Analytical Quality by Design: A Tool for Regulatory Flexibility and Robust Analytics
Table 5
Selection of DOE tools in analytical quality by design.
Design
Number of variables and usage
Advantage
Disadvantage
Full factorial design
Optimization/2–5 variables
Identifying the main and interaction effect without any confounding
Experimental runs increase with increase in number of variables
Fractional factorial design or Taguchi methods
Optimization/and screening variables
Requiring lower number of experimental runs
Resolving confounding effects of interactions is a difficult job
Plackett-Burman method
Screening/or identifying vital few factors from large number of variables
Requiring very few runs for large number of variables
It does not reveal interaction effect
Pseudo-Monte Carlo sampling (pseudorandom sampling) method
Quantitative risk analysis/optimization
Behavior and changes to the model can be investigated with great ease and speed. This is preferred where exact calculation is possible
For nonconvex design spaces, this method of sampling can be more difficult to employ. Random numbers that can be produced from a random number generating algorithm
Full factorial design
Optimization/2–5 variables
Identifying the main and interaction effect without any confounding
Experimental runs increase with increase in number of variables