d e f es 
AICOS News
15.03.2020
Coronavirus: continuation of our activities
Independently of the coronavirus epidemic... 
more Linie
20.02.2020
AICOS Technologies at PCH Meetings 2020 in Lyon
Lyon - 25.-26.11.2020: International business... 
more Linie
05.01.2020
Our training schedule for spring / summer 2020 is on-line!
You want to deepen your knowledge in data... 
more Linie
Search
Go
Infos > Design of Experiments (DoE)

Who is afraid of statistical Design of Experiments (DoE)?

Reservations against statistical Design of Experiments (DoE)

When discussing with persons from R&D who critisize statistical Design of Experiments (DoE), they mention more or less the same points. These are addressed in the following.

Prejudice ...

 

... and the truth

Statistical Design of Experiments (DoE) requires too many experiments. „Trial-and-Error“ in general needs more. And the software cannot force you to continue a series of experiments if you are already content with the currently reached result and decide to stop there.
The process is varying too much for a statistical analysis.  In particular when there is much noise, statistics is needed in order to discover the signal!
It is too early for statistics, we first need to collect some data. „garbage in, garbage out“: no systematic approach = no reliable results. Moreover: who guarantees that you will find sufficient time "later" for your analysis..?
We only vary one factor at a time in order to avoid confusion.  In this way, you miss the optimum. This procedure is moreover dangerous: interactions cannot be detected.
Everything is much too complex.  This would be true in fact if one had to perform statistical Design of Experiments (DoE) by hand. Modern software tools such STAVEX, however, facilitate the employment of statistical Design of Experiments (DoE) and yield easily interpretable analysis reports as well as intuitive graphical representations.

But: statistical Design of Experiments (DoE) tries to replace my professional competence!

On the contrary: due to the increased efficiency you have more time to actually use your competence. Most effort is needed before performing the experiments.
More often than not, only the systematic approach of statistical Design of Experiments (DoE) leads to the discovery of effects, which allow a deeper insight into the underlying mechanisms. For instance, in a crystallisation optimsation, statistical Design of Experiments (DoE) revealed that the crystal structure sensitively depends on the factor settings. This was the reason for a surprising drop in the yield, as it lead to a clogging of the filters.

Any questions about statistical Design of Experiments (DoE)?

We are at your disposal! Please call us at +41 61 686 98 77, or use the contact form.

powered by Swisscom AG
c