Using Control Charts to Identify Product Process Problems

Project managers use control charts to spot production process problems before they spin out of control. On a control chart, production data is plotted and analyzed for specific trends. Control charts are used more as a preventive measure rather than for detection or rejection of quality problems. This is extremely useful since it is cheaper to prevent mistakes than to correct them.

Control charts compare quality, cost, and time issues to an established norm. They indicate permissible behavior so that aberrations are easy to identify. Analysis of control charts determines whether a process is stable or whether corrective action needs to be taken.

Control charts can also help determine sources of variation. Variation is the range the observations fall around the process mean or average. Variation is different for every product or process since each has different characteristics.

• Common cause variation - is random variation common to any process. This type of variation requires management decisions to change the basic processes. Common cause variation is caused by chance and requires no corrections.
• Special cause variation - happens at the operational, or production, level. This variation is indicated by exceeding a control limit or a persistent trend towards the limit. Special causes exist when the variation in a process exceeds allowable standards. Corrective action is then required.

Variation is also categorized by time.

• Short-term variation - can be caused by changes in suppliers or workers' performance.
• Long-term variation - occurs in cases of tool wear, environmental changes, or increased administrative control.

There are two types of control charts classified by the type of data they collect. Variable control charts are used with continuous data in which all numerical values are possible. Variable charts are useful when measurements from a process are variable such as diameters, electrical output, or chemical concentrations.

Attribute control charts are used with discrete data, or when data can only have a certain value, or range, such as "1" for "yes" and "2" for "no" in a conformance test. Attribute charts analyze data such as conforming/non-conforming, pass/fail, go/no go, or yes/no measurements.

The use of these various charts depends on what type of quality measurement is desired. The most common type is the X bar chart, or process average chart.

Limits on a control chart are often called the three-sigma limit because most companies operate within the 3 sigma limit. In a normal distribution, 99.73 percent of the measurements lie within X bar ± 3s, or within the UCL and LCL. Some companies now employ a six-sigma limit in their quality control. This allows only 2 defects per billion. This exactness in quality is so expensive that it is only possible over very large production runs.

The high figure indicates a high degree of variation because more of the observations fall away from the average. Therefore, the taller the curve shape, or the bell curve, the lower the standard deviation will be.

Control charts can be interpreted in many ways depending on their patterns and line shifts. Experience is the greatest aid to understanding a chart. Control charts tell when to look for trouble but not where the cause lies. Control charts also indicate when to leave a process alone. Variation can be unnecessarily introduced by an operator trying to fine-tune a machine to near perfection, when the control chart indicates the operator could leave the process alone. Charts are interpreted by runs, trends, periodicity, and hugging.

Quality control inspectors also use the Rule of Seven to determine if a process is out of control. If seven or more consecutive observations are found to be on one side of the mean, then it is out of control. The reason it is said to be out of control is that there is only a 1.56 percent statistical chance of random variation that the run of seven would fall on one side of the mean.

One of the most useful quality control tasks is ensuring a process is in control, by identifying the existence of a problem. Control charts are a valuable tool in determining whether or not a project or process is in control. To be able to read control charts, you need to be familiar with the different control chart types and their components, and the various methods of interpretation.