Integrating 2K Factorial DOE and General Linear Model (GLM)

2K factorial DOE ( Design Of Experiment ) is a popular tools to either modeling your process or optimizing your process . Speaking about optimizing your process , a process owner of a process might ask you question like “ From your 2K result , can you tell me the percentage of contribution of each factor to the process output ? “ .
To answer this question , we cannot directly know the percentage of each factor contribution to the process output because in 2K result , it does not show the value of SEQ SUM SQUARE as shown in regression or General Full Factorial . Thus you need to integrate the result of 2K with GLM .

Let’s have a look at below case study :

Let say you executed 2K DOE in order to maximize your process output by working on 3 factors as below with 2 replicates ( total 16 runs ):

Factor               low (-1) level      High (+1) level
A                      195                   205
B                      Vtech                PolyAll
C                      Smooth             Ribbed

Your task is to find the optimum setting for each factor that can mazimize your process output and you also need to report the contribution of each term ( factor and interaction ) towards the process output .

Proceeding with 2K DOE analyze, you will get the final model as below :

Factorial Fit: Y versus A, B, C

Estimated Effects and Coefficients for Y (coded units)

Term       Effect    Coef  SE Coef      T      P
Constant           78.463   0.8050  97.47  0.000
A         -11.300  -5.650   0.8050  -7.02  0.000
B          -9.150  -4.575   0.8050  -5.68  0.000
C           3.350   1.675   0.8050   2.08  0.062
B*C        -3.525  -1.763   0.8050  -2.19  0.050

 

S = 3.22004     PRESS = 241.306
R-Sq = 89.18%   R-Sq(pred) = 77.11%   R-Sq(adj) = 85.25%

 

Analysis of Variance for Y (coded units)

Source                  DF     Seq SS    Adj SS   Adj MS      F      P
Main Effects            3     890.54    890.54  296.847  28.63  0.000
2-Way Interactions   1    49.70      49.70   49.703     4.79  0.050
Residual Error          11  114.06    114.06   10.369
Lack of Fit               3      12.25   12.25      4.082      0.32  0.810
Pure Error                8    101.81    101.81   12.726
Total                      15  1054.30

 

 

 

JUNE

Lean Six Sigma Black Belt 2

(15th June 2009 -19th June 2009)

LEAN LEADERS

(17th June 2009 - 18th June 2009)

MINITAB ADVANCE

(22h June 2009 - 25th June 2009)

JULY

SIGMA XL Pulau Pinang

(13th July 2009 - 14 July 2009)

Lean Six Sigma AWRAIENESS

(15th July 2009-15th July 2009)

Lean Kaizen Manufacturing

(20th July 2009 - 23th July 2009)