Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY
Edited by: B.H.V. Topping, J.M. Adam, F.J. Pallarés, R. Bru and M.L. Romero
Marshall Quotient and Resilient Modulus Predictions for Asphaltic Concrete Mixes using Neural Networks
Civil Engineering Department, University of Umm Al-Qura, Makkah, Saudi Arabia
M.H. Alawi, "Marshall Quotient and Resilient Modulus Predictions for Asphaltic Concrete Mixes using Neural Networks", in B.H.V. Topping, J.M. Adam, F.J. Pallarés, R. Bru, M.L. Romero, (Editors), "Proceedings of the Seventh International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 106, 2010. doi:10.4203/ccp.94.106
Keywords: resilient modulus, Marshall quotient, artificial neural networks, asphaltic concrete mixtures.
The road system in Saudi Arabia is one of the main transportation systems, which has developed in rapid manner. Presently, the road construction program is still under way in all regions of the Saudi Arabia. The main function of this program is to connect the cities, towns and villages as much as possible throughout the kingdom, Therefore, it is necessary to have roads with excellent pavements from a structural and functional point of view.
Some empirical formulas were developed for the determination of the optimum bitumen content (OBC) [1,2]. Reference  provides estimates of the optimum bitumen content from the aggregate gradations of asphaltic concrete mixtures in the Makkah area using neural networks (NNs). They concluded from the study that the OBCNN prediction was fairly close to the corresponding actual values of the optimum bitumen content with an average error of 1.80%. The correlation coefficients of 0.9802 and 0.9679 were obtained for the testing data for optimum bitumen content prediction. The results indicated that the OBCNN could predict the optimum bitumen content of asphaltic concrete mixtures with adequate accuracy required for practical design purposes.
Estimation of Marshall stability from optimum bitumen content and aggregate gradations of asphaltic concrete mixtures in the Makkah area using the neural network was studied by Alawi et al. . Checking errors were calculated using two training algorithms: the back propagation (BP) algorithm and the cascade correlation (CC) algorithm. They concluded that the predicted values were fairly close to the corresponding actual values of optimum bitumen content with an average error of 3.841%. The correlation coefficients of 0.9792 and 0.9749 were obtained for the training and test data for stability prediction. Also it is found that the training algorithm of CC yielded slightly more accurate forecasts compared to that of the BP algorithm .
The neural network method was used in this research for predicting the resilient modulus and the Marshall quotient for asphaltic concrete mixtures. To determine these properties using neural networks, firstly; samples were collected from different regions in the Makkah area during construction and secondly tested at Umm Al-Qura University laboratories for bitumen content, gradation of aggregate, bulk density, Marshall stability, Marshall quotient and resilient modulus determination. The neural networks method is applied to the data set to determine the Marshall quotient and resilient modulus at 60°C. The Marshall quotient and resilient modulus predictions were fairly close to the corresponding actual values. The results indicated that the bitumen content, compacted density and Marshall stability could predict the resilient modulus and Marshall quotient of asphaltic concrete mixtures using neural networks with adequate accuracy required for practical design purposes.
purchase the full-text of this paper (price £20)