| Abstrakt: |
Partial replacement of conventional stabilizers with environmentally-friendly and sustainable materials has recently turned out to be a common approach to enhance the strength and stiffness properties of earthen materials. In this regard, natural zeolite and expanded polystyrene (EPS) beads are used in this study to be added to clay samples conventionally stabilized with lime. Accordingly, a systematic survey is conducted to inspect the changes in mechanical strength and shear stiffness of untreated and lime-zeolite treated reconstituted clays (with various proportions of kaolinite and montmorillonite) due to the incorporation of EPS beads. To this end, first, the optimum percentage of lime was obtained from a series of pH tests. Then, the optimum content of lime replacement with zeolite yielding the highest unconfined compressive strength (UCS) was estimated which appeared to be 25%. Afterward, all the samples were lightened by adding 0.1% and 0.25% weight ratios of EPS beads to the fine materials (η ) and the results were compared with similar samples without EPS beads. Several standard Proctor compaction, swelling, UCS, and non-destructive bender element (BE) tests were carried out on lightened samples to assess their physical properties, strength characteristics, shear wave velocity ( V s ) and small-strain shear modulus ( G max ). The samples' swelling and compaction properties decreased by incorporating the EPS beads. Strength and stiffness characteristics of lightened composites were observed to be acceptable when 0.1 % EPS beads (denoted as the optimum percentage) were included in the studied mixtures. Furthermore, to interpret microscale phenomena, scanning electron microscopy (SEM) micrographs of the interfaces of the typical untreated and treated samples are provided, which demonstrate the interlocking zones, formation of cementitious gels and the considerable volume of the clay medium occupied by deformable EPS beads leading to the significant reduction in the soil weight. Finally, using the group method of data handling (GMDH)-type neural network (NN), several statistical modelings are carried out to predict the UCS, V s , and G max of the studied mix designs, which resulted in highly accurate models. [ABSTRACT FROM AUTHOR] |
