The natural tolerance is the variation present in my in-control process. So I’m not going to
estimate this unless my process is in-control (stable and predictable over time).
Recall, we’ve already found this process to be in statistical control by inspecting the X-bar and R
charts.
So, we can now investigate the tolerance of this process if specification limits of 10 and 18
seconds are imposed.
Recall, we’ve already found this process to be in statistical control by inspecting the Xbar and R
charts.
So, we can now investigate the tolerance of this process if specification limits of 10 and 18
seconds are imposed.
Tolerance Chart
There is no option for a tolerance chart in JMP, but we can make a very similar looking graph
using the Variability /Gage platform. You must stack your data ahead of time and enter the
specification limits in the column information.
Cp
The Cp is a numerical measure that captures the information displayed in this histogram below.
It is a ratio of the required tolerance to the natural tolerance.
Cpk
The Cp fails to consider if the process is centered or not. There might be different
consequences for being out high than low (you can always cut something shorter but you can’t
make something longer).
This is not a new topic! Think about what you already know from Normal Theory…
Joe’s Salsa Company is considering using a new container for its salsa. According to their
research, the glass containers for the salsa should have a strength between 185 and 325 psi.
Joe’s vendor has provided the following data for evaluation. Is there evidence that the process
is in control? Is it capable of producing jars with a breaking strength between 185 and 325?
• Next, let's investigate monitoring statistics that can summarize what's going on.
• Use Cpl and Cpu to provide an estimate of fallout and the necessary assumptions when
using these ppm quantities