Simplified methods for the design of landfill double composite liners using neural network

Double composite liners (DCLs) have been widely used in landfills to protect the surrounding environment. This study aims to develop simplified empirical equations for calculating breakthrough times of DCLs based on analytical equations or experimental data. An artificial intelligence neural network called group method of data handling (GMDH)-type neural network was used to perform equation simplification. New empirical equations in polynomial formats were obtained by a layer-summation method and a series of numerical experiments based on analytical solutions for contaminant transport in double composite liners. The accuracy of empirical equations is demonstrated by comparing them with the existing solutions and numerical results. The performance of four types of DCLs were then investigated. The mean absolute percentage errors (MAPEs) for each type of DCL with different leachate heads and soil liner thicknesses were all lower than 10%. Additionally, a trend for the improvement of the GMDH equation accuracy with the increase of head loss of the upper composite liner (Δh₁) was observed. The presented equations can perform well in high leachate head conditions (e.g. >5 m) where DCLs are required.