---

Statistics are playing a greater and greater part in the improvement of the fabrication process. Statistic Process Control is a method of production control based on statistics analysis. It favours the development of self-control and helps towards guaranteeing an optimum level of quality in each stage of fabrication. _ The aim of SPC its to priviledge the consistant preventative measure of:
  Evaluating the aptitude of the process in relation to the specifications.
  Permanently analysing its performance in relation to a situation of reference.
  Intervening, not simply when faulty goods are produced, but as soon as a drift is detected, in relation to the aforementioned situation of reference.

SPC systems make it possible to react afterwards or in real time on the production resources as a whole, in full knowledge of the facts, and by following the impacts or effects of the applied instructions. An activity usually dedicated to the engineering department or to the research department, statistics systems are used in production, particularly in the automotive industry and in process industries.

Objectives:
  To maintain a process in a nominal situation and tolerance limit situation.
  To identify process variations to establish comparative rules of evaluaton.
  The follow-up of modifications to validate its actual improvement.

SPC tools all work in the form of sampling in accordance with laws of statistics. The results are presented in different graphics, highlighting the variability of the process studied.

PARETO diagrams are based on statistics properties of normal principles (GAUSS distribution) and are practical in the experimental principle named 80 - 20 (20% of breakdowns generate 80% of stoppages). The result is that a limited action (reacting on only 20% of the causes of breakdowns) can have a great financial impact (80% of stoppages are removed or reduced). To do this, SPC systems have tools which automatically supply statistics data and present the results in the appropriate graphics. The more classic representation is the histogram in percentage and of decreasing value; and the histogram with cumulative values. The control cards represented in graphics, deliver a user-friendly tool, making it possible to predict faults, as opposed to being subjected to them, and to react in time so as to avoid generating high costs due to poor quality.

The unavoiadable forms of the SPC tools are control cards, generally of 2 types:
  Control cards of a measurable magnitude (weight, temperature, size...) also named “measuring cards”
  Control cards of a non-measurable magnitude (scratch, impurity, aspect...) also named “attribute cards.”

The control cards of a measurable magnitude authorise the reception of the results of samples m of size n (the most common values are m=20 and n=5). The calculation rules make it possible to determine the higher and lower control limits. Mean cards (X/) and extended cards (R) are usually used. X/ and E cards are also used.

The capability measures the aptitude of the process, in order to produce parts or products which comply with specifications. The capability measurement is done by sampling methods, which can use control cards. SPC provides an easy means of measurement of the process capability and of the indicators Cp ((high tolerance - low tolerance) / 6 sigma), Cm and Pp for dispersion, and Cpk, Cmk and Ppk for centering.

SPC rules are valid only if the dispersion of the process follows a normal rule. It must be verified. The Henry slope and the Chi2 test, or other methods, enable this to be verified. SPC provides a control means of distribution with a formula enabling the implementation of different tests: Henry slope and the Chi2 test, Kolmogorov, Shapiro-Wilk etc.