Out-of-Control Point on a Bottling-Line Fill Chart
A beverage bottling line fills 500 mL bottles. Net-content tolerance is 495 to 505 mL. Every 30 minutes, operators weigh five consecutive bottles and plot the subgroup mean and range on an X-bar and R chart. The last 20 subgroups gave a grand mean of 501.2 mL and an average range of 6.8 mL. Using the standard constants for n=5, the X-bar limits are approximately 497.3 to 505.1 mL. One subgroup mean plotted at 505.8 mL. The operator immediately turned the filler adjustment down. The next four subgroup means averaged 498.9 mL, and one individual bottle weighed 494.6 mL. A gauge study completed last month reported a gauge repeatability and reproducibility of 3.8 mL, about 38% of the tolerance width. The filler has 24 valves; no valve-specific data are charted. Product changeover occurred two hours before the high subgroup.
1. Was the operator's adjustment justified? Explain using control-chart and measurement-system evidence.
Read the worked answer
The point is an out-of-control signal, but the adjustment was not justified as executed. A single subgroup beyond a 3-sigma limit should trigger investigation, not immediate tampering. A product changeover two hours earlier is a plausible special cause, and the gauge R&R of 3.8 mL is 38% of the 10 mL tolerance, so the 505.8 mL result is within measurement noise of the limit. Turning the filler down moved the process mean to 498.9 mL and produced a 494.6 mL bottle, below the legal minimum. That is over-adjustment: reacting to common-cause variation increases variability. The correct sequence is to verify the measurement, check the changeover and valve behavior, confirm a special cause, and then adjust only if the process mean is genuinely off-target. Recalculate limits after a known process change.
2. What actions should be taken before the chart is used again for routine adjustment?
Read the worked answer
Repeat the gauge study with a proper design that includes the 24 filler valves and the operators. If GRR remains above about 10-30% of tolerance, improve the scale or fixture, increase sample size, or use an individuals chart for net content. Stratify data by valve; a single valve may be causing the high fill. Verify the scale calibration with certified weights. Recompute control limits only from stable, post-changeover data. Train operators on the difference between a signal and a cause. Add a rule for individual bottles and legal net content, because subgroup means can hide a single underfill.
3. How should the 494.6 mL bottle be handled?
Read the worked answer
Quarantine the 494.6 mL bottle and retain the lot. Re-weigh it on a calibrated scale and check the fill record for that valve and time. If confirmed, treat it as a net-content nonconformance: evaluate the lot against the applicable legal and customer requirements, document the check, and correct the valve. Do not simply re-label or discard the record. The chart is a process tool; conformance is decided on individual packages. Notify quality and production, and include the event in the MSA and valve-maintenance review.