Affordable Access

Publisher Website

Chapter 7 Other designs for screening experiments

Elsevier Science & Technology
DOI: 10.1016/s0922-3487(05)80011-5
  • Design


Publisher Summary This chapter discusses two design principles for the construction of screening experiments. Designs for screening experiments can be obtained from two-level factorial designs. In such designs, the number of runs will be a power of two, 4, 8, 16, 32,…, 2k. There are, however, situations where this may be inconvenient. An example of this would be running a screening experiment with 17 variables. The smallest fractional factorial design, which can accommodate that many variables would be a 217-12 design with 32 runs. In a screening experiment with that many variables, it would be sufficient to have the estimates of the average result and the main effects—that is, to approximate the response surface by a plane. For this, 18 experiments is the minimum. A fractional design with 32 runs would, therefore, contain a number of unnecessary runs, which are not used for the purpose of the experiment. This would be wasteful, especially if the individual runs are time-consuming or expensive. The chapter discusses two principles for the construction of screening designs: design by a Hadamard matrix and D-optimal design. Through such designs, many inconveniences can be overcome.

There are no comments yet on this publication. Be the first to share your thoughts.