Troubleshooting

Why Your Bottling Line Numbers Are Wrong: Common Mistakes and Fixes

The mistakes that quietly wreck your fill-rate, OEE and giveaway numbers on a bottling or canning line, each with a symptom, a root cause, and a numeric fix.

Symptom: your Filler OEE reads 92 percent but the warehouse receives far fewer good cases than that implies. Root cause: you are mixing speed loss into availability, or counting rejected containers as good. OEE is availability times performance times quality, and each must use the same clock. If the filler runs 6.5 hours of a scheduled 8, fills at 380 of a rated 400 bottles per minute, and rejects 1.8 percent, real OEE is 0.8125 times 0.95 times 0.982, about 75.8 percent. A 92 reading almost always means quality or micro-stops were dropped. Recompute in the Filler OEE tool with one consistent time base and every reject counted.

Symptom: liquid giveaway costs thousands per month but the fill target looks fine on paper. Root cause: filling to the nominal label volume instead of accounting for measurement variation. A 750 mL wine bottle held at a 754 mL mean to guarantee no underfills gives away 4 mL per unit, and across 40,000 bottles that is 160 liters of free product. The fix is statistical: set the target so mean minus two to three standard deviations still clears the legal minimum, not the label. Tighten filler standard deviation from 3 mL to 1.5 mL and you can lower the mean by roughly 4.5 mL. Model the money in Liquid Giveaway Cost before changing setpoints.

Symptom: pack-out and fill-rate numbers disagree by several percent and nobody trusts either. Root cause: unaccounted container loss between the filler and the case packer. Bottles break, cans crush, and rejects get pulled, so 100,000 filled units may only yield 98,700 packed. That 1.3 percent gap is not scrap noise, it is real material you paid for. Track it explicitly in Container Loss and reconcile Fill Rate against Pack-Out Rate at the same shift boundary. If the two tools use different start and stop times, the delta is a timing artifact, not a process loss, and chasing it wastes maintenance hours.

Symptom: cap or closure consumption runs 2 to 4 percent over units produced and stockouts hit mid-shift. Root cause: treating one cap per bottle as a hard rule while the capper spits, cross-threads and rejects closures. A torque reject rate of 1.5 percent plus 0.8 percent feed jams means you consume 102.3 caps per 100 sealed bottles. If purchasing orders exactly to unit count, you run short. Use Cap Usage and Closure Usage with the real reject fraction built in, then add safety stock equal to at least one hour of loss. At 400 per minute that is 24,000 caps, not a rounding allowance.

Symptom: label roll changeovers keep stopping the line unexpectedly. Root cause: estimating roll life from label count only, ignoring web length, splice losses and the 30 to 60 seconds each splice costs. A roll of 5,000 labels at 380 per minute lasts about 13.2 minutes, so a 10-hour run needs roughly 45 rolls and 45 splice events. If each splice averages 45 seconds, that is 33.75 minutes of stops you never scheduled. Plan changeovers with Label Roll Timing so roll depletion aligns with existing stoppages, and stage rolls before the reel empties rather than reacting when the sensor trips.

Symptom: efficiency looks strong on running hours but weekly output falls short. Root cause: excluding CIP and sanitation from the denominator. A dairy or juice line may burn 90 to 120 minutes per day on clean-in-place, and if Line Efficiency is computed only on post-CIP uptime, it hides a 15 to 20 percent scheduled loss. Report efficiency two ways: mechanical efficiency on available time, and overall on scheduled time including CIP. Feed the sanitation window into the CIP Downtime tool and hold it separate from breakdowns, because conflating a planned 100-minute wash with unplanned faults sends maintenance chasing a problem that does not exist.

Symptom: rated line speed and actual throughput never match no matter the settings. Root cause: unit and rate confusion, usually bottles per minute versus containers per hour, or nominal filler rate versus system rate limited by the slowest machine. A filler rated at 500 bpm behind a labeler that tops out at 420 bpm delivers 420, not 500, and averaging the two gives a fantasy number. Always compute Fill Rate at the constraint, convert every station to the same unit and time base, and verify with a stopwatch count over 60 seconds. A single decimal or a per-hour figure mistaken for per-minute throws estimates off by a factor of 60.

Symptom: a new SKU's numbers look right in isolation but the whole line underperforms after the switch. Root cause: forgetting that container size changes every downstream ratio at once. Moving from 500 mL to 330 mL cans raises can count per liter, cap or seam consumption, label draw and case-pack ratios together, and a spreadsheet updated for only one of them drifts fast. Rebuild Pack-Out Rate, Container Loss and closure counts on the new format before committing to a schedule. The catch-all check is a mass balance: liters in must equal liters filled plus giveaway plus loss, within about 0.5 percent, or an input is wrong somewhere.

Published 2026-07-01.