Prediction of Undrained Shear Strength from Index Properties of Soils Found in Bonga Tow

Abstract

For many foundation Engineering problems, such as the carrying capacity of shallow foundations and piles, lateral Earth pressure on retaining walls, and the stability of dam and embankment slopes, the shear strength of the soils is a crucial element. Due to a lack of resources and limited time to get soil engineering parameters, geotechnical engineers frequently work to create statistical models that are tailored to a certain place and soil type, especially for analysis and design purposes. The study aims to predict undrained shear strength from the Index Properties of soils found in Bonga town by statistical analysis using statically packaged software for social science (SPSS V.27). The methodology used for this thesis included field identification, sampling, laboratory tests, preliminary analysis of data using scatterplot, develop model, and evaluation of model adequacy by checking the assumption and hypothesis test for regression coef icient. The design of the study is an experimental study that requires laboratory tests conducted on 35 representative samples by purposive sampling technique. Using the results obtained from laboratory tests, linear, quadratic, exponential, power, and cubic regression were performed, and the fit model was selected based on the coef icient of determination, model significance, and fulfillment of the assumption of regression. For all test procedures, the American Society for Testing & Material (ASTM) standard was used. For the study area the laboratory test result of natural moisture content (NMC), liquid limit(LL), plastic limit(PL), unit weight of soil, and specific gravity value range from 34.04 % to 52.34 %, 56.5-92.7%, 30.92-50.79%, 15.94-19.01, 2.64 to 2.80 respectively. The undrained shear strength of the study area ranges from 48.58kPa to 106.2kPa. Correlation and regression analysis shows that undrained shear strength was significantly correlated with natural moisture content, liquid limit, Plasticity index, and dry unit weight, and not significantly correlated with specific gravity. From the study, the best Model is obtained from multiple linear regression (MLR) analysis by logarithmic transformation and given by: 𝑙� 𝐶 = 0.600209+0.092788Ɣd+0.003034PL; coef icient of determination (R 2) =0.923, P-value <.001, Tolerance=0.847 and Variance inflation factors (VIF) =1.18, Durbin Watson=1.555 with 96.2% of accuracy. The validation of the model was confirmed using a control sample from the study area as well as from another area. Using the developed model, the undrained shear strength parameter can be figured as well and it is expected to have wide application in the construction to minimize the cost, ef ort, and time for laboratory tests of the undrained shear strength of the study area. In light of the study findings researcher recommends it is better to perform a database preparation of laboratory test results from all parts of Ethiopian towns and perform a predictive model for engineering properties of soil using advanced predictive modeling technique