Abstract Central composite design, a type of response surface methodology (RSM), was combined with Derringer and Suich desirability function approach (DFA) in this study to optimize multi-additives for expansive soil improvement. The additives, which ranged from 2 to 8 for Ordinary Portland cement (CM) and 0–20 for both sawdust ash (SDA) and quarry dust (QD), were all added by air-dried weight of the expansive soil. The soil properties evaluated as the responses include the unconfined compressive strength (UCS), California bearing ratio (CBR) and differential free swell (DFS). Second order polynomial regression models were developed for prediction of the responses and validated statistically using analysis of variance. The derived models had respective R2 values of 0.865, 0.872 and 0.951 for UCS, CBR and DFS, which show that the models possess good predictive capability. Combined effects of the additives on the various responses were assessed using both surface and contour plots. Confirmatory tests and scanning electron microscopy analysis performed on the optimum combinations of additives confirmed clearly that the RSM can be employed to optimize additives for expansive soil modification. Furthermore, an optimum combination of additives (5 CM, 20 SDA and 20 QD) that optimizes the three responses concurrently using the Derringer and Suich DFA, had a composite desirability of 0.8367 which is reasonable in comparison with the ideal composite desirability of 1.0000. This is an indication that the responses were optimized favorably.
展开▼