Protocol - Design of Experiment (DoE)
DoE is a statistical method to analyze the interactions among experimental factors in order to identify their optimal combinations.
Category:
Quality methodologies
Last revision: Apr 11, 2016
Author(s):
R. A. Fisher
Contact
Name: Giovanna L. Liguori
Address: Institute of Genetics and Biophysics Via Pietro Castellino, 111 Naples 80131 ITALY
Email: giovanna.liguori@igb.cnr.it
Figure legend:
Steps
Description | Temperature | Time | Note |
---|---|---|---|
Output definition and standardization of its computation
|
|||
Selection of factors
|
|||
Determination of minimum and maximum levels for all factors
|
|||
Creation of a screening design using a full factorial design
|
Begin with a large number of potential factors or large ranges level for each factor. The Screening allows you to eliminate the ones with little effect on the response or fix ones whit very strong effect.
|
||
Experiment execution
|
|||
Model construction
|
|||
Residual analysis
|
|||
Factors and Interaction analysis
|
|||
Model refinement
|
|||
Comparison of full factor model and refined model
|
|||
Creation of a Modeling design to study center points
|
Once identified the most important factors in the Screening experiment modeling design can be used to obtain a model that can be used to identify factors settings that optimize the response.
|
||
Experiment execution
|
|||
Model construction: Fit a quadratic model that has a linear main effects
|
|||
Model refinement
|
|||
Comparison of full factor model and refined model
|
|||
Identify factor settings that optimize the response
|
|||
Model validation
|
Quality validation:
Yes
Validation info
The methodology has been used to produce results published in peer-reviewed articles
Related Protocols: