Design & Analysis of Taguchi Robust Design Experiments - Design Experiment Tab

Select the Design experiment tab of the Design & Analysis of Taguchi Robust Design Experiments dialog to select from a variety of standard Taguchi robust designs. The logic of using orthogonal arrays to conduct experiments is explained in the Introductory Overview.

Arrays

Select an orthogonal array in the box provided. Note that the number of runs, maximum number of factors, and number of factors with 2, 3, 4, and 5 levels is indicated for each possible array. The Experimental Design module generates the most common arrays used in robust design experiments. In the unlikely case that your design does not "fit" any of the arrays listed here (note that you can analyze designs with up to 65 factors), refer to Taguchi (1987) or Phadke (1989) for a listing and description of additional orthogonal arrays. Also note that the standard 2(k-p), Plackett-Burman, 3(k-p), and Box-Behnken designs can be used to conduct robust design experiments. The arrays generated by the Experimental Design module are displayed in this dialog.

Choosing an array

When choosing an array, it is not necessary that your actual design include exactly as many factors as are supported by the respective array. You may simply leave some columns of the array unused. As long as all experimental runs are performed, the resulting design is still balanced, and all columns of the array are orthogonal to each other.

Estimating interactions, creating new factors

In some arrays, some two-way interactions are not confounded and can be estimated. Also, in some cases one can combine columns in the array to create factors with more levels. These techniques are described in detail in, for example, Phadke, 1989.