Hi All,
I have a question regarding the following example to demonstrate why DOE is
a good idea versus the change one variable at a time approach used by many
engineers.
Consider an experiment with factors X1 and X2 and a response Y. In a vary
one factor at a time approach I might use the following cross design (Y
comes out of the page)
X1
*
* * *
*
X2
Where I fix X1 and make measurements at 3 different values of X2, and then
fix X2 and make an additional 2 measurements at different values of X1.
The DOE approach is standard 2^2 factorial design with a centre point
X1
* *
*
* *
X2
My question is: Are these 2 designs not identical one is just a rotation of
the other by 45 degrees?
So my vary one factor at a time example is not a good one?
Best Regards,
Richard.
|