Capability statistics are wonderful things. They tell you how well your process is meeting the specifications that you have. But there are so many capability statistics that it’s worth taking some time to understand how they’re useful together.

Two capability statistics that are hard to keep straight are Cp and Cpk. Their names are different by only a single letter. A single letter that, by the way, doesn’t really explain anything about how these two statistics are different.

Definition of Cp

The equation for Cp is often written ET / NT. ET stands for “engineering tolerance,” which is the width between the specification limits. NT stands for “natural tolerance,” which is the width that should contain almost all the data from the process. Traditionally, NT is six times the standard deviation.

We often describe Cp as the capability the process could achieve if the process were perfectly centered between the specification limits.

Definition of Cpk

The equation for Cpk is more complicated: [minimum(mean - LSL, USL - mean)] / (0.5*NT). LSL stands for “lower specification limit,” and USL stands for “upper specification limit.”

 …