Statistical process control

• Quality management
• Environmental management
• Health and safety
• Integrated management
• Process improvement
• Contact the MCR
• Customer feedback
• Become a registered consultant

Statistical process control (SPC), as its name implies is based on checking the process, not the product. If materials and processes are OK then, in theory, it is not possible to produce defective product. SPC generates a 'picture' of a process.

All processes are inherently subject to variation and common causes (background 'noise') are associated with the working environment eg differences in batches of material, things that wear out etc. Common causes are generally easy to spot and give rise to the normal distribution. If we only have common causes the results will be predictable.

SPC does not solve any problems - it identifies where and when there is a problem and identifies where an adjustment to the process, or corrective action, is required to eliminate the possible occurrence of defective product. It also identifies where adjustments are not required. Subsequent analysis of data can then be used as a tool for quality improvement. SPC helps us to work smarter, not harder and should be documented and operator-driven with adequate training in use and implementation given to all concerned.

Special (or assignable) causes of variation are usually hidden, not easy to spot and happen infrequently. They could seriously affect quality. The effect of a special cause may not necessarily result in items out of specification, even with slightly more variation the items may still be acceptable but we have lost control of the process and can no longer be sure that it will continue to produce good parts.

A control chart is a graphical comparison of process performance data compared to control limits based on previous performance of the process. Note that the chart is process-based not requirement-based ie the actual requirements of the output from the process are not taken into consideration. The prime use of a control chart is to detect special causes of variation in the process.

Control limits are drawn on an SPC chart as a guide to the past operation of the process and also to predict the future operation of the process whilst it is in control.

The position of the control limits is calculated from the collection of data collected at random when the process is operating under normal conditions. Such samples will represent the distribution of the output from the process. A control chart will show if something changes, even slightly and there are established ways that a control chart can indicate an out of control condition.

Many SPC applications consider attributes only eg defectives (number of faulty items, regardless of magnitude of fault) or defects (number of faults per item, again regardless of magnitude of fault). Attribute charts can be used to measure one specific characteristic (were there any defectives for a specific reason or how many of a specific defect was observed) or many characteristics (were there any defectives for any reason or how many defects were observed for any reason).