You can look into center point replicated , or un replicated, 3^2 factor design, and Central composite Design, their analysis and associated properties, in this regard.

Dear Ahmed Kira,

If you have two factors with three levels, and you just want to assess the effects of each one on your experience, you can perform your experience with a three levels full factorial design containing 9 experimental runs with different combinations of both factors. But, if you need to optimize your experience and find the best suting of your two factors, it is more convenient to perform a central composite design of response surface methodology, but at this case you need to run more than 9 expériences. Best regards

2 factors at 3 levels? I would do a full factorial with n = 3 depending on how expensive and complicated the experiment is. Time and cost usually determine whether I choose a full factorial design or something smaller.

Ahmed Y. Kira

you can use CCD or BBD in RSM (DX7).

10-12 test.