SPC - Stamping parts - Small batch size


Starting to get Involved
Hello everybody

Our company has followed the requeriments stated in PPAP Handbook from Ford, and it is requested to our stamping suppliers perform a SPC of 125 parts for CC.
Due to timing, for PPAP just 1 production run has been carried out ( mean, with off tool & off process conditions), so the 125 parts to be measured should be form the same batch.

For those batches from pre-series production, the bath quantity are not really bigs. Quantities are usually below 1000 parts

Do you consider such task to have added-value for this technology taking into account the batch sizes? In my opinion, a 30 parts size sampling would already be highly representative, as stamping variations should be low in the same batch. Additionally, to be able to measure from different meters of the coil, you could pick 6 groups of 5 parts for each during the production...

What is your experience regarding SPC for stamped product? How have you usually handled it?

Thanks, regards

Jim Wynne

I don't know what the Ford requirements are, but unless I've been out of the loop for too long, the default (AIAG) "significant production run" for PPAP should be a minimum of 300 pieces, with SPC based on 25 subgroups collected at intervals during the production run, with a total of at least 100 data points. This may or may not be appropriate in all circumstances, so the customer should be consulted if some other method is to be used.


Starting to get Involved
Hello Jim

Yes, you are right, a production of minimum 300 prices and, for SPC a minimum of 100 data points.

However, PPAP Handbook from AIAG version is much older (1994).

Accordig to Ford CSR (PPAP 4.0, Version 2013)

11 Initial Process Studies (General) (PPAP section
Demonstration of initial process capability (Calculation of Ppk) (for all Special Characteristics except Critical Characteristics):

The Ppk index of the initial process study must be calculated using a data set that is statistically stable, in control and normally distributed or of the expected distribution (see note 1 below). Stability, control and normality are determined using a data set of at least 25 subgroups and a subgroup size of at least 5 for a minimum of 125 measurements using rational sampling methods (see the AIAG SPC manual).

Anyway, I understand that it can be quite interesting in some specific processes, but for others an exception of the minimum numbers of parts could be applied, as both texts (AIAG and Ford Handbook) indicate

Thanks for the support!


Haste Makes Waste
I agree with Jim. I would explain this to my STA Engineer and see if he/she will change the requirements.


Stop X-bar/R Madness!!
Trusted Information Resource
Not too many people buy into the 30 part sample any longer. But be sure you measure the exact same place on each part, because you want the between part variation, without within part variation. The 5 pc sample assumes your process is normal. I would not. I would take the number of parts you are going to make, divide it by 125 and sample at that rate. E.g. if 250 parts, every other part. I would plot it in time ordered sequence to see that variation, and I would plug it into a curve fitting program, such as Distribution Analyzer at variation.com, and report best fit capability. Look closely at the p value - a p value of 1.00 means the data fits the distribution perfectly. You cannot expect Cpk or Ppk to be very good, because a new die should never have data near the center of the specification - or it will only have half of the life it should have.

I would expect the STA to have very limited knowledge of the proper application of capability to stamping.
Top Bottom