THESIS SUBMITTED TO THE DEPARTMENT OF CIVIL ENGINEERING ARBA MINCH UNIVERSITY INSTITUTE OF TECHNOLOGY, SCHOOL OF GRADUATE STUDIES ARBA MINCH UNIVERSITY

Abstract

Geotechnical characteristics of the subgrade soils can affect the road quality and serviceability in the construction of pavement, since all soils are not used as a subgrade material. The existing has several failurities that weaken the road performance to carry the loads and to give comfortable services for the people of the area due to its performance is affected by rutting‟s, pothole. The aim of this paper is correlating CBR with index properties and characterizing subgrade soil of the road by linear and multi-linear regression analyses. To achieve the objective of this research, ten pits were selected from different representative parts of the existing road at 2km interval and twenty soil sample were collected, two from each pit at the depth of 1.2m to 1.5m. The laboratory test results presented that the NMC ranges from 21.09% – 29.85%, Percentage finer (silt & clay) ranges from 86.55%– 97.46%), LL ranges from 47.76% – 55.06%, PL ranges from 29.92%-39.55%, PI ranges from 9.98%-19.90%, Gs ranges from 2.56%– 2.70, Gravel ranges from 0.10%– 1.82%, Sand ranges from 2.26%– 13.13%, Clay ranges from 51.22%– 69.21%, Silt ranges from 25.69%– 42.69%, OMC ranges from 21.0%-33.00%, MDD ranges from 1.41g/cm3 - 1.56 g/cm3, CBR ranges from 3.28%-12.79%. Based on these results the study area was characterized as clay soils, and for the soils classification under A-7-5 according to AASHTO soil classification system. From all soil samples the subgrade strength class is characterized based on CBR result i.e., 3.28%-12.79%, it shows highly compressible soil and its plastic limit is very high. Most of the soils of the study area fell under A-7-5, and which indicate that the soils are highly clay according to the AASHTO classification. The developed correlation entailed a moderate determination coefficient of Model 21 with R2=0.9 using single regression analysis, indicates that model equation for CBR, have better strength of association, with CBR values. That model shows the best strength of association among all others based on relationship of predicted values and data of various soil properties. The models for CBR are the function of eight independent variables; i.e., CBR =fn (LL, PL, P200, OMC, MDD,) with recommended equations: CBR = -0.0635(LL3) + 9.6179(LL2) - 486.17(LL) + 8206.7. But the simplicity of utilization the best model for CBR is the fn (LL).i.e. .