1 - 2 of 2
ItemDeveloping prediction models for soil profile and parameters using Artificial Neural Network(University of Khartoum, 2016-03-22)The objective of this research is to propose a unified neural network model to predict the type of soil layer and to estimate soil parameters in specified locations in Khartoum state. The study used 76% of the soil investigation data previously available (data of 82 sites) as input data to predict the remained data (data of 26 sites) what was never seen before from the network. The source data of coordinates and elevations was collected from aerial photographs where was the soil data was obtained from geotechnical projects performed in Building & Road Research Institute (BRRI) of University of Khartoum. After the neural network has been successfully trained, its performance was tested on a separate verifying set. The results indicate neural network is a useful technique for prediction of type of soil layer and parameters. In addition, previous studies were well enhanced to develop the proposed unified model; this is because the previous studies did not take into account the validation process of the models performance during the training operation. The proposed unified method gave a success rate of 55% to 86% for the validation of its ability to predict the soil profile. For evaluating the ability of the models to predict the soil parameters, the correlation coefficient was used as an indicator of the relation between the measured and predicted values. It gave values in range between 0.4 and 0.5 for most models except for the shear model which produced 0.05. This small value is back to the large variation in the input data of the shear model. Finally, it can be concluded that Artificial Neural Network (ANN) has the ability to predict the soil classification and soil parameters in the Khartoum with an acceptable degree of accuracy and can be developed further to produce a geotechnical map for Sudan.
ItemGeotechnical Properties of Clayey Sand (SC) soils(UOFK, 2015-05-14)Montmorilloinic clayey sand (SC) soils tend to have relatively high values of L.L & P.I and noticeable variations of volume change as their moisture content changes. The expansive clay in (SC) soil tends to increase with the increase of its colloidal properties. In this research, it is intended to examine the behaviour of an artificial and natural soils made by combining different ratios of sand with clay of known characteristics. The number of soil types tested in this program can give good idea about the geotechnical properties of (SC) soil. The effect of increasing sand fraction on soil was determined at different types of clay minerals. The use of artificial and natural soil samples is intended to clarify the effect of sand fraction on the behaviour of sand-clay mixture. Increasing sand fraction content will affect the volume change characteristics whereas it changes the properties of expansive clay and then changes the whole soil structure. Series of index tests, compaction, consolidation and swelling tests were performed to understand the geotechnical properties of (SC) soil. Package SPSSWIN was used for the regression analysis of laboratory tests data. The test results were analyzed and presented. The research concluded that artificial soils with 5% bentonite content and not more than 15% are classified as having low swelling potential and recommended for landfill liners of low hydraulic conductivity. Increasing the sand fraction from 7% to 10% will give low volume change characteristics and high strength properties for (SC) soils for using under foundations or fill materials.