We use cookies to optimize our website and to provide you with the best experience. By continuing to browse this site you agree with our cookie policy
O'PEEP'S CPK EXPLANATION
Imagine you park your rather big car in your rather tiny garage. This occasionally results in scratching your door panels? You are sometimes closer to one wall or to the other? The more distance you keep from those nasty walls, the less likely a scratch, the better your Cpk value.
Of course, you are not able to always park exactly centered even though you are always fully awake. For this short-term variation the Cpk tells you about your performance.
The Cpk Index compares the distance from the process center to the nearest Specification Limit and to the process spread.
The greater the Cpk, the better the process fits between the specifications. The underlying concept is that a process which is “more on target” is more capable than a process that is wider distributed. Have a look at the two processes below. Process A is better than Process B even though through sorting you have the same number of “Out of Specification events” in both processes.
Many organizations strive for a Cpk of 1,5. Is this a good idea?
Enjoy our reading of 3min
As you can see in the graphs, Cpk varies depending on where the center of the distribution is with respect to the specification limits but also depending on the spread (standard deviation) of the distribution:
"Cp" stands for Critical Process Capability whilst the "k" comes from Japanese and means "Katayori" which is bias.
If a process is normally distributed, Cpk and process yield are linked. The metrics process yield and scrap rate are also measures for process capability. You typically find them expressed in percent or ppm (parts per million).
For example, a Cpk of 1.33 means that less than 0.01% of the total production is scrap. See our table below.
Click here to learn more about the differences.
What are the core ingredients it takes to benefit from Process Capability? Enjoy our reading
In practical applications, statistical software does not ask if your process data is long- or short-term. Thus e.g. Minitab is estimating short-term variation from your data by using: