Department of Civil Engineering
Permanent URI for this collection
Browse
Browsing Department of Civil Engineering by Author "AKIŞ, Ebru"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
Item UTILIZATION OF POWDERED GLASS AND EXPANDED POLYSTYRENE BEADS (EPS) AS ADDITIVES IN CLAYEY SOILS(2022-06-14) ÇİĞDEM, Öykü Yağmur; AKIŞ, EbruWaste management is gaining importance since the impact of climate change on human life becomes more apparent. In this study, it is aimed to investigate the effect of waste materials on the improvement of high plasticity clay. Glass powder and expanded polystyrene beads (EPS) are selected as waste materials since both additives have the lowest recycling rate (4.43% and 4.47%, respectively) among solid wastes. The effect of these materials is evaluated both individually and together with various experiments such as the Atterberg limit test, standard proctor test, one-dimensional swell, and unconfined compressive strength tests. Additive percentages are set at 0.3%, 0.9%, and 2% of the dry weight of the soil sample for EPS and 2%, 4%, and 6% of the dry weight of the soil for glass powder. Test results show that the addition of waste glass powder causes an increase in unconfined compressive strength and a decrease in free swell values. When EPS is used as the only additive, a decrease in the unconfined compressive strength and free swell values is observed. Furthermore, the combination of 4% of glass powder and 0.9% of EPS addition causes the most effective improvement in both unconfined compressive strength and free swell values. In addition to the experimental studies, the obtained data from this study, and the test results of related studies in the literature are used together to build up datasets. These datasets are used to perform multilinear regression (MLR) and artificial neural network (ANN) analyses. As a result of MLR analyses, prediction equations are proposed to estimate the free swell or unconfined compressive strength of the improved soils depending on the additive contents, free swell, or unconfined compressive strength of untreated soils. When the same dataset is used for ANN analysis, similar results with the MLR analyses are observed. Furthermore, if the provided data is limited to the Atterberg limits only, the ANN method provides reliable predictions of those parameters. A 5-fold cross variation is performed to verify that the results of ANN analysis do not depend on the selection of the data sets. Both MLR and ANN methods show that it is possible to predict the free swell and unconfined compressive strength of treated soil samples with high correlation coefficients.