STATCON - Software for Design of Experiments

Developed in the 1930ties Design of Experiments (DoE) was originally developed to optimize agricultural processes. Based on that it evolved to the standard method for product- and process-optimization in the whole industry. This is true for pharmaceutical applications, developments in automotive and especially for semiconductor industry. DoE can even be applied for psychological experiments or the optimization of direct marketing campaigns.

Central idea of DoE is to precisely plan which experiments are done - before the analysis begins. That way one can make sure to do just the right experiments to solve a specific problem. An important benefit of DoE is the ability to plan the right number of experiments: You will not do to many and not do too few.

Classical Design of Experiments

Classical Design of Experiments is still the most common way to apply DoE. Classical DoEs are clearly structured and easy to interpret. For the most real-world applications a classical Design of Experiments will be the adequate solution.

Classical Design of Experiments is a foundation pillar of more general paradigms like Six-Sigma and Quality-by-Design.

Optimal or Modern Design of Experiments

For some problems classical Design of Experiments is just not flexible enough. The doctrine of Optimal Design of Experiments follows the main principle to adjust the DoE to the problem and not the problem to the DoE. For this purpose - partially complex - mathematical optimality criteria (most often: D- and I-Optimality) are used to provide plans with more flexibility.

Optimal DoEs grant solutions for problems where classical Design of Experiments does not help any more.

Design of Experiments for Mixtures

A special application of DoE are mixture experiments. That are applications where each individual experiments implies merging multiple components. This kind of experiment has some implications for the setup of a design and for the way how it is analyzed.

There are special variants of Classical and Optimal DoEs, developed to address just this topic.

Split-Plot Designs

A very important special application in DoE are so called Split-Plot-Designs. These are addressing problems with factors that are hard to change. In Classical and Optimal DoEs it might happen, that some factors should be varied fairly often. E.g. for temperatures this could be a problem due to economic reasons.

Split-Plot-designs are a solution to this problem, as the allow you to research the effect of hard-to-change factors without varying them too often.