The main aim of this thesis is to compare actual power production from an existing wind farm with power production prediction by WindSim, which is a CFD tool based on the nonlinear flow model. The wind farm that is being worked on is located in Northern Sweden and has high orographic complexity with forested hilly terrain. There is 1 year record of met-mast wind measurements and nearly 2 years record of production data.Firstly roughness and height contours data are put as an input in order to simulate and generate wind fields over the complex terrain. In addition forest model is used to get more detailed roughness height. After generating wind fields existing turbine locations and 1-year wind speed measurement are imported.The results show that how accurate are the CFD calculations to solve turbulence in a complex terrain. Comparison between actual production data with energy production results by simulations is the main approach of this thesis work to validate the simulations.The results indicate that both WAsP and WindSim have overestimated energy production and wind speed as well. However particularly with WindSim forest module CFD calculations have more accurate results than without forest module and WAsP estimations.

Thunderstorms that occur in areas of complex terrain are a major severe weather hazard in the intermountain western U.S. Short-term quantitative estimation (QPE) of precipitation in complex terrain is a pressing need to better forecast flash flooding. Currently available techniques for QPE, that utilize a combination of rain gauge and weather radar information, may underestimate precipitation in areas where gauges do not exist or there is radar beam blockage. These are typically very mountainous and remote areas, that are quite vulnerable to flash flooding because of the steep topography. Lightning has been one of the novel ways suggested by the scientific community as an alternative to estimate precipitation over regions that experience convective precipitation, especially those continental areas with complex topography where the precipitation sensor measurements are scarce. This dissertation investigates the relationship between cloud-to-ground lightning and precipitation associated with convection with the purpose of estimating precipitation- mainly over areas of complex terrain which have precipitation sensor coverage problems (e.g. Southern Arizona).The results of this research are presented in two papers. The first, entitled Toward Development of Improved QPE in Complex Terrain Using Cloud-to-Ground Lighting Data: A case Study for the 2005 Monsoon in Southern Arizona, was published in the Journal of Hydrometeorology in December 2012. This initial study explores the relationship between cloud-to-ground lightning occurrences and multi-sensor gridded precipitation over southern Arizona. QPE is performed using a least squares approach for several time resolutions (seasonal -June, July and August-, 24 hourly and hourly) and for a 8 km grid size. The paper also presents problems that arise when the time resolution is increased, such as the spatial misplacing of discrete lightning events with gridded precipitation and the need to define a "diurnal day" that is synchronized with the diurnal cycle of convection. The second manuscript (unpublished), entitled An Improved QPE Over Complex Terrain by Using Cloud-to-Ground Lightning Occurrences, provides a new method to retrieve lightning-derived precipitation at 5 minutes and 5 Km time and space resolutions. A stationary model that employs spatio-temporal neighboring (Space and Time Invariant model -STI) improves upon the least squares method in the first paper. By applying a Kalman filter to the STI model, lightning-precipitation is retrieved by a dynamic model that changes in time. The results for seasonal and 5 minutes time resolution show that the dynamic model improves the retrievals derived by the STI model.

This thesis focuses on modelling of wind turbine wakes when they are affected by real complex terrain features, such as hills and forests, and also examines the effect of the rotational momentum imparted to the downstream wake from the rotor blades. Modelling work is carried out using the commercial Computational Fluid Dynamics (CFD) solver FLUENT. Motivation for this project was the fact that there is currently limited knowledge on several issues that affect the operation of a wind farm in a complex terrain environment. Wind developers normally use commercial, easy-to-use software (such as WAsP) to predict the potential wind farm output , which are based on simple linear models to model wakes and wind flow orographic effects and have been calibrated for cases of simple terrain. In cases of complex terrain, they are expected to give errors due to arising non-linearities. After a review of the relevant literature, the chosen CFD procedure is explained. This involves the use of 3-D Reynolds Averaged Navier-Stokes equations using the Reynolds Stress Model for the turbulence closure, in order to account for the anisotropy in atmospheric turbulence. The Virtual Blade Model in FLUENT is demonstrated as a useful tool for modelling the rotor effects without the need of meshing the rotor geometry in detail and avoiding significant computational cost. The approach is initially validated with the widely documented Nibe measurements, which involved full-scale observations of a single wake over at terrain. The model is also tested in the case of a wind turbine operating at the summit of an ideal, Gaussian hill. The wake development is examined in detail and in comparison with another CFD approach. Most notably, a slight divergence is found in the wake path as it evolves downwind. Additionally, the proposed approaches of modelling the neutral atmospheric ow over a real hill and over a forest are validated with full-scale measurements. Ultimately, the work includes the modelling of real wind farms over complex terrain and validating the results with measurements. A coastal complex terrain wind farm is initially examined and results are validated with SCADA measurements and compared with results using the WAsP wind modelling software. Finally, a wind farm over hilly terrain and near forests is also considered and the effect of the forest in the wake is studied. Results are also validated with full-scale measurements.

