screening designs

**Screening Designs** are **experimental designs optimized for identifying the vital few significant factors from a large number of potential factors** — using a minimal number of runs to determine which of many candidate process variables actually affect the response, before investing in detailed optimization. **Key Screening Designs** - **Fractional Factorials**: $2^{k-p}$ designs that test $k$ factors in $2^{k-p}$ runs using aliases. - **Plackett-Burman**: Economical 2-level designs in $N = 4n$ runs for up to $N-1$ factors. - **Definitive Screening**: 3-level designs that can detect curvature and 2-factor interactions. - **Supersaturated**: More factors than runs — for initial rough screening only. **Why It Matters** - **Factor Reduction**: Screening reduces 20-50 candidate factors to the 4-8 that truly matter. - **Efficiency**: 12-run Plackett-Burman can screen 11 factors — far fewer than the 2048 runs for a full $2^{11}$ design. - **First Step**: Screening is the essential first stage of any systematic process optimization. **Screening Designs** are **finding the vital few from the trivial many** — efficiently identifying which process parameters truly drive quality from a large candidate list.