Page 595 - Design for Six Sigma a Roadmap for Product Development

Design Optimization: Advanced Taguchi Robust Parameter Design 551

(a) High variation (b) Low variation
Low S/N
High S/N

(c) Nonlinearity (d) Linearity
Low S/N
High S/N

(e) Low sensitivity (f) High sensitivity
Low S/N High S/N
Figure 15.10 S/N in various cases.

15.3.3 Two-step optimization procedure
and examples
Taguchi dynamic robust parameter design proposes the following
preparation and two-step optimization procedure:
Step 0 (preparation). Using the layout described in Table 15.1, run
the experiment and collect data. For each inner-array run, compute
the sensitivity β and S/N.
Step 1. Run a complete Taguchi DOE data analysis; using dynamic
S/N as response, find the control factor combination to maximize S/N.
Step 2. Run another Taguchi DOE data analysis; using sensitivity β
as response, find sensitivity adjustment factor(s), which do(es) not
affect S/N, and use it (them) to tune β to the target value.
We will use the following examples to illustrate this two-step opti-
mization procedure.