(Received 29 August 2013; accepted 9 April 2014)
Published Online: 2014
| ||Format||Pages||Price|| |
|PDF Version||12||$25||  ADD TO CART|
This study presents artificial neural network (ANN) models to estimate the specific surface area of fine-grained soils as an alternative to sophisticated laboratory procedures. Geotechnical properties of 206 soils were measured experimentally based on ASTM standards. Soil input parameters used in database development were particle size at 10 %, 30 %, and 60 % finer, coefficient of curvature, coefficient of uniformity, percentage of silts and clays, percentage of soil passing sieve No. 200, fineness modulus, liquid limit, plastic limit, plasticity index, and activity. This data was used to train, test, and develop ANN models based on the backpropagation algorithm. Performance of ANN estimation was reliable when comparing the predictions with target outputs. Results indicated that the suggested ANN models exhibited excellent fit of the data as measured by the coefficient of determination and mean-square-error values. Thus, the developed ANN models could be used as a simple prediction tool to estimate soil-specific surface area reliably and efficiently as a rapid inexpensive substitute for cumbersome laboratory techniques.
Dept. of Civil Engineering, The Univ. of Jordan, Amman,
Dept. of Civil Engineering, Al-Isra Univ., Amman,
Stock #: GTJ20130146