Insyt
All charts in the library

Flow, process and networks

Statistical process-control chart

Compares process statistics with statistically derived control limits to detect unusual variation.

  • Plot
  • Process monitoring
  • Exception detection

What it answers

Does the process show evidence of unusual variation?

Shape
Sequential points and control limits
What it shows
Potential special-cause variation.

Example

The question

Is the packing line running steadily, or has something changed?

Average packing time per shift with control limits (seconds)

Average packing time per shift with control limits (seconds)Points for 24 shifts around a centre line of 62 seconds with control limits at 56 and 68. Shift 18 breaks the upper limit, and shifts 21 to 24 sit well above the centre, signalling a real change.Seconds per order54586266701357911131517192123Centre 62Upper limit 68Lower limit 56
Average packing time per shift with control limits (seconds)Points for 24 shifts around a centre line of 62 seconds with control limits at 56 and 68. Shift 18 breaks the upper limit, and shifts 21 to 24 sit well above the centre, signalling a real change.Seconds per order5458626670159131721Centre 62Upper limit 68Lower limit 56
Points for 24 shifts around a centre line of 62 seconds with control limits at 56 and 68. Shift 18 breaks the upper limit, and shifts 21 to 24 sit well above the centre, signalling a real change.

Illustrative data. It belongs to no company.

When to use it

Use it when

Monitoring a process using an appropriate stable baseline.

Avoid it when

Specification limits substituted for control limits or an unsuitable chart type.

How to read it

Compare sequential statistics with chart-specific control limits and run rules.

Note

Control limits are not specification limits. Choose a chart suitable for the data.

The data you need

Layout
Long table
Fields
Time: date/sequence; Measurement: number; or subgroup observations and counts
One row is
One process observation or one member of a sampling subgroup. What is the grain?
Rules
Choose chart type first. Individuals charts use sequential measurements; subgroup charts need subgroup IDs; proportion charts need defect and eligible counts.
Optional
Specify stable baseline and chart-specific limits. Do not substitute specification limits. Counts example: Period,Defects,Inspected.
Preparation
Choose a chart suited to the data and subgroup structure; estimate limits from a suitable baseline and apply declared signal rules.
Units
Units depend onchart type: raw measurement, subgroup mean/range, count or proportion. Use chart-specific denominators and limits.
Example of the input layout: Long tableA long table: one row per observation, few columns. New data adds rows.TimeMeasurementor subgroupAABBC
A long table: one row per observation, few columns. New data adds rows.

Worked example

Supplied centre 10, LCL8, UCL12; observation 13 exceedsUCL. This signals under the stated chart rule; it does not by itself identify the cause.

TimeMeasurement
110.2
210.1
310.4
410.0
All charts in the library