Fluxo de água e gás em camada de cobertura final de solo compactado em aterro sanitário

Abstract

The cover layer applied to Municipal Solid Waste (MSW), in landfills, generally made of compacted soil, as it is in direct contact with the external environment, presents changes in its physical, chemical and biological conditions. These variations can occur due to the exposure of the soil to natural environmental agents that cause the layer to lose its tightness over time, in terms of simultaneous flows of water and air, thus prevailing the unsaturated condition of the soil in semi-arid regions. As a result, excess flows can harm the useful life of the landfill, the safety of its operation and negatively affect public health. The existing ways to monitor these parameters only use the measurement of the water permeability coefficient, based on analytical approaches, numerous experimental and numerical analyses, which are insufficient to predict the behavior of the covering layer in the face of water and air flows. Therefore, it is necessary to use predictive models, for example, Artificial Neural Networks (ANN), which are versatile tools that have the capacity to make decisions and make them available for use and application in other landfills. Thus, the objective of this work is to propose predictive models using ANN to predict soil permeability coefficients for water and air in the cover layer of a landfill in a semi-arid region. As a methodology, an experimental design was used to carry out tests in the laboratory and in the field, to verify the conditions of the layer and soil used. Through statistical analyses, the soil parameters that influenced water and gas flows through the cover layer were defined. Subsequently, by developing multiple linear regression models and numerical modeling using neural networks, based on a synthetic database, the topology and architecture of the ANN were defined. The results indicated that the physical variation of the soil used in the cover layer influences gas emissions and water infiltration. The compaction efficiency parameters have less variability, indicating uniformity in the energy applied, however, they do not reflect the homogeneity of the soil in the layer. The optimum compaction moisture value is similar to the air entry point of the characteristic curve, and at this point the unsaturated water and air permeabilities are similar. The soil parameters that most influence unsaturated soil permeability to water and air are the degree of saturation, volumetric moisture and gravimetric moidture. The developed ANN showed good performance for predicting soil water and air permeability coefficients. Given this, it can be concluded that soil parameters that are related to the water content present in the soil have greater influences on flows. The execution and monitoring of the cover layer must be regulated by standards that include the environmental and physical-chemical parameters of the soil. The developed ANN can be applied to analyze flows in landfills with similar characteristics.