Thesis Simulación multiescala del viento sobre terreno complejo mediante el método embebido WRF-LES y asimilación variacional de datos 4D
Loading...
Date
2019-11
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Mecánica
Campus
Campus Casa Central Valparaíso
Abstract
Con el fin de lograr una correcta predicción del recurso viento en terreno complejo para zonas muy localizadas, se llevaron a cabo una serie de simulaciones numéricas meteorológicas multiescala con datos reales de alta resolución utilizando el modelo atmosférico WRF y una clausura LES para la turbulencia. El acoplamiento de la meso y microescala se logró a través de la técnica de dominios anidados hasta llegar a una resolución de aproximadamente 2 [m]. Para corregir las desviaciones e incertidumbres propias de una simulación numérica, se propuso utilizar un esquema de asimilación de datos cuatridimensional en el dominio mas interior.
La validación de la implementación se basa en 4 casos. Los primeros dos corresponden a una simulación en el sitio de pruebas de turbinas en Høvsøre, Dinamarca, el cual es un terreno cuasi-plano ampliamente estudiado. La primera simulación valida el acercamiento numérico y la segunda muestra la influencia de la asimilación de datos en la capa límite considerando 5 niveles de un mástil meteorológico ubicado en el centro del dominio.
Los últimos dos casos corresponden a la aplicación de la misma metodología en terreno complejo para simular la colina de Bolund ubicada también en Dinamarca. Estas dos simulaciones exponen: (i) el comportamiento del modelo para un flujo de viento neutralmente estratificado con burbuja de separación y, (ii) la influencia de la asimilación de datos multipunto utilizando la información de 8 mástiles en 3 niveles cercanos a la superficie.
Los resultados obtenidos muestran que es posible obtener predicciones más certeras y que replican el comportamiento turbulento del viento a las escalas simuladas y que, además, la asimilación de datos mejora esta predicción en el caso de terreno plano en un 10 %. En terreno complejo, la asimilación de datos no logra mejorar la solución debido a la cercanía de las mediciones con el suelo y los forzamientos inducidos por el terreno, sin embargo toda la investigación da pie a un uso operativo de la metodología y los códigos propuestos.
In order to achieve a correct wind resource prediction in complex terrain for localized areas, a series of multiscale meteorological numerical simulations with high resolution real data was carried out using the WRF atmospheric model and a LES turbulence closure. The meso-microscale coupling was done through nested domains until a resolution of approximately 2 [m] was reached. To correct the numerical simulation deviations and uncertainties, a four-dimensional data assimilation scheme in the innermost domain was proposed. The implementation’s validation is based in 4 cases. The first two correspond to a simulation at the turbine test site in Høvsøre, Denmark, which is a widely studied quasi-flat terrain. The first simulation validates the numerical approach and the second one shows the influence of the data assimilation in the boundary layer considering 5 levels of a meteorological mast located at the domain center. The last two cases correspond to the application of the same methodology but in complex terrain to simulate the Bolund hill, also located in Denmark. These two simulations exposes: (i) the behavior of the model for a neutrally stratified wind flow with a separation bubble and, (ii) the multipoint data assimilation influence using information of 8 mast at 3 levels near the surface. The obtained results shows that it is possible to obtain more accurate predictions that replicate the turbulent wind behavior at simulated scales and that, in addition, data assimilation improves the flat terrain prediction by 10 %. In complex terrain, the data assimilation fails to improve the solution due to the proximity of the measurements with the ground and the terrain induced forcing. However, all the research gives rise to an operational use of the proposed methodology and codes.
In order to achieve a correct wind resource prediction in complex terrain for localized areas, a series of multiscale meteorological numerical simulations with high resolution real data was carried out using the WRF atmospheric model and a LES turbulence closure. The meso-microscale coupling was done through nested domains until a resolution of approximately 2 [m] was reached. To correct the numerical simulation deviations and uncertainties, a four-dimensional data assimilation scheme in the innermost domain was proposed. The implementation’s validation is based in 4 cases. The first two correspond to a simulation at the turbine test site in Høvsøre, Denmark, which is a widely studied quasi-flat terrain. The first simulation validates the numerical approach and the second one shows the influence of the data assimilation in the boundary layer considering 5 levels of a meteorological mast located at the domain center. The last two cases correspond to the application of the same methodology but in complex terrain to simulate the Bolund hill, also located in Denmark. These two simulations exposes: (i) the behavior of the model for a neutrally stratified wind flow with a separation bubble and, (ii) the multipoint data assimilation influence using information of 8 mast at 3 levels near the surface. The obtained results shows that it is possible to obtain more accurate predictions that replicate the turbulent wind behavior at simulated scales and that, in addition, data assimilation improves the flat terrain prediction by 10 %. In complex terrain, the data assimilation fails to improve the solution due to the proximity of the measurements with the ground and the terrain induced forcing. However, all the research gives rise to an operational use of the proposed methodology and codes.
Description
Keywords
Energía eólica, Simulaciones meteorológicas, Potencia eléctrica