U Chart
U chart is a statistical process control chart for defects per unit (nonconformities per unit), used when each inspected item or area can contain multiple defects and the subgroup size varies. It plots the number of defects divided by the number of units inspected, with control limits based on the Poisson distribution, adjusting for varying sample sizes.
On the shop floor, a u chart tracks the rate of defects per unit when each piece can have multiple nonconformities and the number of pieces inspected changes between lots. For example, in a CNC machining cell, inspectors count burrs, chatter marks, or scratches per part, but lot sizes vary from 10 to 500 pieces. The u chart normalizes by plotting defects per unit (u = c/n) rather than raw counts, with control limits that adjust for each subgroup's size using a Poisson-based formula. In edgebanding, operators monitor defects per linear meter of banding—like glue voids, edge lift, or chip-out—across cabinets of different lengths. The center line represents the average defect rate from historical data. When a subgroup's defect rate falls outside the varying limits, the process likely needs investigation for special causes such as dull tooling, resin variation, or operator error. This chart is preferred over a c chart when inspection opportunity is not constant, providing a true picture of process performance over time.
When is a u chart preferred over a c chart?
When the number of inspected units, area, or opportunity for defects changes between subgroups; the c chart is for constant sample size, while the u chart normalizes defect counts per unit.
What distribution assumption is usually behind the limits?
The u chart commonly uses a Poisson model for defect counts per unit.
What does a point above UCL mean on the shop floor?
The defect rate for that subgroup is statistically higher than expected from the baseline process, indicating a likely special cause such as dull tooling, resin starvation, poor fixturing, operator setup error, or incoming material variation.