Aplicação de técnicas para estimação de parâmetros e discriminação de modelos em processos de adsorção

• Autor(a) principal: Gonçalves, Cliff Iuri de Souza
• Data de Publicação: 2017
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFS
• Texto Completo: http://ri.ufs.br/jspui/handle/riufs/17084
• topic: Adsorção; Estimação de parâmetros; Discriminação de modelos; Planejamento experimental; Adsorption; Parameter estimation; Model discrimination; Experimental planning; ENGENHARIAS::ENGENHARIA QUIMICA

Adsorption is a surface phenomenon that depends on the interfacial tension between the fluid and the pores of the adsorbent. The determination of the adsorption isotherms is a fundamental step in understanding the behavior of the process as well as designing new optimized and improved operating conditions. This work presents a literature review on the estimation of parameters in adsorption processes, showing that many authors perform linearization of functions that should be treated as non-linear, as well as the use of variables that are inference and are treated as exit variables, inducing serious problems to discriminate models. In addition, the use of the equilibrium constant 'K' to determine the thermodynamic parameters as enthalpy and entropy of the system is quite recurrent in the literature. The most consistent approach is to describe this temperature effect in a single model. Therefore, the objective is to make use of parameters estimation procedures and experiment planning techniques to discriminate models in adsorption processes. Thus, experiments were conducted in stirred tank to determine the isotherms of acetic acid in activated carbon at different temperatures. Concentrations of 8, 15 and 22 g / L were used, with temperatures of 35, 45 and 55 °C, with all points in triplicate. The HPLC analytical methodology was used to determine acetic acid concentrations. Langmuir (L), Langmuir Freundlich (LF), Freundlich (F) and Langmuir Tóth (LT) were selected among the vast number of models, because they had smaller amounts of parameters and behavior of the isotherms closer to the experimental data. It was also performed the estimation of parameters and discrimination of the isotherms models at each individual temperature and taking into account the effect of the temperature in the isotherm model. The results showed that it is possible to discriminate models using few experiments, with materials of low laboratorial cost to obtain a single adsorption isotherm model, whether with temperature dependence or not. This work showed the importance of the proper use of parameter estimation tools and model discrimination in the adsorption scenario, as well as the selection of models that represent the experimental data in the best possible way. We also compared the magnitude of the systematic errors that are found in the estimation process when using the linearization of the models or the use of correct evaluation techniques of input and output data. The importance of this work is given in the creation of a rationale that can be applied in other more costly processes and in industrial scale, reducing their operational costs.