Material Insights · Understanding the Material World

Statistical quality control

Control charts tell normal variation from real problems, so quality is built in.

Making things · Production systems · No. 495

  • Electrical and chemical age
  • Readiness 9 of 9
  • Process
Making things · Technology

A memo about a page long, written in 1924, taught factories to stop chasing every wobble and act only on real faults.

Chart: Illustration: Statistical quality control

Where it comes from

Chart: Map of where it comes from

Walter Shewhart, a physicist at Western Electric, proposed the control chart in a memo of May 1924 while studying inspection data for the Bell telephone system. His book of 1931 set out the method in full.

Three things to know
  1. Shewhart's chart sampled parts during production instead of waiting for the end of the run.
  2. With limits at three standard deviations, a process working normally gives a false alarm only 0.27% of the time.
  3. During the Second World War the American Standards Association issued war standards for control charts in 1941 and 1942.
The journey
  1. 1 SampleMeasure a few parts at regular times during production
  2. 2 PlotMark the sample averages or ranges on a chart
  3. 3 LimitsDraw lines three standard deviations either side of the centre
  4. 4 SignalA point outside the lines points to a real cause
  5. 5 ActFind and remove the cause, then keep watching
Covered in What engineers do: rules of thumb, the method and the kinds of engineering knowledgeEngineeringControl and systems: feedback, stability and how machines are made to behaveEngineeringDesign, failure and safety: how engineering knowledge growsEngineering
Go deeper pick a tab to open it in full