In this modern era, some of us are fortunate enough to have a process that allows cheap, instantaneous, non-destructive inspection of 100% of what we make. For everyone else, capability analysis is extremely useful. Consider wine making as an example: measurements are more difficult. As soon as you add hydrogen peroxide to a sample of wine to start measuring its chemical characteristics, you’re not putting that wine back into a bottle to sell. Getting the information that you need quickly and from as few samples as possible makes a big difference.
We use capability analysis to assess how well a process is performing with respect to specification limits. At first glance, if the measurements that you have are within specifications, then you think the process is performing well. Capability analysis goes beyond this simple, binary approach to provide insight into how well a process meets specifications. Consider the histograms below that show measurements of pH values. In the finished product, the values are to be within the specification limits of 3.3 and 3.7.
This webinar is focused on measurement system analysis, control charting, and capability analysis.
In Process 2, the measurements come close to the specification limits. In both processes, all of the measurements are within the specification limits. Because the histograms show a sample from the process, we would be much more worried that Process 2 could have measurements outside of the specification limits. Capability analysis distinguishes something like Process 1 from Process 2 so that we know where our improvement efforts are most valuable.
Many different measures of capability are in use. Two of the easiest to understand are Pp and Ppk.
Pp is a ratio that compares the distance between the specification limits to the estimated range of the process.
When the estimated range is the same as the distance between the specification limits, then Pp is 1.
The narrower the estimated range, the better the process performs, which gives higher values of Pp. Although different products and processes have different standards, you’ll often see 1.33 as a goal for Pp.
When a process is not centered, Ppk is a better representation of the processes capability to provide products that are in specification to customers. Instead of the distance between the specification limits, Ppk uses the distance from the average of the measurements to the closer specification limit. Instead of the estimated process range, Ppk uses ½ of the estimated process range. The result is a measurement that considers where the average of the process is in relation to the specification limits.
For example, the Pp value of 3.46 indicates that the process performs well even though more than half of the measurements are outside of the specification limit. The Ppk value of -0.02 indicates that the process performs poorly. Because the process performs poorly, Ppk is the better measure to see that the process needs improvement.
When you are making decisions about process improvement, it’s often not enough to know whether a sample of products are all within specification limits. We need a more detailed way to describe how effective a process is at meeting customer specifications. If you need to know where to focus your improvement efforts, then you need a capability analysis.