通用电气绿带DOE读本.pptVIP

GBWorkshop_10_FractionalFactorial_Rel_1.1a ? 2001 General Electric GBWorkshop_10_FractionalFactorial_Rel_1.0a ? 2001 General Electric Fractional Factorial Designs Number of Runs Required... The factorial strategy is an efficient approach to experimentation as compared to “one at a time.” This can result in a large number of runs, even with a relatively small number of factors. For a 2 level factorial with k factors…..Number of run=2k. Information From 2-level Factorial Designs Example of Information Available... Reducing the Size of a Factorial Experiment Equipment delays allow you time to run only 4 trials in the allotted time. Which 4 trials will you choose? Choosing the Half Fraction Choosing the Half Fraction Properties of a Properly Selected Half Fraction The design is nicely balanced, that is, each factor is studied the same number of times at each level. (Equal number of + and -). The design collapses into a full factorial. Should any factor turn out not to matter, the result is a full factorial in the other two factors. The design covers much of the region of interest. Constructing a Half Fraction for Four Factors List the full factorial for three factors. This is called the base design. The fourth factor is assigned the values of the 3-factor interaction of the other three factors. Recall: a 3-factor interaction column is obtained by multiplying the three main effect columns. So D = ABC. General Rule for Constructing a Half Fraction Define the base design as a full factorial for the first k-1 factors. Assign the kth factor to the interaction of the first k-1 factors from the base design. This interaction is found by multiplying the factor level settings together for the first k-1 factors from the base design. Number of runs = N = 2k-1. Exercise Half fraction of a 25 factorial. Generate a half fraction experiment fork = 5 factors, A, B, C, D, and E. Answers Trade-offs A 2-factor Experiment With Confounded Effects Confoun