Utilizing computational neural networks for evaluating the permeability of compacted clay liners

The design of lined waste-storage facilities is significantly influenced by the permeability of the liner. The permeability of compacted clay liners, in turn, is influenced by factors such as clay type and composition, compaction type and effort, and operating conditions. The complexity of the permeation process makes it difficult to predict analytically the permeability from these factors. As a result, empirical regression models are frequently used to predict permeability. In this paper, permeability prediction models are developed using computational neural networks (CNNS). The developed CNN models are used to predict the permeability of compacted clay for a known set of soil properties and field and laboratory conditions. Moreover, the models are used to determine the relative importance of the various input parameters to the model output. Also, a comparison between regression models and neural networks for predicting permeability is presented and the advantages of utilizing CNN methodology over regression techniques in model development are highlighted.