[pt] O objetivo principal do presente trabalho foi investigar computacionalmente o escoamento e a dispersão de poluentes atmosféricos sobre topografias complexas tridimensionais em escala de laboratório. Foram realizadas simulações numéricas de escoamentos neutros e estavelmente estratificados sobre colinas e também sobre terreno plano. A modelagem matemática, baseada na solução das equações gerais de conservação, inclui o modelo de tensões de Reynolds para a turbulência e um modelo de duas camadas para o tratamento do escoamento na região próxima à parede. O código comercial Fluent (Versão 6.0.12), que emprega o método de volumes finitos, foi utilizado nas simulações computacionais. Os resultados numéricos foram comparados a dados obtidos em experimentos em túnel de vento disponíveis na literatura. Também foram realizadas comparações com resultados obtidos com a utilização do modelo (k menos épsilon) clássico. A comparação entre os resultados obtidos com as diversas modelagens numéricas e os dados experimentais mostrou que a utilização conjunta do modelo de tensões de Reynolds e do tratamento em duas camadas produziu os melhores resultados na predição do escoamento. O desempenho dessa modelagem foi particularmente superior na representação da recirculação no escoamento na região a jusante da colina. Com relação ao cálculo das concentrações, os resultados obtidos foram razoáveis nas regiões mais distantes da fonte quando comparados aos experimentais. Na região mais próxima à fonte emissora, foram calculadas concentrações excessivamente altas ao nível do solo. Estas discrepâncias foram atribuídas ao fato de ter-se utilizado um modelo de difusividade turbulenta isotrópica para os cálculos da dispersão turbulenta do poluente. Ainda assim, os campos de concentrações apresentados mostraram importantes aspectos qualitativos relativos ao problema como, por exemplo, os efeitos da estabilidade atmosférica na dispersão do poluente, que foram adequadamente previstos. / [en] The main objective of the present work was to investigate computationally the flow and the dispersion of atmospheric pollutants over three dimensional complex topographies in laboratory scale. The investigations included the numerical simulation on the neutral and stably stratified flows over hills and flat terrain. The mathematical model was based on the solution of the general conservation equations and included the Reynolds stress model for turbulence and a two layer zonal model for the flow treatment in the near wall region. The commercial code Fluent (Version 6.0.12), which is based on the finite volume method, was employed in the computational simulations. The numerical results were compared to data obtained in wind tunnel experiments, available in the literature. Comparisons were also made with results obtained by employing the standard (k less épsilon) model for turbulence. The comparisons between the experimental data and the numerical results showed that the combined use of the Reynolds stress model and the two layer treatment provided the best results for the flow representation. This modeling approach was particularly superior in representing the flow recirculation on the leeside of the hill. The predicted concentrations results were reasonably good at regions far away from the emission source. In the near source regions, the ground level concentrations were overestimated by the numerical modeling. These discrepancies were attributed to the employment of an isotropic turbulent diffusivity model in the turbulent dispersion calculations. Nevertheless, the calculated concentration fields represented well important qualitative features of PUC Rio.

