Because the elastometric connector experiments described in Section 12.3 required physical experimentation, Song and Lawson (1988) suggested us-

ing a single array design to save on the number of experiments required.

Table 12.21 shows a resolution IV 27−2 fractional factorial design and the resulting pull-o force. The factor names and levels are the same as those shown in Table 12.3 of Section 12.3. The generators for the design were F = ABC and G = ABD.

(a) What is the dening relation for this design, and how many of the 12 control-by-noise factor interactions can be estimated clear of other main eects or two-factor interactions?

(b) Using the FrF2 package, can you nd a 32-run resolution IV design that has more control-by-noise factor interactions clear? If so, what are the generators?

(c) Calculate a set of 31 saturated eects for this design, and make a normal plot to determine which eects, interactions, and confounded strings of two-factor interactions appear to be signicant. Is there a clear interpre-

tation of any confounded strings of interactions?

(d) Make interaction and contour plots of any two-factor interactions be-

tween control factors and noise factors and use them to choose the level of control factors that will minimize the eect of noise factors.

(e) Are there any adjustment factors? If so, what levels should be chosen to maximize the pull-o force?