
PE with planned stops inside
It penalizes whoever did the setup. Prorating helps, but a great period can still be contaminated by earlier records.
Schedule adherence + OEE
For plants that have mastered OEE: the plan foresees every machine minute, and adherence also controls the time OEE doesn't see.
What this research changed in the way we build ProMOS®
The data showed that OEE is the output of a system. That's why, since then, ProMOS® keeps the indicators that move with it on the same platform.
Maintenance connected to production
MTTR and MTBF come from the same stops that feed OEE. The ProMOS® CMMS and MES share that data, with no re-typing.
Capability in the operator's sight
With r = 0.94, process stability is a production matter. QualityWeb SPC shows variation in real time, by product.
Safety as a habit, not a brake
Observations logged in EHS didn't steal productivity. Easy logging, right from the shop floor, lets the culture grow without stopping the line.
A tag in seconds
The maintenance-tag result showed the value of fast logging. In LABEL, an anomaly is opened in seconds and goes straight to whoever fixes it.
An indicator immune to tweaks
Data captured from the PLC and validated at entry keeps speed loss from turning into a “planned stop” on paper. The number has to be reliable before it is pretty.
The right indicator at the right time
OEE for operations, TEEP for capacity, schedule adherence for mature plants. We help you choose in our OEE Consulting.
H2W Systems · applied research + 20 years on the shop floor
The study's limits
Source: SOUZA, Jonas de. Eficiência Geral de Equipamento (OEE): um estudo da associação entre o OEE e outros indicadores da atividade produtiva [Overall Equipment Effectiveness (OEE): a study of the association between OEE and other production indicators]. Master's dissertation (Professional Master's in Business Administration) — Faculdade Campo Limpo Paulista (FACCAMP), Brazil, 2017. Advisor: Prof. Dr. Manuel Meireles.
Questions about OEE and other KPIs
Is OEE related to workplace safety?
In this research, yes: days with more recorded safety observations had higher OEE (Spearman r = 0.19, significant at 1%). Caring about safety did not cause more stops, speed loss or scrap on the lines studied.
How are OEE, MTBF and MTTR related?
Both were strongly associated with OEE: higher MTBF (r = 0.56) and lower MTTR (r = −0.70) went together with higher OEE. MTTR had the stronger link because, on fast lines, every stop also affects performance and quality during the restart.
Are OEE and TEEP the same thing?
No. TEEP multiplies OEE by calendar utilization and is affected by periods with no scheduled production. In the research, the two had no significant association (r = 0.03). Use OEE to measure operational efficiency and TEEP to measure capacity use.
Does process capability influence OEE?
It was the strongest association in the study: r = 0.94, with 87% of OEE variation explained by process capability. Stable processes tend to stop less, run closer to ideal rate and generate less scrap.
What is the best KPI to measure industrial productivity?
It depends on the plant's reality and maturity. The indicator must allow fair comparisons between shifts, lines, operators and products and, above all, be reliable: with a known calculation, accepted by everyone and immune to tweaks. For most operations, OEE broken down into availability, performance and quality is the starting point.
Want to see these indicators together in your plant?
ProMOS® brings MES, CMMS, SPC, EHS and anomaly tags onto a single platform — with OEE calculated from machine data. Start with the “What is OEE” guide or talk to us.
