Theoretical Analysis and Development of an Artificial Neural Network Model to Evaluate Earthen Dam Slope Stability

Abstract

In the design of earth dams, it must be considered thatthe water leakage through the earth dam generatesupward and pore pressure, in addition to leakageforces that cause internal erosion, which has a directinfluence on the structural stability of this system.Also, the rising and dropping in the water level has adirect effect on the stability of the dam's face slope.One way to solve these issues is the installation of acore or a horizontal water drainage system. Thepresent study relied on the GEO-Studio computer toolto evaluate cross-sectional models of earthen dams bydetermining the safety factor under differentsituations represented by a change in filter type, andthe flow state as a result of raising and lowering thewater level at the dam reservoir and the full fillcondition of the dam reservoir. The research foundthat the existence of a core substantially contributedto improving the safety coefficient for the case of risingthe water level (2m) and rapidly rising by assigning itthe greatest safety coefficient values. The absence of afilter had an opposite influence on the safetycoefficient by decreasing it. Also, the factor of safetyfor the downstream slope was affected by less than 5%for different flow conditions, compared with thehigher effect generated by the upstream slope.Furthermore, an artificial neural network model withan accuracy ratio of more than 97% was developed forthe predicted safety factor.

Keywords

ANN, Core, Earth dam