Split-Plot Designs |
This example revisits the tablet production scenario in Chapter 3, "Full Factorial Designs," with an alternate analysis suggested by Fossceco (1999). The tablet production process has several stages, which include batch mixing, where ingredients are combined and mixed with water, and pellet production, where the batch is processed into pellets that are compressed to form tablets. It is more convenient to mix batches and randomize the treatments within each batch than to mix a new batch for each run. Thus, this experiment calls for a standard two-stage split-plot design.
The moisture content and mixing speed for the batch constitute whole-plot factors, while the factors that control the variety of ways that pellets can be produced from a single batch are subplot factors. The responses of interest are measured on the final tablets. Table 12.1 shows all the variables involved and the stage with which they are associated.
Table 12.1: Factors and Responses in the Tablet Formulation ExperimentVariable | Name | Low | High | Description |
Level | Level | |||
Whole Plot | FLUID | 90 | 115 | Moisture content (%) |
MIX | 15 | 45 | Mixing time (min) | |
Split Plot | EXTRUDER | 36 | 132 | Extruder speed (rpm) |
SCREEN | 0.6 | 1.0 | Screen size (mm) | |
RESID | 2 | 5 | Residence time (min) | |
DISK | 450 | 900 | Disk speed (rpm) | |
Response | MPS | Mean particle size (micron) |
Copyright © 2008 by SAS Institute Inc., Cary, NC, USA. All rights reserved.