Finite-element model updating is the process of using measured data from a structure to update a numerical model representation of the structure. The measured data can represent either the static or dynamic properties of the structure. This document reviews and evaluates several methods of finite-element (FE) model updating, including direct, indirect, and control-based methods for the dynamic case. It is important to have a correct finite-element model obtained using model updating methods either to assess the current condition, or to modify the structure from its current state. In this study, three types of methods were evaluated; direct, indirect, and control based finite-element model updating methods. Each method was first used to update a simple example model for two separate cases. For the first case, the entire set of measured modal parameters were used; and for the second case, only a sub-set of the eigenvalues were used. These examples provide insights into the advantages and disadvantages of various methods. The model updating methods are also used to update a full-scale 42 degree of freedom model. Since it is not practical to measure all the degrees of freedom, the model was reduced using the SEREP model reduction method, down to 18 degrees of freedom. This was done to evaluate the effectiveness of the model updating methods on a real structure. Detailed methodologies and a comparison between the relative advantages and disadvantages between various model updating methods are highlighted in this thesis.

Finite-element model updating is the process of using measured data from a structure to update a numerical model representation of the structure. The measured data can represent either the static or dynamic properties of the structure. This document reviews and evaluates several methods of finite-element (FE) model updating, including direct, indirect, and control-based methods for the dynamic case. It is important to have a correct finite-element model obtained using model updating methods either to assess the current condition, or to modify the structure from its current state. In this study, three types of methods were evaluated; direct, indirect, and control based finite-element model updating methods. Each method was first used to update a simple example model for two separate cases. For the first case, the entire set of measured modal parameters were used; and for the second case, only a sub-set of the eigenvalues were used. These examples provide insights into the advantages and disadvantages of various methods. The model updating methods are also used to update a full-scale 42 degree of freedom model. Since it is not practical to measure all the degrees of freedom, the model was reduced using the SEREP model reduction method, down to 18 degrees of freedom. This was done to evaluate the effectiveness of the model updating methods on a real structure. Detailed methodologies and a comparison between the relative advantages and disadvantages between various model updating methods are highlighted in this thesis.

Análises numéricas confiáveis do comportamento mecânico de materiais como borrachas, dependem muito de uma calibração precisa do modelo constitutivo hiperelástico utilizado. Estes modelos são calibrados ajustando as curvas teóricas aos dados experimentais, obtidos por meio de ensaios usuais. Em muitos casos as amostras de matéria prima desses elastômeros já não se encontram disponíveis ou é impossível fabricar os corpos de prova requeridos. O objetivo deste trabalho é verificar a possibilidade de encontrar constantes constitutivas testando o próprio componente, ao invés dos usuais ensaios de tração, compressão e cisalhamento. A abordagem proposta consiste em criar uma rotina de programação associada a uma função custo onde, a partir de uma estimativa inicial de constantes constitutivas, sejam realizados processos iterativos de otimização buscando aproximar as curvas de força × deslocamento teórica e experimental. Um componente automotivo será utilizado nos estudos e dois modelos constitutivos hiperelásticos serão testados. As equações de tensões nominais dos modelos hiperelásticos serão utilizadas para predizer o comportamento teórico dos ensaios usuais, de forma a verificar a qualidade das constantes obtidas. Conclui-se que é possível utilizar o ensaio da própria peça para caracterizar o material hiperelástico, com resultados comparáveis aos que seriam obtidos com os ensaios típicos para esta aplicação. / The reliable numerical analysis of the mechanical behavior of rubber-like materials depends strongly on accurately calibrated hyperelastic constitutive models. Such models are calibrated by fitting theoretical curves against experimental data obtained in well known tests. In many cases samples of the original elastomer are no longer available or it is impossible to manufacture the specimens required by the standard tests. The aim of this work is verify the possibility of finding the constitutive constants by testing the actual component instead of the usual tensile, compression and shear tests. The proposed approach consists in creating a programming routine with a cost function that, starting from an initial estimate of the constitutive constants, iterate through an optimization algorithm in order to fit the theoretical force × displacement curves to the experimental ones. An automotive component will be used during the studies and two hyperelastic constitutive models will be tested. The nominal stress equations for the hyperelastic models are used to predict the standard tests behavior, to assess the quality of the constants obtained. The results shown that is possible to characterize an hyperelastic material by testing the actual component, with results comparable to those which would be obtained with standard tests.

