In Design of Experiments (DoE) how to choose the next sample to take?

Question

I have just recently completed the 1st iteration in my DoE where I had ~20 variables that generated 900 combinations. From those combinations, it simulated outputs that were higher than others.

I plan to do a second iteration of the cycle to validate my model and narrow the search space. Do I just take the top 10% highest value or do random sampling for my second iteration?

PS. It is an optimization problem so I want to achieve the highest output.

Thanks

Answer 1

For something like this you would/could use a fractional design e.g. via FrF2.