Hi All,

Could someone explain the distinct difference/strengths/weaknesses when talking about DOE: Full Factorial & Fractional Factorial Experiments & Response Surface Experiment.....appreciate if explained with examples.....Cheers

 

Kiran Varri

 

Hi Kiran,

Though your query looks quite simple, it is difficult to explain with out the understanding of certain basics of DOE. It would be easy if you had an oppurtunity to conduct an experiment and had the experience of DOE. It is important to understand the follwoing

a) Confounding or Alias

Resolution

c) Two, three, four way interactions

d) Blocking

e) centerpoints Etc

In a broad sense Factorial designs are applied mostly study mutiple factors at a same time and typical relationship is linear while Response surface is used when you suspect curvature

Generally less factorial means certain treatment is missed out and the equation may not be accurate. Various methods are available to mitigate based on the cost of experiment and other constraints

You can reach me at 9702777754. I was part of Motorola for 7 years

Srinivas

MBB

 

Hello Kiran,

Lets understand theory of each (of the above) one by one. Too many questions and responses may lead to some confusion.

1. Full Factorial: runs all possible combinations and develops a complete model

Vs

Fractional Factorial: rather than all possible combinations it runs only a subset of combinations and thus producing a fractional model.

Can you now guess what could be the adv/ disadv of these 2 models wrt time/ resources required and deriving conclusions with some degree of confidence based on the models produced thereby??