Mathews Malnar and Bailey, Inc. Quality engineering, applied statistical consulting, and training services for R&D, product, process, and manufacturing engineering organizations.

Ppk Study Examples

The output from Ppk Study for each of the example problems provided with the software is shown below. Each case includes a short interpretation of the graphs. The four graphs produced by Ppk Study are:

1. A histogram with superimposed normal curve. The specification limits will also be displayed if they fall within the range of the graph.
2. A normal probability plot with superimposed fitted line and the Anderson-Darling normality test p value. The specification limits will also be displayed if they fall within the range of the graph.
3. A run chart with center line and 99% control limits.
4. A running Ppk chart with superimposed Ppk goal and 95% Ppk confidence limits.
   

Capable.ppk: The process appears to be normally distributed and in control. The lower bound of the 95% CI for Ppk exceeds the minimum requirement of Ppk > 1.33, so it appears that the performance of this process is acceptable.

Insufficient Data.ppk: Although this process appears to be normally distributed and in control, the sample size is too small to confirm that the process performance is actually acceptable. More data from this process are required.

Not Capable.ppk: This process appears to be normally distributed but the run chart indicates that the process is out of control.

Not Normal.ppk: The histogram and normal probability plot indicate that this process is not normally distributed so that the calculated process performance statistics are compromised.

Out of Control.ppk: The run chart indicates the process was wildly out of control while these data were collected so the process performance statistics are compromised.

Outlier1.ppk: Although there is evidence to suggest that this process is normally distributed, in control, and that Ppk meets its minimum requirement, there is an outlier present in the data which puts any claims about process performance at risk.

Outlier2.ppk: When further data were taken from the process considered in the preceding example (Outlier1.ppk), it became evident that the suspected outlier was not a single rare event. The histogram and normal probability plot indicate that the right tail of this distribution is exaggerated relative to that of a normal distribution and that, despite the fact that the process performance statistics exceed their goal, they are misleading - the actual defective rate of this process is much higher than they suggest.

Runout.ppk: The histogram, normal plot, and run chart show the classic behavior of a process experiencing run out – radial deviation from a target position in two dimensions. These data are not normal, so should not be analyzed using the usual process performance statistics.

Tool Wear.ppk: Despite the hints from the histogram and normal plot that this process is normally distributed, the run chart shows that this process is drifting, such as if tool wear was affecting part size. Although the process performance statistics appear to be excellent, they are invalid because of the instability in the process.

Return to MM&B Software

Return to MM&B Home Page