Design of Experiments (DoE) Designer

Design of Experiments (DoE) is a systematic approach for investigating how multiple factors affect a process response. Instead of varying one factor at a time (OFAT), DoE varies several factors simultaneously in a structured way, which is more efficient, more statistically powerful, and reveals interactions between factors that OFAT can never detect.

When to Use Each Design

• 2-Level Full Factorial: Characterization of 2–5 factors when you want every main effect and every interaction.
• 3-Level Full Factorial: Smallest designs (2–3 factors) where you want a fully saturated quadratic model.
• Plackett-Burman: Start here when you have many factors (up to 23) and don't know which ones matter. Highly efficient screening in 8–24 runs.
• Box-Behnken: Response surface / optimization for 3–5 factors. Avoids extreme corner points (e.g., max pH × max conductivity), which is often desirable in bioprocessing.
• Central Composite (Face-Centered): Response surface design for 2–8 factors. Combines factorial and axial points to fit a full quadratic model while staying within the specified factor ranges.
• Central Composite (Rotatable): Same structure as face-centered, but axial points extend beyond the factor range to give uniform prediction variance at all points equidistant from the center. Best for 2–6 factors.

Practical Tips

• Always include center points (typically 3) to estimate pure experimental error and detect curvature.
• Randomize run order to protect against time-based confounding (reagent degradation, column aging, operator drift).
• Block by day when experiments span multiple days or buffer batches. This removes inter-day variability from your factor effects.
• Set realistic factor ranges. Don't choose ranges wider than operationally feasible or you'll waste runs in regions you'd never use.