Synthace aims to provide scientists with an easier way to get started on using DOE in their experiments without feeling immediately overwhelmed. In practice, this translates to limiting the number of DOE design options that you can create.
Following the 80-20 rule, Synthace provides 20% of the design functionality and this is sufficient to cover 80% of use cases encountered in the lab. Alas, this trades off against the ability to make other types of designs that are needed in particular cases, which remains the domain of dedicated statistical software.
Here are a few instances where you will need to use a statistical package to create your DOE design and import it into Synthace:
Larger or more complex optimal designs
Synthace’s optimal designs have the following limitations and characteristics:
“Main Effects only”: up to 20 factors with 2 levels each. D-optimal criterion.
“Up to Two-Factor Interactions": up to 12 factors with 2 levels each, or 9 factors with 3 levels. D-optimal criterion.
“Up to Quadratic Effects”: up to 8 factors with 3 levels each. I-optimal criterion.
Here are a few cases in which you would need to use 3rd party software for creating an optimal design:
More factors than listed above for the given model type
A specific model not included in the above three choices
A Bayesian-optimal design (in other words, one which doesn’t require all model parameters to be estimable)
More runs in the design than required for a full-factorial in the specified factors (controls and replicates are additional to this and have no restrictions)
A lot of factors with more than 2 levels - there is a limit on the time it takes to create a design which may be hit even for smaller numbers of factors if those factors have very large numbers of levels
Optimality criteria which don’t fit into the three cases above
Large or complex split-plot designs, split-split plot designs, split plots in which you want a specific number of whole plots
Any type of space filling design other than latin hypercube
There are plenty of other choices - uniform, maximum entropy, minimum potential etc. which you may want to use, however for most scientific purposes the differences are not important.
Any conventional design type other than a full factorial, such as:
Fractional factorial designs
Taguchi designs
RSM designs such as Box-Behnken, central composite designs etc. Synthace supports optimization using quadratic optimal designs.
To learn how to export your factors from Synthace and import designs from third party tools back into Synthace, click here.