Response Surface Designs: Part-1 Addition of Centerpoint

[@vigneshdinagaran]

Nice explanation... 👍

[@instituteofqualityandrelia7902]

Thank you!

[@shawkathussain3421]

Hi, great video as always. I have a question: in your previous videos you showed how to calculate effect size for a 2 level DoE design. How would that translate if it is a 3 level, to work out average effect?

[@instituteofqualityandrelia7902]

The primary reason for three levels is to validate curvature. This can be better done by addiing center point as the third level. Ans if you want to model curvature properly, response surface designs are better option. As there are three levels, main effects of each factor will have two degrees of freedom (effect of level 1 to 2 and level 2 to 3). Montgomery metions that each main effect can be represented by a linear and quadratic component, each with one degree of freedom. So it becomes little complicated. Refer to Design of Experiments book by D.Montgomery for details. Apologise for the late response.

[@divyang089]

Explain why cetner point per block 3?

[@instituteofqualityandrelia7902]

Basically center points are used to estimate standard deviation. So three is the sample size suggested by the statisticians. Increasing will be good but will mean more budget and time.