The U chart monitors the defect rate per unit: the total defects in a batch divided by the batch's inspection unit size (count, area, length, etc.) gives the average defects per unit. Based on the Poisson distribution, it solves the problem of comparing defect counts when inspection unit sizes are unequal.
Use it when the inspection unit size changes from batch to batch - for example when the number of products sampled per day differs, the plated area checked per week varies, or fabric roll lengths differ. With a fixed unit size, the C chart (defect count) is simpler and more direct.
The U chart limits form a stepped band because they vary with each batch's unit size - this is normal. A point above the upper limit means that batch's defect rate is significantly high - investigate the special cause. Several consecutive points above the center line indicate the defect level has risen overall and needs systematic improvement.
The center line is the average defect rate ubar (total defects divided by total inspection units); each batch's limits are ubar +/- 3*sqrt(ubar/n), where n is that batch's inspection unit size. The tool computes and plots the limits per batch from the actual n.