Glossary

Cp · potential process capability: Cp compares the width of the tolerance with the natural spread of the process (six standard deviations within subgroups). It ignores where the process is centred, so it shows what the process could achieve if it were perfectly centred.
Cpk · process capability index: Cpk measures how many three-sigma widths fit between the process mean and the nearest specification limit, using the short-term (within-subgroup) sigma. It accounts for both spread and centring.
Pp · process performance: Pp is like Cp but uses the overall standard deviation of all values, so it includes shifts and drift between subgroups. It describes the long-term performance the customer actually sees.
Ppk · process performance index: Ppk is like Cpk but uses the overall standard deviation. The AIAG PPAP manual uses Ppk for the initial process study with an acceptance level of 1.67.
Cpm · Taguchi capability index: Cpm penalises any distance from the target value T, not only distance from the limits. It is useful where being on target matters, for example for mating parts.
Cm and Cmk · machine capability: Machine capability is measured on a short, undisturbed run of usually 50 consecutive parts on one machine, for example at machine acceptance. Because the run excludes long-term influences, VDA volume 4 uses 8 and 4 standard deviations instead of 6 and 3, and typical requirements are Cmk ≥ 1.67 or 2.0.
Sigma within · short-term standard deviation: The standard deviation estimated from the variation inside rational subgroups. It represents the inherent, short-term noise of the process and is used for Cp, Cpk and the control limits.
Sigma overall · long-term standard deviation: The ordinary sample standard deviation of all values. It contains short-term noise plus all shifts and drift and is used for Pp and Ppk.
d₂ constant · range to sigma conversion: d₂ is the expected range of n values from a standard normal distribution: 1.128 for n = 2, 1.693 for n = 3, 2.059 for n = 4, 2.326 for n = 5. Dividing the average range by d₂ estimates the standard deviation. Related constants A₂, D₃ and D₄ give the control limits of X-bar and R charts.
PPM · parts per million defective: Defective parts per million. Observed PPM counts the values outside the limits in the sample. Expected PPM is calculated from the fitted distribution and estimates the long-run defect rate, including defects too rare to appear in the sample.
Sigma level · Six Sigma performance scale: The number of standard deviations between the process mean and the nearest limit. By Six Sigma convention a long-term shift of 1.5 sigma is assumed, so a 6 sigma process corresponds to 3.4 defects per million opportunities.
%GRR · gauge repeatability and reproducibility: The share of the observed variation that comes from the measurement system (repeatability plus reproducibility). AIAG: below 10 % acceptable, 10 to 30 % may be acceptable, above 30 % not acceptable. It can also be expressed as a percentage of the tolerance.
Repeatability and reproducibility · EV and AV in Gage R&R: Repeatability (equipment variation, EV) is the variation when one operator measures the same part several times with the same gauge. Reproducibility (appraiser variation, AV) is the variation between the averages of different operators. Together they form the Gage R&R.
ndc · number of distinct categories: How many groups of parts the measurement system can reliably tell apart within the part variation of the study. AIAG requires ndc ≥ 5. An ndc of 1 means the gauge cannot distinguish the parts at all.
Cg and Cgk · gauge capability (type 1 study): Indices from a type 1 gauge study, where one reference part is measured 25 to 50 times. Cg compares the gauge spread with 20 % of the tolerance, Cgk also includes the bias. Usual requirement: both ≥ 1.33.
Kappa · agreement beyond chance: Kappa measures agreement between ratings corrected for the agreement expected by chance. 1 is perfect agreement, 0 is chance level. Cohen's kappa compares two raters, Fleiss' kappa any number. In attribute MSA kappa ≥ 0.75 is good and < 0.40 poor.
RPN and action priority · FMEA risk ranking: The risk priority number multiplies the severity, occurrence and detection ratings (1 to 10 each). Because very different risks can have the same RPN, the AIAG-VDA FMEA handbook (2019) replaces it with an action priority table (High / Medium / Low) that gives severity the most weight.
AQL · acceptable quality limit: The worst tolerable process average for sampling inspection, in percent nonconforming. Together with the lot size and inspection level it gives a sampling plan: sample size n, acceptance number Ac and rejection number Re (ISO 2859-1, ANSI Z1.4).