Discussion of “Optimization of carpet waste fibers and steel slag particles to reinforce expansive soil using response surface methodology” by M. Shahbazi, M. Rowshanzamir, S.M. Abtahi, S.M. Hejazi [Appl. Clay Sci., doi:10.1016/j.clay.2016.11.027]

Abstract Shahbazi et al. (2017) adopted the response surface methodology (RSM) technique to evaluate the combined efficiency of slag treatment and fiber reinforcement in the stabilization of an expansive soil mixture. The swell percent ( S p ), swelling pressure ( P s ), and unconfined compressive strength ( q u ) were represented as a function of three independent variables (i.e. slag content, fiber content, and fiber aspect ratio) by the second-order polynomial regression model, and ANOVA analysis was adopted to allocate a contribution percentage to each component of the regression model. It was noticed that the reported fitting parameters for S p and P s provided ambiguous predictions, and thus were recalculated to avoid any misleading information for readers. The statistical technique proposed by the authors is limited to allocating contribution factors to regressional components rather than evaluating the net favorable impact of each independent variable on the desired dependent variable. Therefore, the partial derivative sensitivity analysis approach was proposed as a simple and practical alternative to address the aforementioned uncertainty. The new approach presented in this discussion not only provides a unique contribution factor for each independent variable but also accounts for the combined favorable contribution offered by all three independent variables.