Contribuição da atualização bayesiana na estimativa de recalques: estudo de caso de um aterro sanitário

Abstract

Predicting and monitoring settlements in urban solid waste landfills is one of the most difficult challenges in this type of structure. This item is provided for in NBR 13.896/97, which recommends the essential stages of designing non-hazardous waste landfills. Many studies and research have been developed throughout recent history to estimate settlements in sanitary landfills, based on theoretical and empirical methods, considering several different factors. Many of these methods reported in the literature are based on empirical procedures associated with experimental data adjustments. These experimental data, in most cases, are obtained in field monitoring. However, the heterogeneity of this type of massif hinders the wide application of the developed formulations. It often makes studying the subject an arduous task, with complex variables to be considered. The present work intends to contribute to the study of the compressibility of sanitary massifs, by updating the predicted settlement of a Waste Treatment Center, called Unit 01, in Jaboatão dos Guararapes/PE, through the application of estimation models of existing settlements, combined with data from the monitoring of the real settlement that occurred in the unit, associated with concepts and theories of statistics. For the development of this study, we chose to adopt the model considered classic and most used to approach the topic, the Sowers Model, which has its basis in the adaptation of Terzaghi's Classical Theory (1973) for soils, and the Yen and Scanlon, which uses the concept of average landfill age. To study the compressibility that occurred in solid waste, data from monthly geotechnical monitoring of 30 surface landmarks with well-defined settlements and with known local waste disposal history were used, in intervals of 720 days ranging from 2019 to 2023. Statistics, based on Bayes' Theory (1812), which allows calculating the probability of an event "A" considering that an event "B" occurred, were used to estimate the posterior updates of the settlement of the surface landmarks used in the region, adopting as a function of likelihood the measured data, and as a priori data the results provided by models. Following the methodology initially proposed by Lacasse (1991) applied to piles, this study provides an application focused on the analysis of settlements in massive solid waste, opening a series of possibilities for this challenging topic. The concept of the Failure Indicator from Guttormsen (1987) evaluates the quality of the updating. The application indicated the reduction of uncertainties and allowed the comparison of two a priori models.