What is Design of Experiments and why use it?

• Are you running experiments involving two or more factors?

• Do you have complex pieces of equipment requiring set-up by adjusting two or more factors?

• Are you using Design of Experiments (DoE) for these tasks?

If the answer to the last question is NO, then you are almost certainly wasting precious resources, and possibly not even obtaining valid results.

DoE is the most efficient way possible to run multi-factor experiments or set up complex equipment, can be easily learned by any engineer or scientist and is the essential tool for designing and optimizing complex products and processes.

How does DoE differ?

The majority of Engineers and Scientists working today use the one-factor-at-a-time approach to experimentation.  This means that factor A is varied while factors B, C, D, …etc are held fixed at their nominal values.  Next factor B is varied while factors A, C, D, …etc are held fixed at their nominal values.  The problem with this approach is that it does not allow you to identify interactions between factors.  This can lead to a choice of factor settings which are far from optimal and a complete misunderstanding of your process or product.

DoE provides you with a means to build a more complete picture of your process or product by allowing you to identify situations where factors interact, helping ensure that you obtain a more accurate model of the process or product.  What’s more, the additional information will be gained at a lower cost in resources than the traditional one-factor-at-a-time approach.   

Taguchi

The Taguchi approach to designing experiments has become popular in certain industrial sectors (e.g. automotive).  Taguchi's experimental designs have some associated problems with regard to efficiency, although Matrex does include a large range of his design arrays (Orthogonal Arrays).