Micrositing é o campo do posicionamento estratégico dos aerogeradores na área do parque eólico visando a con guração mais promissora em termos econômicos ou de produção. Dado que em terrenos complexos as características do vento variam na área do parque eólico de forma não-linear, emprega-se a modelagem numérica do vento por CFD para extrapolar os dados medidos para toda a região. O presente trabalho consiste no desenvolvimento de uma metodologia de micrositing em terrenos complexos capaz de auxiliar no projeto do layout e seleção do tipo e altura de eixo do aerogerador que maximiza o fator de capacidade (FC) utilizando softwares livres de código aberto. A metodologia consiste na simulação do vento para várias direções de incidência, assimilação das medições, convers ão de velocidade em densidade de potência, ponderação por frequência de ocorrência de cada direção, sobreposição, seleção das coordenadas dos aerogeradores e cálculo do FC para diversas con gurações de tipo e altura de eixo dos aerogeradores. Veri cação, validação e seleção das constantes do modelo de turbulência é realizada anteriormente às simulações Veri cou-se que o modelo k- produziu um escoamento horizontalmente homog êneo e que o melhor desempenho na validação foi obtido com a escolha de constantes para escoamentos atmosféricos. A metodologia foi demonstrada em uma região de terreno complexo em que o FC do parque eólico proposto apresentou caráter convergente com o re no progressivo da malha, porém oscilatório em termos do número de direções simuladas. Por m, obteve-se FC brutos superiores a 40% para as cinco melhores con gurações e de aproximadamente 52% no melhor caso, indicando bom potencial eólico. A metodologia foi capaz de preencher uma lacuna na literatura cientí ca de micrositing ao possibilitar o planejamento do layout, tipo de aerogerador e altura de eixo, bem como a estimativa da produção e FC brutos da usina em terrenos complexos. Além disso, a estrutura de trabalho com o uso de recursos computacionais livres e de código aberto reforça o caráter de desenvolvimento contínuo, compartilhamento e transparência da metodologia. / Micrositing is the eld concerned with the strategic positioning of wind turbines in the wind farm area aimed at the most promissing con guration economically- or yield-wise. Given the wind characteristics vary non-linearly across the wind farm area in complex terrain, numerical wind modeling with CFD is employed to extrapolate the measured data to the whole site. The present work consists of the development of a micrositing methodology in complex terrain capable of assisting the layout project and selection of wind turbine type and hub height that maximizes the capacity factor (CF) using free and open-source software. The methodology consists of simulating the wind for a number of incoming directions, assimilation of measurements, conversion of wind speed into power density, weighing by frequency of occurrence of each direction, overlapping, selection of wind turbine coordinates and CF calculation for a number of wind turbine types and hub heights. Veri cation, validation and selection of turbulence model constants is performed previous to the simulations It was veri ed that the k- model is able to sustain horizontally-homogeneous ow and that the classic atmospheric ow constants performed best in the validation step. The methodology was demonstrated in a complex terrain region for which the proposed wind farm CF showed converging behavior with progressive mesh re nement, however oscillating with the number of wind directions simulated. Ultimately, CF greater than 40% were obtained with the ve best performing con gurations and approximately 52% in the best case scenario, suggesting good wind potential. The methodology was capable of lling a major gap in the scienti c literature of micrositing for allowing the layout planning, selection of wind turbine type and hub height, as well as gross production estimates and CF for the wind farm in complex terrain. Additionally, the free and open-source-based framework strengthens the continuous development, sharing and transparency of the methodology.

A model has been developed using the general-purpose Navier-Stokes solver CFX4 to simulate Atmospheric Boundary Layer flow over complex terrain. This model has been validated against the measured data from the Askervein Hill experiment, and has been shown to perform well. The CFD model is also compared to the WAsP linear model of wind flow over topography, and a significant improvement is noted for flow over complex topography. Boundary conditions, gridding issues and sensitivity to other solver parameters have all been investigated. An advanced roughness model has been developed to simulate flow over forest canopies, using a resistive body force within the canopy volume. The model is validated against measured data for simple 2D cases, and for a complex 3D case over real topography. The model is shown to give a more physically realistic profile for the wind speed in and just above forest canopies than the standard roughness length model used in most CFD simulations. An automated methodology for setting up CFD simulations using the models described has been developed. A custom pre-processing package to implement this has been written, to enable the use of the CFD methodology in a commercial environment.

