There’s a rumor that the statistical measurements of Cpk and Ppk are no longer useful for statistical process control (SPC). This simply isn’t true. However, Cpk and Ppk are over-relied upon in manufacturing.
Although these powerful performance measurement calculations are very useful in manufacturing, you probably aren’t using them to your full advantage. Read on to learn how to do so.
Coming to Terms with Terms
The obvious place to begin our discussion of Cpk and Ppk is to focus on why these measurements are over-used. Four metrics are commonly used in manufacturing: Cp and Cpk, and Pp and Ppk. Let’s discuss these metrics and the terms that are often associated with them.
Cp, Cpk, Pp, and Ppk are all process capability indexes. They are used to define the ability of a process to produce a product that meets your manufacturing requirements. They help simplify the management of statistically controlled processes.
Your specifications define your product requirements. In other words, they define what is expected from an item for it to be usable (and sellable).
Standard deviation is a measurement of the variability of measured values. In the context of SPC, standard deviation is typically used to describe product variation caused by the manufacturing process. It indicates how much a product can be expected to vary from the average. The standard deviation is normally fixed for a process that’s under statistical control and can only be affected by a process change that affects the variability in the process.
Finally, the mean is the arithmetic average of a group of values.
Into the Weeds!
To make sure we’re speaking the same language, let’s dive into some more granular definitions.
Cp and Pp are both indicators of what a process is capable of. As with all process capability indexes, the higher the number the better, because this means process variation is very small compared to the spread of the specification limits. These metrics don’t convey actual performance, just a comparison of the variation to the specification limits. Cp uses short-term standard deviation and Pp uses long-term standard deviation.
Cpk and Ppk are both indicators of how a process is actually performing. They compare the mean and standard deviation to each specification limit separately to convey a sense of how centered a process is within the specification limits. Just like Cp and Pp, Cpk uses short-term standard deviation and Ppk uses long-term standard deviation.
Making comparisons between the metrics is powerful. If Pp tells us what is possible with a process, then Ppk can never be greater than Pp. The closer Ppk is to Pp, the better centered our process is. The same holds true for Cp and Cpk.
One Is the Loneliest Number
One of the reasons so many people like these metrics is that most folks don’t use all four. They tend to use Cpk or Ppk. Why? Because these indexes are good for determining at a glance how your process is performing, and folks like to compare them. People like single numbers: How much does that car cost (and compare the price to other cars)? What score did you get on your math test (and compare it to other test scores)? People like looking at a single number and comparing it to others. It’s convenient.
However, looking at just one number can be dangerous. One number does not tell the whole story.
The real power of these indexes is when you compare them. If your Cpk is greater than 1, you know you’re running within spec—but that’s all you know. Cp tells you how well you could be doing. Although your Cpk shows that you’re within spec, you could be quite far from your ideal state.
Further, although Ppk is similar to Cpk, the difference could be significant if you’re delivering a batch of product from many different production lines. Relying on short-term standard deviation makes sense if everything is delivered from a single process—but your customers don’t see that variation. They open your product and see the variation across ALL the processes that produced the production run. This is a better case for using Ppk, but there’s still the trick of rolling it up across all those processes.
A False Sense of Security
Many people over-rely on Cpk and Ppk simply because they have a lot to keep track of in a manufacturing facility. They just want to know how they’re doing overall. They want to know their Ppk value so they can move on to the next thing.
However, statisticians will tell you this isn’t an ideal approach. You can get a false sense of security from Cpk or Ppk.
Enact Stream Grading to the Rescue
Stream grading in Enact® avoids this false sense of security. Stream grading is a streamlined way to make comparisons. It simplifies your understanding of process performance while also making it more powerful. These two benefits seem contradictory, but they aren’t. Let’s explore why.
Grading is a fallout-based metric, just like capability indexes. There’s an A-B-C component to a grade that’s just like Cp. The A-B-C assumes a centered process and uses short-term standard deviation. This tells us how well a process could run if all sources of variation were minimized. The A-B-C component is an indicator of your equipment or technology.
There’s also a 1-2-3 component to a grade that’s just like Ppk (with a twist). The 1-2-3 uses long-term standard deviation and the process mean to understand actual process performance and compares that performance to the A-B-C component. The 1-2-3 component is an indicator of your operations (equipment setup, operator training, etc.).
The biggest benefit of stream grading is that based on the grade, you know what kind of issues you’re facing and can prioritize your resources to address those issues. Another massive benefit of stream grading is that stream grades are designed to be rolled up—so you can see the performance for a given feature (e.g., net weight) across all the processes and parts where the feature is measured. This is something capability indexes just weren’t designed to do.
To make things visually simpler, grades in Enact stream grading are color-coded on a green-yellow-red spectrum for easy interpretation. Here’s an example of how grading can be presented :
*Click the image for greater detail
It’s easy to see on this report that “Net Contents” at the “London” site is having issues, with a red “C3” grade. Let’s take a look at some of the other grades in this table:
A1 – The “A” indicates the process is capable of running with almost 100% yield and the “1” indicates the process is running with at least 95% of that capability. This means the equipment is highly capable and the operations are doing a great job running to that capability. Processes with this grade are model processes for others to reference.
A3 – The “A” indicates the process is capable of running with almost 100% yield and the “3” indicates the process is running with less than 90% of that capability. This means the equipment is highly capable, but the operations aren’t leveraging it. This is likely “low hanging fruit” for quick improvements via some training or setup changes.
C1 – The “C” indicates the process isn’t capable of running with a yield greater than 99.73% and the “1” indicates the process is running with at least 95% of that capability. This means the equipment is having issues and the operations are doing the best they can running the equipment. This is likely an opportunity for equipment maintenance or replacement.
C3 – The “C” indicates the process isn’t capable of running with a yield greater than 99.73% and the “3” indicates the process is running with less than 90% of that capability. This means the equipment is having issues and the operations are also having issues. This is likely a candidate for a major improvement initiative.
These pieces of knowledge make it easy to identify which sites need help and what kind of help they need. Grading is made to be rolled up, which means these grades can be drilled into for more discovery. Although a site can have an overall grade, there are likely processes within that grade that are running well and others that need to be addressed. This drill-down functionality means it’s easy to find the sites that need assistance and then determine more specifics about where they need help. For example, drilling into the London site shows this:
*Click the image for greater detail
With an understanding of grading, it’s easy to drill into a site and see where it needs assistance. You won’t get these kinds of insights from a table of Cpk or Ppk values. And trying to gain these insights across all the processes in a site—especially across many sites—is almost impossible without stream grading.
Get the Answers You Need
Enact allows you to conquer your over-reliance on effective but often over-simplified tools like Cpk and Ppk. Why pore over tables of individual numbers when you can see the result of your quality data summarized into an actionable result that shows you where to dig in and helps you understand the details once you do?
You’re busy, and you want answers that allow you to breathe easier and move on. But you also need accurate and complete answers. Get them with Enact stream grading.