Integrated ANN model for earthfill dams seepage analysis: Sattarkhan Dam in Iran

Development and extension of the civilizations need optimum management of water resources. In this way, construction of dams to control, store and transfer water is one of the oldest and most important activities of engineering, which is nowadays regarded as one of the biggest and costliest projects of the Civil Engineering. In arid and semi-arid countries, due to low amount of rainfall, and rapid increase in water demand, it is necessary to control and optimize the available water resources. Therefore, dam construction on the path of surface flows is one of the basic choices to reach such an optimized goal.

One of the traits of earthfill dams is related to their cheap body material, which may be found in the construction site. Because of this, earthfill dams in comparison to other types of dams are more economical. In addition of economic aspects, in some cases constructing limitations make building of the earthfill dam necessary. In the past, design of the earthfill dams was completely based on the empirical knowledge and faith in the sections which had good performance, but in the last decades the behavior of dams, especially the collapsed dams, has been studied using modern soil mechanics. The most important improvement in this case is the analysis of seepage through the body of an earthfill dam and its influence on the stability of the dam. Terzaghi’s one-dimensional consolidation theory for saturated soils introduced an approach to seepage problems [1].

Over the past few decades, significant improvement has been achieved in the issue of seepage modeling through the dams; for this purpose several methods have been suggested to model seepage analysis. Models based on their involvement of physical characteristics generally fall into three main categories: black box models, conceptual models and physical based models [2]. The conceptual and physical based models are the main tools for predicting variables and understanding the physical processes involved in a system. However, they have a number of practical limitations, including the need for large amounts of field data, sophisticated programs for calibration using rigorous optimization techniques, and a detailed understanding of the underlying physical process [3]. If sufficient data are not available, and accurate predictions are more important than understanding the actual physics of the situation, black box models remain a good alternative method and can provide useful predictions without the costly calibration time [4].

Recently Artificial Neural Network (ANN) as a black box model has been widely used for forecasting in many areas of hydraulic, hydrology and water resources. Also, a presented literature survey by Shahin et al. [5] reveals that ANN has been successfully applied to several geotechnical engineering problems. Goh [6] presented an ANN model to predict the friction capacity of piles in clays. Ni et al. [7] proposed a methodology of combining fuzzy sets theory with ANN for evaluating the stability of slopes. A number of hypothetical natural slopes were evaluated by both ANN and an analytical model, and the results of the ANN model were in a good agreement when compared with the analytical model. Sivakugan et al. [8] explored the possibility of application of neural networks to predict the settlement of shallow foundation on the granular soils. Similar studies have been conducted in different areas including soil properties and behavior, liquefaction, design of tunnels and underground opening, soil permeability and hydraulic conductivity, soil swelling and classification of soils by applying ANNs [5].

For full text: click here

(Author: Vahid Nourani, Elnaz Sharghi, Mohammad Hossein Aminfar

Published by Sciedu Press)