Thesis ANÁLISIS DEL COMPORTAMIENTO DEL ESCALONAMIENTO Y CÁLCULO DE LA DEFLEXIÓN POR ALABEO TÉRMICO EN EL DISEÑO DE PAVIMENTO DE HORMIGÓN
Date
2016
Authors
ARANCIBIA MUÑOZ, ALEJANDRA FRANCISCA
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
El modelo de escalonamiento de la GEMDP fue calibrado en Chile utilizando losas con juntas transversales esviadas en pavimentos de hormigón, para lo cual se emplea un factor de conversión para convertir el efecto del escalonamiento en las losas con Juntas Transversales Esviadas (JTE) a losas con Juntas Transversales Rectas (JTR). El primer objetivo de este trabajo es la verificación del factor de conversión al considerar el efecto de las JTE en las losas de pavimento. El análisis del escalonamiento en losas con JTE se realiza mediante el programa de elementos finitos EverFE, tomando como variable el ángulo de las juntas transversales para conseguir la deflexión deseada. Con el modelo de escalonamiento de la GEMDP, se calcula el escalonamiento resultante al aplicar un eje simple como carga en un sistema de 6 losas cuadradas de dimensión 3,5 [m], con una distribución de 2 x 3. Con lo anterior, se calcula el factor que relaciona los escalonamientos de losas con JTR y con JTE, teniéndose una tendencia polinomial cuadrática de este factor al ir variando el ángulo de la junta transversal. Además, se obtiene una relación lineal entre los escalonamientos de losas con JTR y con JTE, que va variando con respecto al sentido del giro y al valor de la inclinación, considerando la posición de la carga en la configuración del sistema de losas.El segundo objetivo es el cálculo de las deflexiones originado por el alabeo térmico mediante la utilización de Redes Neuronales Artificiales (RNA). Estas deflexiones que se localizan en el borde y esquina de las losas son producto de los gradientes de temperatura y de humedad. Para obtener el alabeo térmico de las losas se escoge la configuración de RNA más óptima para correlacionar el comportamiento de los datos. Los datos necesarios para alimentar la RNA se obtienen en las ciudades de Arica, Valparaíso, Santiago, Concepción, Puerto Montt y Punta Arenas. El uso de RNA para el cálculo de las deflexiones es parte de la metodología del modelo de escalonamiento de la GEMDP, destacándose que los resultados se obtienen a partir de datos generados en ciudades de Chile. El cálculo de las deflexiones por el alabeo térmico se lleva a cabo con las RNA, donde se ingresa el espaciamiento entre juntas transversales, el radio de rigidez relativa y el gradiente de temperatura adimensional de Korenev, entregando como dato de salida la deflexión producida en la losa. Para alimentar la RNA, se genera una base de datos usando los datos climáticos y del suelo de las ciudades, ingresándose al programa de elementos finitos ISLAB2000 para conseguir la deflexión en cada caso. El programa que se utiliza para las RNA es el software matemático MATLAB, creándose un código para reagrupar los errores y los coeficientes de correlación que se generan en cada configuración. De esta forma, se escoge la configuración consistente a tres capas ocultas con 10, 6 y 4 neuronas por capa, lográndose modelar las deformaciones de la base de datos de manera óptima.
MEPDG’s faulting model was calibrated in Chile using slab with skewed tranverse joint in concrete pavement, for which a conversion factor is used to transform the faulting efect in slabs with skewed tranverse joint to a slabs with straight tranverse joint. The first objetive in this work is to verify the conversion factor to consider the skewed tranverse joint efect on pavement slabs. The faulting analysis on slabs with skewed tranverse joint is done by the finite element program EverFE, the angle of the joint is taking as a variable to get the required deflection. Using the MEPDG’s faulting model, the resulting faulting of applying a single axle load in a set of six square-slabs of dimension 3.5 [m], with a 2x3 distribution, is calculated. Considering this, the factor that relates the faulting of slabs with straight tranverse joint and skewed tranverse joint is calculated, obtaining a quadratic polinomial behavior considering tranverse joint angle variable. Also, a linear relationship between faulting of slabs with straight tranverse joint and skewed tranverse joint is obtained by varying the direction and value of rotation, considering the position of the load on the system configuration slab.The secund objetive is to calculated curling deflections using Artificial Neural Networks (ANN). This deflections caused by temperature and moisture are located at the edge and corner slabs. An optimal setting ANN is chosen to obtain curling deflection slab to correlate the behavior of the data. The data needed to feed the ANN are obtained in cities like Arica, Valparaíso, Santiago, Concepción, Puerto Montt and Punta Arenas. The application of ANN to calculated curling deflections is part of MEPDG’s faulting model, but applied in Chilean conditions. The input data to calculated curling deflection using ANN are transverse joint spacing, radius of relative stiffness and Korenev’s nondimensional temperature gradient, it gives as an output data slab deflection. To feed the ANN, a database is generated using the climatic and soil data of cities, thus entering the finite element program ISLAB2000 for deflection in each case. The program used for ANN is the mathematical software MATLAB, creating a code to regroup errors and correlation coefficients generated in each configuration. Thus, the configuration that achieves optimally model the behavior of the database consist of three hidden layers with 10, 6 and 4 neurons in each layer.
MEPDG’s faulting model was calibrated in Chile using slab with skewed tranverse joint in concrete pavement, for which a conversion factor is used to transform the faulting efect in slabs with skewed tranverse joint to a slabs with straight tranverse joint. The first objetive in this work is to verify the conversion factor to consider the skewed tranverse joint efect on pavement slabs. The faulting analysis on slabs with skewed tranverse joint is done by the finite element program EverFE, the angle of the joint is taking as a variable to get the required deflection. Using the MEPDG’s faulting model, the resulting faulting of applying a single axle load in a set of six square-slabs of dimension 3.5 [m], with a 2x3 distribution, is calculated. Considering this, the factor that relates the faulting of slabs with straight tranverse joint and skewed tranverse joint is calculated, obtaining a quadratic polinomial behavior considering tranverse joint angle variable. Also, a linear relationship between faulting of slabs with straight tranverse joint and skewed tranverse joint is obtained by varying the direction and value of rotation, considering the position of the load on the system configuration slab.The secund objetive is to calculated curling deflections using Artificial Neural Networks (ANN). This deflections caused by temperature and moisture are located at the edge and corner slabs. An optimal setting ANN is chosen to obtain curling deflection slab to correlate the behavior of the data. The data needed to feed the ANN are obtained in cities like Arica, Valparaíso, Santiago, Concepción, Puerto Montt and Punta Arenas. The application of ANN to calculated curling deflections is part of MEPDG’s faulting model, but applied in Chilean conditions. The input data to calculated curling deflection using ANN are transverse joint spacing, radius of relative stiffness and Korenev’s nondimensional temperature gradient, it gives as an output data slab deflection. To feed the ANN, a database is generated using the climatic and soil data of cities, thus entering the finite element program ISLAB2000 for deflection in each case. The program used for ANN is the mathematical software MATLAB, creating a code to regroup errors and correlation coefficients generated in each configuration. Thus, the configuration that achieves optimally model the behavior of the database consist of three hidden layers with 10, 6 and 4 neurons in each layer.
Description
Catalogado desde la version PDF de la tesis.
Keywords
DEFLEXION POR ALABEO TERMICO , ESCALONAMIENTO , EVERFE , GEMDP , ISLAB2000 , JUNTAS TRANSVERSALES ESVIADAS , REDES NEURONALES ARTIFICIALES