Experimental part of the ruggedness test

The experiments are performed according to the chosen design and a response or a number of responses are measured. The sequence in which the experiments are performed can influence the estimation of the effect of a factor [36]. The reason for this lies in the fact that the measurements can be influenced by different sources of error. Each measurement is influenced by uncontrolled factors that cause random error. Measurements can also be influenced by systematic errors or by systematic errors caused by drift (linear drift due to time-dependent factors). The occurrence of systematic errors or of drift will affect the estimation of the effects of the factors fi om the design [36]. 

If only random errors would occur the experiments could be performed in any order. Performing the experiments in a randomized way allows to make more correct estimations of the effects when there are also systematic errors. These errors are due to uncontrolled factors which vary between sets of experiments [36]. An example of this type of error could be the factor room temperature . Suppose that the factor room temperature is not controlled or examined during the performance of a design and that it takes more than one day to execute the complete design. It is then possible that a different room temperature during the different days introduces systematic differences between the sets of experiments carried out on those days. 

Especially the full and fractional factorial designs are best performed in a random order to avoid the influence of systematic errors because they are constructed so that one factor is at one level in the first N/2 experiments and at the other in the last N/2 experiments. The Plackett-Burman designs can be considered as randomized when they are performed in the sequence that is described in the original papers of Plackett and Burman. 

Sorting the experiments for organizing reasons or experimental constraints as is proposed in some publications [6,17,19] should preferably be avoided. In some experimental designs it is however too difficult and time-consuming to perform the design in a random order. This is for instance the case when one of the factors is the column manufacturer or batch number of the column . In this case two columns are used and a performance of the design in random order is not really indicated. Experiments performed with the different columns must be grouped. The experiments with each column still can be executed in random order. 

If systematic errors due to drift are expected then one can perform the design in a well defined randomized way so that the calculated main effects are not biased by the drift [36]. These designs are called anti-drift designs and they are described for full and fractional factorial designs. However, the interactions effects calculated from these designs are still biased by the drift. 

---

Carrying out the experiments described in the design. This is the experimental part of the ruggedness test ... 

---