Made available in DSpace on 2014-06-11T19:22:35Z (GMT). No. of bitstreams: 0 Previous issue date: 2008-02-29Bitstream added on 2014-06-13T18:49:38Z : No. of bitstreams: 1 silva_ds_me_ilha.pdf: 714350 bytes, checksum: f3f8101e373024cd4f64ef7d8c6923d8 (MD5) / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) / O objetivo do trabalho consiste em estudar e implementar um método de identificação de modelos de funções de transferência que utiliza como entrada de teste um degrau e que foi proposto inicialmente em Kosaka (2005) e compreende duas fases: obtenção de dados referentes à saída do sistema após a aplicação da entrada degrau e a composição de um sistema matricial formado por uma matriz de Toeplitz a partir desses dados. Na solução deste sistema matricial, estima-se os parâmetros da função de transferência do sistema. Neste estudo é proposta uma generalização do método de identificação, descrito em Kosaka, para funções de transferências instáveis. Esse novo método tem como base a multiplicação, no domínio do tempo, da saída da planta y(t) por uma função exponencial do tipo e-at, sendo a um número real positivo. A constante “a” deve ser suficientemente grande, de modo que y(t)e-at ->0 quando t ->infinito. Com esse procedimento o método identifica uma função de transferência estável (G(s+a)) e então é identificada a função de transferência da planta (G(s)). Os resultados da avaliação mostram que o método generalizado e proposto para funções de transferência instáveis também pode ser aplicado em funções de transferência estáveis. O método generalizado fornece melhores resultados quando é comparada a resposta ao degrau da função de transferência estimada pelo método generalizado com a resposta ao degrau do método proposto originalmente por Kosaka em relação à resposta ao degrau da função de transferência da planta com ruído branco na saída. / The objective of this work is to study and implement a method of identifying models of transfer functions which uses as input one step and that was initially proposed in Kosaka (2005).This method comprises two phases: obtaining data relating to the output of the system after application of the step and the composition of a system matrix composed of by a Toeplitz matrix from these data. With the solution of this matrix system, it is estimated the parameters of the transfer function of the system. In this study is proposed a generalization of the identification method, described in Kosaka, for unstable transfer functions. This new method is based on the multiplication in time domain of the plant's output y (t) by an exponential function of the type and e- at,, where a is a positive real number. The constant a must be large enough so that y (t)e- at ->0 when t -> infinite. With this procedure the method identifies a stable transfer function (G (s + a)) and then is identified the transfer function of the plant (G (s)). The results of the evaluation showed that the method widespread and proposed for unstable transfer functions can also be applied to stable transfer functions. The generalized method showed better results when compared with the step response of the transfer functions estimated by the method with the step response of the transfer function estimed by the method originally proposed by Kosaka (2005), regarding the step response of the transfer function of the plant with white noise in the output.

Resumo: O objetivo do trabalho consiste em estudar e implementar um método de identificação de modelos de funções de transferência que utiliza como entrada de teste um degrau e que foi proposto inicialmente em Kosaka (2005) e compreende duas fases: obtenção de dados referentes à saída do sistema após a aplicação da entrada degrau e a composição de um sistema matricial formado por uma matriz de Toeplitz a partir desses dados. Na solução deste sistema matricial, estima-se os parâmetros da função de transferência do sistema. Neste estudo é proposta uma generalização do método de identificação, descrito em Kosaka, para funções de transferências instáveis. Esse novo método tem como base a multiplicação, no domínio do tempo, da saída da planta y(t) por uma função exponencial do tipo e-at, sendo a um número real positivo. A constante "a" deve ser suficientemente grande, de modo que y(t)e-at ->0 quando t ->infinito. Com esse procedimento o método identifica uma função de transferência estável (G(s+a)) e então é identificada a função de transferência da planta (G(s)). Os resultados da avaliação mostram que o método generalizado e proposto para funções de transferência instáveis também pode ser aplicado em funções de transferência estáveis. O método generalizado fornece melhores resultados quando é comparada a resposta ao degrau da função de transferência estimada pelo método generalizado com a resposta ao degrau do método proposto originalmente por Kosaka em relação à resposta ao degrau da função de transferência da planta com ruído branco na saída. / Abstract: The objective of this work is to study and implement a method of identifying models of transfer functions which uses as input one step and that was initially proposed in Kosaka (2005).This method comprises two phases: obtaining data relating to the output of the system after application of the step and the composition of a system matrix composed of by a Toeplitz matrix from these data. With the solution of this matrix system, it is estimated the parameters of the transfer function of the system. In this study is proposed a generalization of the identification method, described in Kosaka, for unstable transfer functions. This new method is based on the multiplication in time domain of the plant's output y (t) by an exponential function of the type and e- at,, where a is a positive real number. The constant "a" must be large enough so that y (t)e- at ->0 when t -> infinite. With this procedure the method identifies a stable transfer function (G (s + a)) and then is identified the transfer function of the plant (G (s)). The results of the evaluation showed that the method widespread and proposed for unstable transfer functions can also be applied to stable transfer functions. The generalized method showed better results when compared with the step response of the transfer functions estimated by the method with the step response of the transfer function estimed by the method originally proposed by Kosaka (2005), regarding the step response of the transfer function of the plant with white noise in the output. / Orientador: Francisco Villarreal Alvarado / Coorientador: Marcelo Carvalho Minhoto Teixeira / Banca: Edvaldo Assunção / Banca: Márcio Roberto Covacic / Mestre

The objectives of this study are to investigate the dynamic behavior of an apple slab subjected to drying at constant external conditions and under changing in the drying temperatures and to determine the effects of temperature and time combinations at different steps during drying on the process dynamics parameters, time constant and process gain of the system. For this purpose, a semi-batch dryer system was simulated by using integral method of analysis. Initially, the dynamic behavior of the drying temperature was investigated by using first order system dynamic model. Process dynamic parameters, time constant and process gain of the system, for change in drying temperature were determined. Secondly, investigation of the drying kinetics of the apple slab was carried out under constant external conditions in a semi-batch dryer. A mathematical model for diffusion mechanism assumed in one dimensional transient analysis of moisture distribution was solved by using explicit finite difference method of analysis. Thirdly, investigation of the drying kinetics of the apple slab was carried out under change in drying temperature at different time steps during drying. Inverse response system model was used for the representation of the dynamic behavior of drying. Process dynamic parameters, time constant and process gain of the system were determined. Model predicted results for apple slab drying under constant external condition and under step change in the drying temperature were compared with the experimental data.

Research on the electrocoagulation (EC) of hexane extractable materials (HEM) has been conducted at the University of New Orleans using a proprietary bench-scale EC reactor. The original reactor configuration forced the fluid to follow a vertical upward-downward path. An alternate electrode arrangement was introduced so that the path of flow became horizontal. Both configurations were evaluated by comparing the residence time distribution (RTD) data generated in each case. These data produced indication of internal recirculation and stagnant water when the fluid followed a vertical path. These anomalies were attenuated when the fluid flowed horizontally and at a velocity higher than 0.032 m s-1 . A series of EC experiments were performed using a synthetic emulsion with a HEM concentration of approximately 700 mg l-1. It was confirmed that EC of HEM follows first-order kinetics, and kinetic constants of 0.0441 s-1 and 0.0443 s-1 were obtained from applying both the dispersion and tanks-in-series (TIS) models, respectively. In both cases R2 was 0.97. Also, the TIS model indicated that each cell of the EC behaves as an independent continuous-stirred-tank reactor.