I have a 20-factor split-plot test where 15 factors are ready and 5 are delayed. The 5 delayed factors all have existing nominal as one of the levels. We must start getting data, so how can I design test that forces 5 delayed factors to nominal. This seems like a very common real world situation, so there must be a standard approach.
I may not totally understand your question. Are you asking how to organize the runs so that the 5 delayed factors have settings that are "none" as a level? If so could you just set up the experiment as a hard to change and hard or very hard to change setting and see if that gives you the order which is more experimentally doable logistically. At the end of the day being the boss and getting started is reality one has to give in on randomness. That is why we verify the experiment.