21.3.6.5. Random Constants
Now, let’s consider the usage of Random Constant. If one has experience for Taguchi’s robust design, he knows the noise factors that are uncontrollable factors in system or environments. As you can see, Taguchi’s robust design concept is to determine the design variables to minimize the variance with respect to those noise factors. Let’s consider the following design variable set.
DV(1) |
Deterministic |
Variable |
- |
- |
DV(2) |
Random |
Variable |
STD |
0.01 |
DV(3) |
Random |
Constant |
STD |
0.01 |
In this problem, the initial meta-models are constructed from the user defined DOE methods. Then, from the meta-model, the approximate variation is evaluated as
\(\tilde{\sigma }{{\left( \mathbf{x} \right)}^{2}}\approx \sum\limits_{i=2}^{3}{\left( {{\left| \frac{\partial \tilde{f}\left( \mathbf{x} \right)}{\partial {{x}_{i}}} \right|}^{2}}\delta _{i}^{2} \right)+...}\)
where \({{x}_{i}}\) and \({{\delta }_{i}}\) are the ith design variable and variation. Also \(\tilde{f}\) is meta-model. In conclusion, Random Constants are Noise Factors in Robust design and DFSS.