Statistical Process Control (SPC) is a group of tools and techniques used to determine the stability and predictability of a process. Graphical depictions of process output are plotted on Control Charts. The first Control Charts were developed by Walter Shewhart at Bell Labs in the 1920’s. At this time, telephone technology was in its infancy with poor reliability. Shewhart used SPC to study variation and reduce special causes of failure. Quality and reliability in phone service increased dramatically as a result of SPC. W. Edwards Deming is credited for introducing SPC to the Japanese after World War II. The resulting rise in Japanese quality and reliability is well documented.

Quality-One provides intelligent Consulting, Training and Project Support of SPC. The proper use of Control Charts brings stability and predictability to key processes. Q-1 has the expertise you need to experience the greatest benefits of your SPC activities.

What Do Control Charts Do?

Control Charts indicate instability of the process being studied. The user of the chart recognizes instability and determines a course of action to remove the Out of Control condition. The subsequent improvement process becomes a new and improved current state. SPC continues until the stability of the process is In Control and the process is capable. SPC is utilized to understand a process and achieve higher levels of quality. The continued use of Control Charts once a process is stable and capable is not recommended as cost effective.

What Control Chart Types Exist?

There are two categories of Control Charts: Attribute and Variables.
Attribute Control Charts are used to monitor an organization’s progress at removing defects that are inherently present in a process.

Examples of Attribute Control Charts include:

  • p chart: Proportion of nonconforming units in a sample.
  • np chart: Number of nonconforming units in a sample.
  • c chart: Number of defects per unit where the sample size is 1.
  • u chart: Average number of nonconformities per unit.

Defects must exist for Attribute Control Charts to be effective and are meant to be prevented, not measured. If defects are present, Attribute Control Charts can be used to track progress. Variables Control Charts should eventually replace Attribute Control Charts.

Variables Control Charts monitor process parameters or product features. A variable’s measurement can indicate a significant change in process performance without producing a nonconformance. As a result, Variables Control Charts are more sensitive to change and are more efficient than Attribute Control Charts. A change in process behavior signals a desired or undesired need for investigation and action. Two primary statistics are measured and plotted on a Variables Control Chart: central tendency and process dispersion. Central tendency is the process Mean or X Bar. Dispersion is either the Standard Deviation or Range.

Examples of Variables Control Charts include:

  • X Bar and R consist of a pair of charts:
    • One to monitor the process Standard Deviation, as approximated by the sample moving range.
    • One to monitor the process Mean.
    • X Bar and R charts plot the Mean value for the quality characteristic across all units in a sample and the range of the quality characteristic across all units in the sample.
  • IMR (Individual Moving Range) consists of a pair of charts:
    • The individuals chart, displays the individual measured values.
    • The moving Range chart displays the difference from one point to the next.
  • Multi-Vari Chart:
    • Similar to X Bar and R except multiple characteristics are plotted against each other.
    • The assumption is that the two or more characteristics are related in some manner.

When Are Control Charts Used?

Control Charts of a variable’s nature are used to monitor and improve characteristics considered to be special. Special Characteristics are features or parameters, directly linked to customer satisfaction holding greater risk than other characteristics of a product or process. Special Characteristics are selected during Failure Mode and Effects Analysis (FMEA). The characteristics at risk require special controls, such as:

  • Error / Mistake Proofing (Poka Yoke)
  • SPC and Process Capability
  • Quality Controls and Inspection

How Are Control Charts Analyzed?

Control Charts use Control Limits determined from process performance. Charts that show Out of Control conditions indicate that current performance of the process is different from the past. It does not mean defects are being produced. This behavior requires investigation and potential improvement to stabilize the process. Control Chart analysis is based on Normal Distribution and the expected percentages under the bell curve.

Examples of Out of Control Behavior include:

  • Run: 7 points on either side of the central tendency line, X Bar in a row (50/50 Probability Rule).
  • Trend: 7 points in one direction but still inside the control limits, not likely under normal conditions.
  • Outside the Control Limits: Any point outside the limit lines is not likely normally.
  • Zone Analysis: Points in zones that do not support the percentages expected under the normal bell curve.

We can train your team to recognize abnormal process behavior and improve stability and process capability.

What is Capability?

Capability or Process Capability refers to the statistical position of the normal distribution compared to the product or process specification. A process is capable when a bell curve is created by +/- 3 Standard Deviation and fits easily inside the desired specification. Indicators of capability are calculated based on the number of Sigma or Standard Deviations fitting between the process Mean and the closest specification.

  • Cp or Cpi is the measurement of the ratio of Six Sigma divided into the allowable specification. Cpi does not indicate how well the process is performing, rather how good it could be.
  • Cpk is a typical indicator used to describe actual process capability. Cpk is used to determine the number of defects that are being produced, even if none have been found up to this point.
  • The Cpk is the capability on the K side of the distribution. The K factor, or side, has the most risk and therefore is the worst of two possible measurements in a bilateral specification.
    • Cpk of 1.33 indicates 4 Sigma Capability or 4/3rds.
    • Cpk of 1.67 indicates 5 Sigma Capability or 5/3rds.

The greater the Cpk the less likely nonconformance will be present.

Ppk is an index similar to Cpk but considers more sources of variation in the process over a longer period of time.

Capability is often misunderstood or considered a difficult concept. Most people are unaware that they are affected by Capability while driving to work. Here is an example of Capability in effect:

  • Specification: width of one lane.
  • Process: our car and its variation.
  • Driving in our lane over a distance, the car easily stays within our allowable variation. We seldom hit the guardrail or oncoming traffic, indicating a very capable process.
  • Cpk is likely to be 1.33 or greater.
  • Larger vehicles have a smaller Cpk and smaller cars a larger Cpk.