# Incomplete treatments in Fractional Factorial Designs

I'm trying to analize a chemical experiment with four factors or independent variables. Initially each factor had three levels, so i proposed a $3^4$ factorial design; the problem was that all the treatments couldn't be measured, so i proposed a $3^{(4-1)}$ fractional factorial design, so i had $3^3 = 27$ treatments.

Later i had this problem where the treatments when the fourth factor is in its "high" level couldn't be measured, so i have now of the $27$ possible treatments just $18$. How could i proceed to analize this treatments?