Program for complex numerical and graphical capability assessment of production processes.

The CAPA software program offers complex numerical and graphical assessment for production processes possessing a variety of distributional properties. It's many features include the ability to evaluate multivariate quality characteristics, job shop production quality assessment, calculation of process robustness and expected percentage of non-conformable products, the control of inputted data, and the validity of evaluation of the chosen index with special tests.  CAPA's core functionality is it's ability to present a thorough and reliable capability evaluation of processes regardless of distributional characteristics following the norms of QS 9000 and VDA 6.1.  Hence great attention is paid to the verification of prerequisites of the data. 

What the CAPA program provides   Calculation of all main capability indexes in all possible situations: Quality characteristics with a normal distribution: Cp,CpK, Cpm,C*pm,CpmK. Quality characteristics with non-normal distribution: estimation of indexes Cp,CpK for any type of distribution, confidence interval of CpK for any type of distribution, estimation of C'p,C'pK,C'pm,C'pmk for a set of chosen distributions. job shop environment (original method)!!! At the same time, it calculates the expected percentage of non-conformable products (symbol NC used) and robustness (symbol R used) of the process.   Data testing outliers will provide numerical and graphical verification of x(1) and x(n) values whether it is extremely high or low. verification of normality with the use of graphs or by three numerical tests.   Capability assessment for multivariate quality characteristics: multivariate indexes MCp and MCpm capability vector for complex process assessment graphical capability assessment for a multivariate quality characteristic.   Capability assessment with Taguchi loss function which expresses the      poor quality financially   Evaluation of the calculated indexes with special tests.   Calculation of basic characteristics. Top of Page

Graphical Methods in CAPA G1: Histogram, Gauss curve, average(the middle yellow vertical line), (xbar-3s, xbar+3s) (the left and right yellow vertical line), tolerances LSL, USL (the left and right vertical full red line), target value T (the middle vertical full red line). G2: Q-Q graph: Quantiles of distribution N(0,1) on the horizontal axis, order statistics x(i) on the vertical axis.  Tolerances: LSL=horizontal bottom red line, USL=horizontal top red line.  If the points lie between the LSL and USL and "close to" the line, then the data are in the tolerance and have normal distribution. G3: Multivariate characteristic of capability White rectangle=tolerance area, i.e. an area of the prescribed location of quality characteristics.  Ellipse with red points = area of the quality characteristics real location.  The yellow cross passes through the means of the quality characteristics (the center of the ellipse), the white cross passes through the target values of the quality characteristics and defining the center point (T1,...,Tk). G4: Outliers (Grubbs' test graphically) Vertical yellow lines = borders whose crossing/exceeding signals that the value (which crossed it) is an outlier (and thus a reason to be excluded).  A random number is assigned to each x as its y-coordinate.  The data is then spread around the x-axis to prevent merging when large amounts of data are to be evaluated. G5: Castagliola's method It shows the point of an empirical distribution function and its color approximation with polygons of different degrees.  The found empirical distribution function is used in the numeric section of the program for an estimation of the index CpK for any type of distribution. Top of Page

Analytical and Graphical Methods in the CAPA Program   Univariate PCI Normal distribution Symmetric tolerance (G1) Non-Symmetric tolerance (G1) One side tolerance Non-normal distribution methods Clement's method Castagliola's method (G5) Confidence interval for CpK (Bootstrap method) Job shop environment   Test of Index Test of Cpm Test of C*pm Cpm(min) calculation   Multivariate PCI Vector Cpm, Cp Capability vector Centralization of processes (The Hotelling's test) Outliers and multivariate Cp (G3) Multivariate Cpm Test of multinormality   Loss function (Taguchi) N tolerance symmetrical N tolerance unsymmetrical S tolerance L tolerance   Tests Test of normality (skewness, kurtosis) The Anderson-Darling's test of normality (G2) The Shapiro-Wilk's test of normality (G2) The Grubbs's test of outliers (G4) Stability of Process Independence of Observations   Numerical characteristics Top of Page