Resources

Q Charts for Process Variances

 /****************************************************************/
 /*          S A S   S A M P L E   L I B R A R Y                 */
 /*                                                              */
 /*    NAME: SHWQCHT4                                            */
 /*   TITLE: Q Charts for Process Variances                      */
 /* PRODUCT: QC                                                  */
 /*  SYSTEM: ALL                                                 */
 /*    KEYS: Shewhart Charts, Short Run Process Control, Q Chart,*/
 /*   PROCS: SHEWHART                                            */
 /*    DATA:                                                     */
 /*                                                              */
 /*     REF: Quesenberry, C. P. (1991). "SPC Q Charts For Start- */
 /*          Up Processes and Short or Long Runs"  Journal of    */
 /*          Quality Technology, Vol. 23, No. 3, pp. 213-224.    */
 /*                                                              */
 /*    NOTE: See Example 4, pg. 220 of above Reference.          */
 /*                                                              */
 /****************************************************************/
 options ps=60 ls=80 nodate;

 /*************************************************************/
 /*                                                           */
 /*  The following data is 20 samples of size 5. The first 12 */
 /*  samples have a mean of 10 and standard deviation of .01. */
 /*  The last 8 samples have the same mean, but the standard  */
 /*  deviation has been doubled to .02.                       */
 /*                                                           */
 /*  The first 2 plots assume the parameters are known, while */
 /*  the last 2 plots assume the parameters are unknown.      */
 /*  Comparing the plots shows how each reacts to an increase */
 /*  in standard deviation.                                   */
 /*                                                           */
 /*************************************************************/

data sample;
   seed = 2934531;
   do i = 1 to 20;
      subgrp = i;
      do j = 1 to 5;
         if ( i < 13 ) then do;
            x = 10 + .01*rannor(seed);
            end;
         else do;
            x = 10 + .02*rannor(seed);
            end;
         output;
         end;
      end;
   run;

 /****************************************/
 /*  Case I:                             */
 /*  Assume mu and sigma are known.      */
 /*  mu = 10, sigma = .01                */
 /****************************************/

 /****************************************/
 /*  This proc call calculates subgroup  */
 /*  means and standard deviations.      */
 /****************************************/

proc shewhart data=sample;
   xchart x*subgrp / nochart
                     outhistory = hist1
                     stddeviations;
run;

data qchart1a;
   set hist1;
   if ( _n_ < 13 ) then do;
      group  = 'Std of .01';
      gcolor = 'yellow';
      end;
   else do;
      group  = 'Std of .02';
      gcolor = 'red';
      end;
   q = sqrt(xn)*( xx - 10 ) / .01;
run;

symbol1 v=dot h=.5 c=white;

title 'Q-Chart for Process Mean (Parameters Known)';
proc shewhart data=qchart1a;
   irchart q*subgrp (group) /
                      mu0        = 0
                      sigma0     = 1
                      cframe     = gray
                      czones     = yellow
                      cinfill    = blue
                      llimits    = 1
                      xsymbol    = 'CL=0'
                      blockpos   = 2
                      cblockvar  = gcolor
                      ndecimal   = 0
                      nolimitslegend
                      novangle
                      nochart2;
   label q      = 'Standardized Value';
   label subgrp = 'Observation Number';
   run;

data qchart1b;
   set hist1;
   if ( _n_ < 13 ) then do;
      group  = 'Std of .01';
      gcolor = 'yellow';
      end;
   else do;
      group  = 'Std of .02';
      gcolor = 'red';
      end;
   q = probit( probchi( (xn - 1)*(xs**2) / .0001, xn - 1) );
run;

title 'Q-Chart for Process Variance (Parameters Known)';
proc shewhart data=qchart1b;
   irchart q*subgrp (group) /
                      mu0        = 0
                      sigma0     = 1
                      cframe     = gray
                      czones     = yellow
                      cinfill    = blue
                      llimits    = 1
                      xsymbol    = 'CL=0'
                      blockpos   = 2
                      cblockvar  = gcolor
                      ndecimal   = 0
                      nolimitslegend
                      novangle
                      nochart2;
   label q      = 'Standardized Value';
   label subgrp = 'Observation Number';
   run;

 /****************************************/
 /*  Case II:                            */
 /*  Assume mu and sigma are unknown.    */
 /****************************************/


data qchart2a;
   set hist1;
   retain lgmean lsum ls2sum;
   if ( _n_ < 13 ) then do;
      group  = 'Std of .01';
      gcolor = 'yellow';
      end;
   else do;
      group  = 'Std of .02';
      gcolor = 'red';
      end;
   if (_n_ = 1 ) then do;
      sum   = xx;
      gmean = xx;
      s2sum = xs**2;
      end;
   else do;
      sum   = lsum + xx;
      gmean = sum / _n_;
      s2sum = ls2sum + ( xn - 1 )*xs**2;
      s2pi  = s2sum / (( xn - 1)*_n_);
      wi    = sqrt( ((xn)*(_n_ - 1 )*xn )/(_n_*xn ) )
                           *(xx - lgmean)/sqrt(s2pi);
      q     = probit( probt(wi, (_n_-1 )*xn ) );
      output;
      end;
   lgmean = gmean;
   lsum   = sum;
   ls2sum = s2sum;
   run;

title 'Q-Chart for Process Mean (Parameters Unknown)';
proc shewhart data=qchart2a;
   irchart q*subgrp (group) /
                      mu0        = 0
                      sigma0     = 1
                      cframe     = gray
                      czones     = yellow
                      cinfill    = blue
                      llimits    = 1
                      xsymbol    = 'CL=0'
                      haxis      = 0 to 20 by 2
                      blockpos   = 2
                      cblockvar  = gcolor
                      ndecimal   = 0
                      nolimitslegend
                      novangle
                      nochart2;
   label q      = 'Standardized Value';
   label subgrp = 'Observation Number';
   run;

data qchart2b;
   set hist1;
   retain ls2sum;
   if ( _n_ < 13 ) then do;
      group  = 'Std of .01';
      gcolor = 'yellow';
      end;
   else do;
      group  = 'Std of .02';
      gcolor = 'red';
      end;
   if (_n_ = 1 ) then do;
      s2sum = (xn-1)*xs**2;
      end;
   else do;
      s2sum = ls2sum + (xn-1)*xs**2;
      wi    = ((xn*(_n_-1) - _n_ + 1)*xs**2) / s2sum;
      q     = probit( probf(wi,xn - 1, xn*(_n_ - 1) - _n_ + 1) );
      output;
      end;
   ls2sum = s2sum;
   run;

title 'Q-Chart for Process Variance (Parameters Unknown)';
proc shewhart data=qchart2b;
   irchart q*subgrp (group) /
                      mu0        = 0
                      sigma0     = 1
                      cframe     = gray
                      czones     = yellow
                      cinfill    = blue
                      llimits    = 1
                      xsymbol    = 'CL=0'
                      haxis      = 0 to 20 by 2
                      blockpos   = 2
                      cblockvar  = gcolor
                      ndecimal   = 0
                      nolimitslegend
                      novangle
                      nochart2;
   label q      = 'Standardized Value';
   label subgrp = 'Observation Number';
   run;

goptions reset=all;