OEE Troubleshooting

OEE and Factory Performance Mistakes That Wreck Your Numbers

A troubleshooting guide to the measurement and data errors that inflate or wreck OEE, availability, throughput, and downtime cost numbers on the plant floor.

The most common OEE mistake is reporting a number above 100 percent or a suspiciously flat 85 percent every shift. Symptom: OEE that never moves regardless of stoppages. Root cause is almost always a wrong ideal cycle time. If the line was rated at 40 parts per minute in 2015 but now runs a heavier SKU at 32 per minute, performance is silently benchmarked against the wrong denominator. Fix: pull the true ideal cycle time per SKU from a 30 minute best-run study, not the nameplate. Re-run the Performance Efficiency calculator per product, and a phantom 92 percent line often drops to a real 74 percent.

Confusing planned and unplanned downtime destroys the availability figure. Symptom: availability sits at 96 percent while operators complain the line is always down. Root cause is scheduling breaks, changeovers, and planned maintenance inside loading time, so they never count against you. Decide your denominator once: OEE uses planned production time, TEEP uses all 1,440 minutes per day. If a shift is 480 minutes with 60 minutes of planned changeover, feeding 480 into the Availability Calculator instead of 420 overstates availability by roughly 4 points and hides real losses.

Micro-stops and speed losses vanish when data is logged by hand. Symptom: availability looks fine at 94 percent but throughput misses target by 15 percent. Root cause: stops under 2 to 3 minutes never get written down, so they leak into performance instead of availability, and often into neither. A line losing 40 micro-stops per shift at 90 seconds each bleeds 60 minutes, about 12 percent of a 480 minute shift. Fix: capture stops automatically from the PLC at 1 second resolution and reconcile logged downtime against the Throughput Gap calculator until the two agree within 5 percent.

Double-counting scrap between performance and quality is a classic unit error. Symptom: OEE reconciles fine but good-parts count does not match finished-goods inventory. Root cause: rework passed on second attempt gets counted as good in the Quality Rate Calculator while the extra cycles are ignored in performance. A part reworked once consumed two cycles but shows as one good unit, inflating both metrics. Fix: count total parts started, not good parts, in the performance term, and count only first-pass good in quality. First-pass yield of 88 percent frequently masks a rolled throughput yield near 79 percent.

Averaging OEE across dissimilar lines produces a meaningless plant number. Symptom: plant OEE reads 82 percent but no individual line hits it. Root cause: simple averaging weights a 12 part per minute packing line the same as a 400 part per minute filler. Fix: weight by planned production time or by units, or better, do not average at all. Run each asset through the OEE Calculator separately and manage the bottleneck. A plant average of 82 percent can hide a constraint asset stuck at 61 percent that caps the entire line.

Ignoring the bottleneck when reading throughput wastes improvement money. Symptom: you buy a faster upstream machine and total output does not change. Root cause: the constraint moved or never was where you improved. If station 3 runs at 28 parts per minute and everything else exceeds 45, only station 3 sets the rate. Fix: identify the true constraint with the Bottleneck Impact calculator before any capital spend, then model recovered units against the Downtime Cost Calculator. One minute of constraint downtime on a line contributing 300 dollars per minute of margin is a 300 dollar loss, not an averaged 90 dollars.

Mixing time units and rate units silently corrupts every downstream figure. Symptom: line efficiency comes out at 140 percent or 12 percent, obviously wrong. Root cause: seconds mixed with minutes, or parts per hour fed into a per-minute field. A 3,600 versus 60 error is a 60x swing. Fix: force one unit basis across the whole model, usually minutes and parts per minute, and sanity-check with the Line Efficiency calculator so results land between 40 and 95 percent. Any efficiency above 100 percent means the ideal cycle time is understated or the count is inflated, so stop and audit inputs before trusting it.

Chasing smart-factory investment without a baseline is the costliest process failure. Symptom: an IoT or automation project delivers sensors but no measurable OEE gain. Root cause: no clean pre-project baseline, so the 6 point improvement claim cannot be defended. Fix: lock 90 days of stable OEE data before go-live, then measure the delta. Feed real recovered minutes and scrap reduction into the IoT ROI Calculator and Automation Payback Calculator. A project promising a 2 year payback on 400,000 dollars needs about 550 dollars per day of recovered value, and vague point improvements rarely clear that bar without hard baseline data.

Published 2026-07-01.