Fig. 3
From: Climbing the mountain: experimental design for the efficient optimization of stem cell bioprocessing
Sequential experimentation in RSM. A two-level factorial design can be used to initially assess the design space for three factors (x 1, x 2, x 3; green). A fractional-factorial design would be more appropriate for processes with many more variables, see text for details. If model predictions suggest that the optimum may be outside of this range, an expanded factorial design can then be run (blue). Once we are confident that the optimum is located within our design space, a more complex CCD experiment can be run in a smaller area of the design space to provide a more accurate model of the process that includes non-linear responses (pink). Finally, the optimum can be located (orange)