What data do you need?
Define a measurable response, controllable factors, feasible factor levels and constraints. Plan randomization, replication and blocking where appropriate before running the experiment.
Worked example
With temperature and pressure each set at two levels, a full factorial design has four factor combinations. Two replicates give eight runs before adding any center points. An interaction means the effect of temperature changes with the pressure setting.
How to use this Six Sigma tool
- Define factors, levels and responses in the DOE workflow.
- Choose a feasible design and collect responses in the planned run order.
- Review main effects and interactions and confirm selected settings.
How to interpret the results
Separate experimental evidence from prediction. Review residuals, uncertainty and practical effect sizes, then confirm promising settings with a follow-up run. A model optimum outside the tested region requires further evidence.
Why use DOE instead of changing one factor at a time?
DOE can estimate multiple factor effects and interactions in a planned set of runs. Changing one factor at a time can miss interactions and does not provide the same information about combined settings.