What is Design of experiments?

The design of experiments (DoE) includes all statistical procedures that should be applied before the start of the experiment. These include: 

- Determination of the minimum number of experiments required to comply with precision specifications 

- The chosen Values for the different factors  

- Take into consideration all the other  

- using different designs, like Full factorial plans, fractional plans, and the response surface method 

- Sequential experimental planning and evaluation (sequential analysis); here data acquisition and evaluation alternate until a predetermined accuracy is reached 

Since experiments require resources (staff, time, equipment, etc. ), the person responsible for the experiment finds himself in a conflict between the accuracy and reliability of his expected results on the one hand, and the necessary effort on the other. Unlike one factor at a time, with the statistical design of experiments, the interaction between influencing factors (= independent variables) and target variables (= dependent variables) is determined as precisely as possible with as few experiments as. This by using different doe designs or design tools, like the full factorial plan, fractional factorial plan, central composite design, response surface method, and a lot of other experiments.  

An important part of the statistical design of the experiment is the determination of the experimental scope in relation to precision specifications such as the risks of statistical tests. Also to determine the levels of the factors to get precise results.  

Purpose and benefit of applying DOE

The intuitive procedures in experiments, such as changing one factor at a time (one factor at a time) or according to the principle of trial and error (trial and error), only produce an optimal test result by chance. The individual effects and interactions of influencing factors are not recognized. 

In contrast, the statistical design of experiments is a method for the systematic planning and statistical evaluation of experiments. The functional relationship between influencing parameters and the results is determined and mathematically described with little effort. The resources required for this, such as personnel, time, and costs, are known and quantifiable before the tests are carried out. 

By using the Randomization and Blocking methods, we will be able to reduce the effect of the other factors, that affect our experiments, but we can’t control them.

The Statistical methods and computer software 

To get a robust experimental design we use a lot of statical methods like analysis of variance, ANOVA, regression models or equations. These methods are precise and help us to understand the process that we examine and support the examiner with analyzing the data. 

To conduct and analyze the data, special software is available to make it easier for statistically less experienced users to carry out planning and evaluation, but this is done at the expense of flexibility. Suitable programs include Design-Expert, GlobalOptimize, Modde, and STAVEX; broader tools with special DoE modules include Cornerstone, JMP, Minitab, STATISTICA, or Visual-XSel, In addition, various simulation packages often contain specially tailored programs or modules for statistical design of experiments.