Methane is one of the primary greenhouse gases, which negatively affects climate change. Itspresence in the atmosphere has increased dramatically over the last century and continues to rise due tohuman activities. Oxidation of methane in landfill bio-covers takes place through the use ofmethanotrophic bacteria which provides a sink for methane. The rate at which methane is biologicallyoxidized depends on several parameters. This study aims to better understand the oxidation of methanein landfill soil covers through modeling methane oxidation with artificial neural networks. Modeling wasdone through laboratory batch reactor experiments, under three levels of saturation limits, three levels oftemperature, for three types of soils, with and without nutrients added. An artificial neural network wasdesigned and served as a numerical prediction tool to describe the expected methane oxidationefficiencies under different levels of applications of moisture content, under different temperatures andnutrient additions. The 3 layered feed forward artificial neural network was trained as a multilayerperceptron with 5 hidden nodes and a root mean squared error of 1.515 and an R2 of 98.6%. The modelwas verified with an error R2 of 98.3% and a root mean squared error of 1.521. Furthermore, the modeldemonstrated that it verifies closely with the experimental results reported by other researchers. Thus, themodel described methane oxidation efficiencies with high competence.
展开▼
