Hybridizing multivariate robust regression analyses with growth forecast in evaluation of shear strength of zeolite–alkali activated sands

A comprehensive experimental program was undertaken to investigate the relationships between B-ratio, void ratio, and principal stress difference at failure of zeolite–alkali activated sands. For this purpose, a series of unconfined compressive strength tests (UCS) was carried out on specimens under different arrangements of zeolite–lime content to fixate an optimum lime content. From interpretation of UCS test results, undrained triaxial shearing tests were performed on samples comprising optimum lime content and various zeolite percentages. With regards to the experimental evidence, a novel time dependent growth forecast was performed to extend the curing ages beyond the conditions of the experimental program. Then, a series of multivariate regression analyses including robust analysis along with error diagnoses estimations was utilized to define new relationships among influential variables. High accuracy and confidence rate of the proposed unified relationships presented by this research will provide a solid framework for the treatment of fine sands with natural zeolite–lime blends.