Thesis TECNICAS DE REDUCCION APLICADAS A MODELOS DE ELEMENTOS FINITOS ESTOCASTICOS
Loading...
Date
2016
Authors
Journal Title
Journal ISSN
Volume Title
Program
Campus
Universidad Técnica Federico Santa María UTFSM. Casa Central Valparaíso
Abstract
Esta tesis presenta una estrategia para abordar problemas de dise~no basados en con abilidad
de modelos de elementos nitos lineales y no lineales sometidos a una excitacion estocastica. La
con abilidad de un sistema estructural se expresa en terminos de la probabilidad de primera excursi
on. Esta se re ere a la probabilidad de que se genere un comportamiento no deseado, el cual
puede ser caracterizado en terminos de respuestas de interes que sean mayores que ciertos umbrales
establecidos, dentro de un intervalo de tiempo.
Los sistemas estructurales estan representados por modelos de elementos nitos sometidos a
excitaciones basales generadas a partir de modelos estocasticos que simulan la aceleracion del suelo
debido a eventos ssmicos. La estimacion de la probabilidad de falla se realiza mediante la aplicacion
de la tecnica de simulacion avanzada llamada Subset-Simulation.
En general, la solucion al problema de dise~no basado en con abilidad es muy demandante
computacionalmente debido al gran numero de analisis requeridos sobre el modelo de elementos
nitos durante el proceso. Es por ello que se propone una tecnica de reduccion combinada con un
esquema de parametrizacion apropiado para abordar el problema de dise~no de manera e ciente en un
espacio reducido de coordenadas generalizadas. Se implementa un metodo basado en subestructuras
para de nir un modelo reducido del sistema estructural. En particular, se consideran los xedinterface
normal modes e interface constraint modes para de nir el modelo reducido. La e ciencia
de la tecnica de reduccion propuesta en el contexto del dise~no basado en con abilidad es demostrada
a traves de ejemplos de aplicacion.
La tecnica de reduccion es nalmente generalizada con el n de obtener modelos aun mas reducidos
sin deteriorar la exactitud de los resultados. Esto se re ere a la implementacion de una tecnica
optimizada de acoplamiento de subestructuras y al uso de coordenadas generalizadas en terminos
de los interface constraint modes. Analisis de validacion muestran que la tecnica propuesta es una
herramienta numerica e ciente para resolver problemas complejos en ingeniera estructural.
This thesis presents a strategy for dealing with reliability-based design problems of linear and nonlinear nite element models under stochastic excitation. Reliability of a structural system is de ned in terms of the rst excursion probability. This refers to the probability that a unwanted behavior occurs , which can be characterized in terms of responses of interest greater than a set threshold, within a time interval. The structural systems are represented by nite element models under base excitations generated by stochastic models simulating ground acceleration for seismic events. The estimation of the failure probability is performed through the application of the advanced simulation technique called Subset- Simulation. In general, the solution of the reliability-based design problem is computationally very demanding due to the large number of nite element model analyses required during the process. Is for this reason that a model reduction technique combined with an appropriate parametrization scheme is proposed to carry out the design process eciently in a reduced space of generalized coordinates. A method based in components or subestructures is implemented to de ne a reduced-order model for the structural system. In particular, xed-interface normal modes and interface constraint modes are considered to de ne the reduced order model. The eectiveness of the proposed model reduction technique in the context of reliability-based design problems is demonstrated by numerical examples. The reduction technique is nally generalized in order to generate even smaller models without deteriorating the accuracy of the results. This refers to the implementation of an enhanced substructure coupling technique and the use of generalized coordinates in terms of the interface constraint modes. Validation analysis show that the proposed technique is an ecient numerical tool for solving complex problems in structural engineering.
This thesis presents a strategy for dealing with reliability-based design problems of linear and nonlinear nite element models under stochastic excitation. Reliability of a structural system is de ned in terms of the rst excursion probability. This refers to the probability that a unwanted behavior occurs , which can be characterized in terms of responses of interest greater than a set threshold, within a time interval. The structural systems are represented by nite element models under base excitations generated by stochastic models simulating ground acceleration for seismic events. The estimation of the failure probability is performed through the application of the advanced simulation technique called Subset- Simulation. In general, the solution of the reliability-based design problem is computationally very demanding due to the large number of nite element model analyses required during the process. Is for this reason that a model reduction technique combined with an appropriate parametrization scheme is proposed to carry out the design process eciently in a reduced space of generalized coordinates. A method based in components or subestructures is implemented to de ne a reduced-order model for the structural system. In particular, xed-interface normal modes and interface constraint modes are considered to de ne the reduced order model. The eectiveness of the proposed model reduction technique in the context of reliability-based design problems is demonstrated by numerical examples. The reduction technique is nally generalized in order to generate even smaller models without deteriorating the accuracy of the results. This refers to the implementation of an enhanced substructure coupling technique and the use of generalized coordinates in terms of the interface constraint modes. Validation analysis show that the proposed technique is an ecient numerical tool for solving complex problems in structural engineering.
Description
Catalogado desde la version PDF de la tesis.
Keywords
CONFIABILIDAD, MODELOS ESTOCASTICOS, TECNICA DE REDUCCION