Análise inversa para determinação de parâmetros de deformabilidade de solos

• Autor(a) principal: Guimarães, Adinele Gomes
• Data de Publicação: 2008
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: LOCUS Repositório Institucional da UFV
• Texto Completo: http://locus.ufv.br/handle/123456789/803
• topic: Análise inversa; Modelos constitutivos; Inverse analysis; Constitutive models; CNPQ::ENGENHARIAS::ENGENHARIA CIVIL::GEOTECNICA

The prevision of soil mechanical behavior demands the use of constitutive models which represent more appropriately its stress-strain relationship. Some models may include parameters difficult to obtain in laboratory and their determination, when involving engineering judgment, is better accomplished by experienced users of a particular model. Alternatively, the parameters can be obtained using inverse analysis. The mathematical procedure consists basically in the formulation of an objective function, that determines the difference between measured laboratory values and those calculated by the model and the selection of an optimization strategy that makes possible the search for the minimum of the objective function. In order to verify the best strategy, some mathematical programming methods (Newton-modified, Gauss-Newton, Levemberg-Marquart, L-BFGS-B) and a genetic algorithm were used in the calibration of the hyperbolic and Lade-Kim models, for a group of triaxial tests in sand samples. A group of cubic triaxial tests in saturated samples of residual gneissic soil was used in the calibration of Lade-Kim model to verify the best way to consider the data in the objective function, considering factors based on the standard deviation, in the covariance and in a maximum scalar factor. Other sets of cubic triaxial and hydrostatic tests, in saturated and non saturated samples of residual gneissic soil, were used to study the use of the inverse analysis in different groups of laboratory results. Tests with unconventional paths of tensions were simulated in order to verify certain parameters of the potential for inverse analysis. Comparisons between numerical and experimental results verify the best identification strategy, indicate the best form of weighting matrix of the objective function and show the capacity and advantages of the inverse analysis in the determination of soil deformability parameters.