Modelagem Cinética e Estimação de Parâmetros na Remoção de Poluentes por Crescimento Microbiano
Modelagem Cinética e Estimação de Parâmetros na Remoção de Poluentes por Crescimento Microbiano
-
DOI: https://doi.org/10.22533/at.ed.892112621011
-
Palavras-chave: modelagem matemática; estimação bayesiana; Monte Carlo via Cadeias de Markov; análise de sensibilidade; remoção de poluentes.
-
Keywords: Mathematical modeling; Bayesian estimation; Monte Carlo via Markov Chains; sensitivity analysis; pollutant removal.
-
Abstract: This chapter presents the mathematical modeling and Bayesian parameter estimation applied to a pollutant removal process based on microbial growth, using the model proposed by Mohiuddin et al. (2022). The system is described by a set of ordinary differential equations representing the dynamics of microbial biomass and the consumption of carbon, nitrogen, and phosphorus substrates. The forward problem was numerically solved in MATLAB® using the ode15s solver, allowing a coherent qualitative reproduction of the experimental behavior of the system variables. Sensitivity analysis indicated that the parameter associated with the maximum microbial growth rate (µMax) and the carbon half-saturation constant (KC) exerts the greatest influence on biomass growth and carbon consumption, while the remaining parameters exhibited reduced sensitivity under the analyzed conditions. A linear dependence between µMax and KC preventing their simultaneous estimation. Based on these results, Bayesian estimation of µMax was performed using the Markov Chain Monte Carlo method with the Metropolis–Hastings algorithm, resulting in an estimated value of µMax = 0,5879 h-1. The use of the estimated parameter led to an good agreement between experimental data and simulated curves, particularly for biomass and carbon, with coefficients of determination of R2 = 0,9080 and R2 = 0,9446, respectively. For nitrogen and phosphorus, the obtained R2 values were 0,8513 and 0,4420, respectively, reflecting poorer adjustment for those dynamics. The results demonstrate the effectiveness of the Bayesian approach for estimating relevant kinetic parameters and for quantifying the uncertainties associated with the proposed model.
- Matheus Bastos do Carmo
- João Victor Tovany Soares da Silva
- Waldecléia Queiroz da Costa
- Miguel Fernando Saraiva Maia
- Ana Paula Souza de Sousa
- Haianny Beatriz Saraiva Lima
- Diego Cardoso Estumano