It is often necessary to group the runs of an experiment into blocks. Runs within a block of runs are more homogeneous than runs in different blocks. For example, the experiment described in Goos (2002), describes a pastry dough mixing experiment that took several days to run. It is likely that random day-to-day differences in environmental variables have some effect on all the runs performed on a given day. Random block designs are useful in situations like this, where there is a non-reproducible shock to the system between each block of runs. In Goos (2002), the purpose of the experiment was to understand how certain properties of the dough depend on three factors: feed flow rate, initial moisture content, and rotational screw speed. It was only possible to conduct four runs a day. Because day-to-day variation was likely, it was important to group the runs so that this variation would not compromise the information about the three factors. Thus, blocking the runs into groups of four was necessary. Each day's experimentation was one block. The factor, Day, is an example of a random block factor.
To create a random block, use the custom design and enter responses and factors, and define your model as usual. In the Design Generation panel, check the Group runs into random blocks of size check box and enter the number of runs you want in each block. When you select or enter the sample size, the number of runs specified are assigned to the blocks.