This study investigates seven parameters for a two-step process for remediating Sucarnoochee soil by agglomeration with petroleum coke. Fine agglomerates produced in the second stage are recycled to the first stage of the next agglomeration experiment. This produces large agglomerates and gives better cleaning than the one-step process described in 2. Sucarnoochee soil, artificially contaminated to its saturation level of 15 weight percent oil, was used. The experiment was statistically designed. Initially experiments were done according to a 2(7-3) fractional factorial design with 1 replicate and 4 center points to screen out the unimportant interactions. This design was augmented with 10 additional axial runs and a response surface model was fit to the data. A nonlinear optimization of the model was done using Lingo(R) software to obtain the optimum levels for the seven process variables. This sequential strategy of experimentation allowed us to model the process and find the optimum levels for the factors in only 30 runs. A complete 3(k) factorial experiment would require considering 2187 treatment combinations. Validation runs at the optimal levels for the process variables gave 170 ppm oil remaining on the soil which was significantly lower than the previous best result of 810 ppm. References: 4
展开▼