You may not need to perform all 27 experiments. In general, the optimum design is obtained when the condition number of the matrix is smallest. If the experiments take a while to perform (or are expensive) you can update and calculate the condition number of the test matrix with each experiment and may find that you have sufficient information before you reach 27 possibilities. This would be the case if the impact of any of the parameters happens to be nearly linear. I know you can calculate the determinant of the test matrix in Excel and you may be able to calculate an estimate of the condition number as well, depending on the structure, or you may be able to find an Add-In on-line for this calculation. See the following link for a description of the condition number: