Search

Efficient DOE – Reducing Runs 25-50%

Industrial applications for optimization or engineering Design of Experiments (DOE) can be perceived as expensive relative to the anticipated outcomes. Minitab and other similar statistical analysis programs that prepare DOEs are often not efficient because the number of degrees of freedom for error is large. This can lead to a significant number of unnecessary runs. United Technical uses Resolution V experiments with the ability to estimate all main effects and all two-factor interactions — while generally requiring significantly fewer runs.


Others claim to be smart enough to omit some interactions in order to make them fit into a standard design (computer-generated), but we take a different approach. We make the experiment efficient from the start after finding out what the customer actually needs. This customization allows United Technical to produce very thorough — but efficient —experimental plans.


We use Hadamard matrices to construct the DOE, this differs from the Yates matrix that is easy to program by computer algorithms. While this may add an hour or two to the program, this additional time is easily recovered from the dramatic reduction in experimental runs — typically a 25% reduction, but sometimes 50% fewer runs, depending on the investigation and size of the DOE. Generally, the larger the DOE, the greater the reduction in runs.


Additionally, United Technical uses a robustness plot approach to present the data. This provides graphical predictions from reduced statistically-based models of the data. Along with the robustness plots, the models are interpreted using the other physical evidence that was generated and combined with the theory behind the process to provide a phenomenological interpretation of results.

These differences in DOE approach makes the result much more science-based and less a purely math-based outcome while also being explicitly customer goal oriented. Contact us today to inquire about your DOE needs.

6 views0 comments

Recent Posts

See All