Modelagem computacional e estimativa de cargas poluentes pontuais despejadas às margens de cursos fluviais

• Autor(a) principal: Lugão, Bruno Carlos
• Data de Publicação: 2022
• Outros Autores: [e-mail]
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UERJ
• Texto Completo: http://www.bdtd.uerj.br/handle/1/18591
• topic: Ipsimpy; Inverse problem; Pollutant loads; Finite differences; Problema inverso; Cargas poluentes; Diferenças finitas; Problemas inversos (Equações diferenciais); Poluentes; Diferenças finitas; CIENCIAS EXATAS E DA TERRA::MATEMATICA::MATEMATICA APLICADA::ANALISE NUMERICA

Economic progress and unplanned urban growth have long contributed to the de terioration of water resources, both quantitatively and qualitatively. In this context, the object of study of this work arises, which aims to identify and quantify the contribution of a point-source pollutant loads, through an inverse problem, released on the margins of fluvial courses without necessarily having knowledge of its functional form. The mathema tical modeling of this phenomenon was performed using the two-dimensional advection dispersion equation, assuming a parabolic velocity profile, solved by the finite difference method using the FTCS (Forward Time Centered Space) scheme. The formulation of the inverse problem involved both parameter estimation, as an optimization problem (scena rio A), and function estimation, as a statistical inference problem (scenario B). The first approach sought to recover constant loads, including the spill point, using the Differential Evolution (DE) algorithm. The second approach involved the identification of transient loads, considering different launch configurations, according to the Monte Carlo Markov Chain (MCMC) method. A computational package was also developed, implemented in Python language, called ipsimpy (Inverse Problem Simple Modeling), which presents it self as a tool capable of centralizing the inverse analysis, whose application can naturally extend to other areas. Aiming to bring the idealized cases closer to real situations, a preliminary calibration step was carried out, involving a real experiment, referring to the instantaneous release of a saline tracer, in order to characterize the dispersive and advec tive parameters of the transport equation. In scenario A simulations, comprising constant loads, the coefficients were properly estimated and the calculated concentrations showed good adherence to the synthetic data, with relative errors for the pollutant mass smaller than 2%. As for the estimation of transient loads, alluding to scenario B, the MCMC sa tisfactorily recovered the functions of interest, even the discontinuous ones, with relative errors of the L2 norm below 7%.