A project improves a process, the report is presented, the team moves on. Six months later the rejects are back. Nothing dramatic happened. A setting was changed, a new operator skipped a step, the warm-up was shortened to save time.
Control is the phase that prevents this, and it is the phase most often skipped.
The X-bar and R chart
For measured characteristics produced in series, the X-bar and R chart is the standard. Every hour, five consecutive parts are measured. Two points are plotted: the mean of the five (X-bar) and their range (R).
The limits come from the process itself, calculated from at least 20 to 25 subgroups:
- center line of the X-bar chart = mean of the subgroup means
- X-bar limits = center line ± A2 × average range, with A2 = 0.577 for subgroups of five
- upper R limit = D4 × average range, with D4 = 2.114
In the bore example after improvement, the mean of means was 24.001 mm and the average range 0.014 mm. The X-bar limits are 23.993 mm and 24.009 mm, the upper R limit 0.030 mm.
Control limits are not tolerances
The tolerance, 23.970 to 24.030 mm, comes from the drawing. The control limits come from the process. A point outside the control limits means the process has changed, even if every part is still within tolerance. That is the moment to act, before defects appear.
Never draw tolerances on a chart of subgroup means. It invites the wrong reaction.
Signals and reactions
Typical signals:
- one point outside a control limit,
- seven points in a row on one side of the center line,
- six points in a row rising or falling.
The reaction is written next to the chart and practiced with every shift: stop, check the tool and the warm-up log, record what was found, inform the setter. A signal without a reaction is just a dot.
Recalculate the limits only after a deliberate process change, not because the chart looks inconvenient.
The control plan
The control plan lists every critical characteristic and process parameter in one table:
- process step and characteristic,
- limits,
- method and gage,
- sample size and frequency,
- record,
- reaction plan.
For the bore, it covers the diameter on the chart, the warm-up cycle in the machine log, the tool counter and the clamping pressure. Each line answers one question: how will we know this has drifted, and what do we do then?
Handover
The project ends when a named process owner signs the handover: control plan released, chart at the cell, work instructions updated, checks after three and six months in the calendar.
Those two checks decide whether the project succeeded. A gain that has held for six months is a result. A gain that lasted until the report is a story.