Control charts are constructed by plotting samples of your process output collected over time. There are several types of control charts, and the one that you choose depends on the type of output that your process produces and the sampling plan that you use to collect it.

Tracking variables. If you record process parameters or things that you can measure, such as temperature, diameter, or flow rates, then you should use one of the variable charts. They tell you how your process is running. Specifically, they compare the actual machine readings to the set point or target values, and they tell you how well your finished product meets specifications. The available variable charts, according to sampling plans, are as follows:

  • Sample of several measurements. If at each collection period you take several measurements, then you should use an X-bar chart with either the R chart or s chart. The R chart measures spread in terms of the sample range while the s chart measures spread in terms of the sample standard deviation.
  • Sample of one measurement. If at each collection period you take a single measurement, then you should use an individual and moving range chart. This chart measures spread in terms of the range of two or more consecutive samples.

Tracking attributes. If you want to track defects, defective units, or other things that you can count, such as defective parts, cracks, bubbles, or late shipments, you should use an attribute chart. They classify defects into categories so you can monitor a specific defect type. They also tell you types of defects and how they occur. The available attribute charts, according to the type of attribute tracked and sampling plan, are as follows:

  • Defective units. In situations where you are tracking whole units that are unusable, you should use a p chart or np chart. The difference between these charts is that the p chart can be used when the size of the sample varies between collection periods.
  • Number of defects. In situations where you are tracking blemishes on units, you should use a c chart or u chart. The difference between these charts is that the u chart can be used when the size of the sample varies between collection periods.

Multiple charts can be used for process control. For example, a radio manufacturer can use a u chart to track cosmetic defects and a p chart to track units that don't work.

General guidelines for assessing process control. From these plots, you can determine whether the process is functioning within prescribed limits. Random variation within those limits indicates the process is stable. Nonrandom variation indicates that something is wrong. Any of the following symptoms in a control chart indicate a possible unstable process:

  • Any number of points outside the control limits.
  • A run of at least 8 points in a row falling on one side of the process centerline.
  • A trend of 6 consecutive points increasing or decreasing in value.
  • A cycle or repetitive pattern.
  • Alternating values of 14 consecutive points that fall on opposite sides of the process centerline.
  • A lack of variability indicated by 15 consecutive points in the center third of the chart.
  • Excessive variability indicated by 8 consecutive points outside the center third of the chart.