Análises numéricas confiáveis do comportamento mecânico de materiais como borrachas, dependem muito de uma calibração precisa do modelo constitutivo hiperelástico utilizado. Estes modelos são calibrados ajustando as curvas teóricas aos dados experimentais, obtidos por meio de ensaios usuais. Em muitos casos as amostras de matéria prima desses elastômeros já não se encontram disponíveis ou é impossível fabricar os corpos de prova requeridos. O objetivo deste trabalho é verificar a possibilidade de encontrar constantes constitutivas testando o próprio componente, ao invés dos usuais ensaios de tração, compressão e cisalhamento. A abordagem proposta consiste em criar uma rotina de programação associada a uma função custo onde, a partir de uma estimativa inicial de constantes constitutivas, sejam realizados processos iterativos de otimização buscando aproximar as curvas de força × deslocamento teórica e experimental. Um componente automotivo será utilizado nos estudos e dois modelos constitutivos hiperelásticos serão testados. As equações de tensões nominais dos modelos hiperelásticos serão utilizadas para predizer o comportamento teórico dos ensaios usuais, de forma a verificar a qualidade das constantes obtidas. Conclui-se que é possível utilizar o ensaio da própria peça para caracterizar o material hiperelástico, com resultados comparáveis aos que seriam obtidos com os ensaios típicos para esta aplicação. / The reliable numerical analysis of the mechanical behavior of rubber-like materials depends strongly on accurately calibrated hyperelastic constitutive models. Such models are calibrated by fitting theoretical curves against experimental data obtained in well known tests. In many cases samples of the original elastomer are no longer available or it is impossible to manufacture the specimens required by the standard tests. The aim of this work is verify the possibility of finding the constitutive constants by testing the actual component instead of the usual tensile, compression and shear tests. The proposed approach consists in creating a programming routine with a cost function that, starting from an initial estimate of the constitutive constants, iterate through an optimization algorithm in order to fit the theoretical force × displacement curves to the experimental ones. An automotive component will be used during the studies and two hyperelastic constitutive models will be tested. The nominal stress equations for the hyperelastic models are used to predict the standard tests behavior, to assess the quality of the constants obtained. The results shown that is possible to characterize an hyperelastic material by testing the actual component, with results comparable to those which would be obtained with standard tests.

Análises numéricas confiáveis do comportamento mecânico de materiais como borrachas, dependem muito de uma calibração precisa do modelo constitutivo hiperelástico utilizado. Estes modelos são calibrados ajustando as curvas teóricas aos dados experimentais, obtidos por meio de ensaios usuais. Em muitos casos as amostras de matéria prima desses elastômeros já não se encontram disponíveis ou é impossível fabricar os corpos de prova requeridos. O objetivo deste trabalho é verificar a possibilidade de encontrar constantes constitutivas testando o próprio componente, ao invés dos usuais ensaios de tração, compressão e cisalhamento. A abordagem proposta consiste em criar uma rotina de programação associada a uma função custo onde, a partir de uma estimativa inicial de constantes constitutivas, sejam realizados processos iterativos de otimização buscando aproximar as curvas de força × deslocamento teórica e experimental. Um componente automotivo será utilizado nos estudos e dois modelos constitutivos hiperelásticos serão testados. As equações de tensões nominais dos modelos hiperelásticos serão utilizadas para predizer o comportamento teórico dos ensaios usuais, de forma a verificar a qualidade das constantes obtidas. Conclui-se que é possível utilizar o ensaio da própria peça para caracterizar o material hiperelástico, com resultados comparáveis aos que seriam obtidos com os ensaios típicos para esta aplicação. / The reliable numerical analysis of the mechanical behavior of rubber-like materials depends strongly on accurately calibrated hyperelastic constitutive models. Such models are calibrated by fitting theoretical curves against experimental data obtained in well known tests. In many cases samples of the original elastomer are no longer available or it is impossible to manufacture the specimens required by the standard tests. The aim of this work is verify the possibility of finding the constitutive constants by testing the actual component instead of the usual tensile, compression and shear tests. The proposed approach consists in creating a programming routine with a cost function that, starting from an initial estimate of the constitutive constants, iterate through an optimization algorithm in order to fit the theoretical force × displacement curves to the experimental ones. An automotive component will be used during the studies and two hyperelastic constitutive models will be tested. The nominal stress equations for the hyperelastic models are used to predict the standard tests behavior, to assess the quality of the constants obtained. The results shown that is possible to characterize an hyperelastic material by testing the actual component, with results comparable to those which would be obtained with standard tests.