Home / DOE / Create Designs / Plackett-Burman Design
Plackett-Burman Design¶
QXL DOE New > Create Design > Screening > Create Plackett-Burman Design
A two-level design that holds up to one fewer factors than it has runs. It estimates main effects.
What this design gives you¶
A Plackett-Burman design gets main effects for a large number of factors in a small number of runs, and it does that by giving up the interactions: every interaction is aliased with main effects, so the design tells you which factors matter and not how they combine.
The run counts are not powers of two, which is the practical difference from a fractional factorial: a 12-run design holds up to 11 factors where the nearest fractional factorial would need 16 runs.
Choosing the design¶
The page is headed Select Plackett-Burman design and asks two things, Select number of factors: and Select number of runs:.
The available run counts are 12, 20, 24, 28, 32, 36, 40, 44 and 48, and only those larger than the number of factors you asked for are offered. A design with n runs holds up to n - 1 factors.
The n - 1 cap holds for every size, including the 40-run design, which takes at most 39 factors.
The factor list runs from 2 to 47, and the box accepts a typed value, so three refusals are reachable here: Number of factors must be an integer greater than 1 and less than 48., Number of runs must be an integer value. and, when the two do not fit together, Invalid number of runs for given number of factors.
Two sizes are built from stored tables rather than by cycling a seed column: the 28-run design, from three hard-coded blocks plus a final all-low row, and the 40-run design, from 39 stored columns. That affects how they are generated, not how many factors they hold.
Center points¶
Plackett-Burman designs get no center points. There is no center points page in the build sequence, and Modify Design > Add/Remove Center Points refuses afterwards with center points are not available for the Plackett-Burman design.
The page sequence¶
Select Plackett-Burman design > Enter Factor Names and Coding > Define outputs (responses) > Enter number of replicates and number of blocks.
See Also¶
- Taguchi Designs, the other screening family
- Two Level Factorial Designs, when interactions are needed
- Create D-Optimal, for a run count you choose yourself
- Define Replicates and Blocks
- Aliasing and Resolution, for how the alias table is built