Abstract

[This abstract is based on the author's abstract.] "Low-cost response surface" methods were applied to derive the optimal engineering design of a plastic fastener with restricted experimental budget and timing requirements. These methods were used because only enough resources were available for half the test runs of standard response surface methods based on central composite or Box-Behnken designs. Low-cost methods provide comparable expected modeling errors to alternative methods under realistic assumptions. The choice of low-cost methods for the application is discussed. The methods are compared with alternatives such as "small response surface designs" and D-optimal designs from several standpoints. It is concluded that the low-cost methods should offer an attractive alternative to standard response surface methods because engineers would generally want to make the cost versus accuracy tradeoff. The snap-fit fasteners that were designed using the low-cost response surface methods hold more than twice the load of the formerly used fasteners with a modest increase in size. A major automotive manufacturer is using the new design to fasten heating, ventilating, and cooling cases.