A first article does not show the process can keep making good parts. Make a significant production run at production speed with production tooling, people and gages, and measure capability on it. Here the full 300 piece run gives Cpk 2.25 and Ppk 1.95. Stopped at piece 150 it would have shown 2.57 and 2.65 and hidden a resin load that moved the average.
Weeks 9 and 10. A first article shows the tool can make good parts. It does not show the process can keep making them. For that, make a significant production run: production speed, production tooling, production people and production gages.
A common run size is 300 pieces. The customer sets the real number, and in this case it is 300.
The rate
The planned cycle is 36 seconds, which is 100 pieces an hour. The run took 188 minutes from first piece to last, including one 8 minute stop at piece 150 to refill the dryer hopper with a new resin load. That is 95.7 pieces an hour over the elapsed time, and 100.0 while the press was running.
The capability
All 300 pieces were measured for snap hook height, in production order, in 60 subgroups of five. The tolerance is 1.10 to 1.30. Sigma within subgroups is the average range divided by 2.326. Overall sigma is the standard deviation of all 300 readings.
The full run. Average 1.2106, average range 0.03085, sigma within 0.01326, overall sigma 0.01529. Cp is 2.51 and Cpk is 2.25. Pp is 2.18 and Ppk is 1.95. This customer's requirement is a Ppk of at least 1.67, so it passes.
Stopped at piece 150. Average 1.2018, average range 0.02967, sigma within 0.01276, overall sigma 0.01235. Cp is 2.61 and Cpk is 2.57. Pp is 2.70 and Ppk is 2.65. It looks better, and it hides something.
What the second half shows
Control limits from the first 30 subgroups: average 1.2018, average range 0.02967. Upper limit 1.2018 plus 0.577 times 0.02967, which is 1.2189. Lower limit 1.1847. Range limit 2.114 times 0.02967, which is 0.0627.
Subgroup 31, the first after the refill, has a mean of 1.2200, above the upper limit of 1.2189. Every subgroup after it sits above the center line. The new resin load moved the average up by about 0.02 mm. The spread did not change. Cpk dropped from 2.57 to 2.25 and Ppk from 2.65 to 1.95. Still above 1.67. But a run stopped at 150 would never have shown it.
The fix goes into the control plan: any new resin load triggers a five piece hook height check before production resumes. Update the PFMEA row for it too, so the two still agree.
Read more: Cpk vs Ppk and SPC control charts for beginners.
The mistake to avoid
Stopping the run when the numbers look good. The first half of a run is the easy half. Resin changes, shift changes and tool warm-up show up late.
- Find your customer's required run size and minimum capability index.
- Check your last run was not stopped early.
- Compare Cpk with Ppk on it. A big gap means something moved during the run.
Open the printable Monday Board
Related on ShopFloorOps
The part, the tool, the program and every number in this book (latch cover LC-5927) are a made-up example. The tool sidebars show ShopFloorOps tools, which I make and sell. Every chapter teaches the step itself, so you can use your own forms. New to a term? See the quality glossary.