Wind turbines connected to the distribution network can be the cause of voltage fluctuations and resulting fluctuations in the light intensity emitted by light bulbs. These fluctuations may cause people disturbance. A model has been developed to obtain a flicker prediction which is useful in the design process of a wind farm. The model is based exclusively in the frequency domain (FD). This new approach allows very fast and efficient evaluation. The impact of individual parameters is often easier to recognise and evaluate in a FD-representation. The following factors leading to flicker disturbances from a wind farm have been considered in detail: The wind spectrum: Effects of terrain and wind farm wakes on the wind turbulence spectrum have been considered and existing models have been expanded. The wind coherence: A new coherence model for large separation distances has been derived for use within a wind farm. Effects of the terrain on the coherence of power produced by turbines within a wind farm have been considered. The wind turbine: A simplified dynamic wind turbine model allows the prediction of turbine specific contributions to flicker for a variety of wind turbines using a minimal set of parameters. The flickermeter: Flicker measurements are found to sometimes neglect the impact of low frequency voltage variations. These are found to be very important for the correct flicker prediction. A new FD-flickermeter has been developed. The model has been validated against experimental data and a sensitivity analysis shows which parameters are most likely to influence the voltage flicker and which are best altered to minimise the flicker.

Micrositing é o campo do posicionamento estratégico dos aerogeradores na área do parque eólico visando a con guração mais promissora em termos econômicos ou de produção. Dado que em terrenos complexos as características do vento variam na área do parque eólico de forma não-linear, emprega-se a modelagem numérica do vento por CFD para extrapolar os dados medidos para toda a região. O presente trabalho consiste no desenvolvimento de uma metodologia de micrositing em terrenos complexos capaz de auxiliar no projeto do layout e seleção do tipo e altura de eixo do aerogerador que maximiza o fator de capacidade (FC) utilizando softwares livres de código aberto. A metodologia consiste na simulação do vento para várias direções de incidência, assimilação das medições, convers ão de velocidade em densidade de potência, ponderação por frequência de ocorrência de cada direção, sobreposição, seleção das coordenadas dos aerogeradores e cálculo do FC para diversas con gurações de tipo e altura de eixo dos aerogeradores. Veri cação, validação e seleção das constantes do modelo de turbulência é realizada anteriormente às simulações Veri cou-se que o modelo k- produziu um escoamento horizontalmente homog êneo e que o melhor desempenho na validação foi obtido com a escolha de constantes para escoamentos atmosféricos. A metodologia foi demonstrada em uma região de terreno complexo em que o FC do parque eólico proposto apresentou caráter convergente com o re no progressivo da malha, porém oscilatório em termos do número de direções simuladas. Por m, obteve-se FC brutos superiores a 40% para as cinco melhores con gurações e de aproximadamente 52% no melhor caso, indicando bom potencial eólico. A metodologia foi capaz de preencher uma lacuna na literatura cientí ca de micrositing ao possibilitar o planejamento do layout, tipo de aerogerador e altura de eixo, bem como a estimativa da produção e FC brutos da usina em terrenos complexos. Além disso, a estrutura de trabalho com o uso de recursos computacionais livres e de código aberto reforça o caráter de desenvolvimento contínuo, compartilhamento e transparência da metodologia. / Micrositing is the eld concerned with the strategic positioning of wind turbines in the wind farm area aimed at the most promissing con guration economically- or yield-wise. Given the wind characteristics vary non-linearly across the wind farm area in complex terrain, numerical wind modeling with CFD is employed to extrapolate the measured data to the whole site. The present work consists of the development of a micrositing methodology in complex terrain capable of assisting the layout project and selection of wind turbine type and hub height that maximizes the capacity factor (CF) using free and open-source software. The methodology consists of simulating the wind for a number of incoming directions, assimilation of measurements, conversion of wind speed into power density, weighing by frequency of occurrence of each direction, overlapping, selection of wind turbine coordinates and CF calculation for a number of wind turbine types and hub heights. Veri cation, validation and selection of turbulence model constants is performed previous to the simulations It was veri ed that the k- model is able to sustain horizontally-homogeneous ow and that the classic atmospheric ow constants performed best in the validation step. The methodology was demonstrated in a complex terrain region for which the proposed wind farm CF showed converging behavior with progressive mesh re nement, however oscillating with the number of wind directions simulated. Ultimately, CF greater than 40% were obtained with the ve best performing con gurations and approximately 52% in the best case scenario, suggesting good wind potential. The methodology was capable of lling a major gap in the scienti c literature of micrositing for allowing the layout planning, selection of wind turbine type and hub height, as well as gross production estimates and CF for the wind farm in complex terrain. Additionally, the free and open-source-based framework strengthens the continuous development, sharing and transparency of the methodology.

Micrositing é o campo do posicionamento estratégico dos aerogeradores na área do parque eólico visando a con guração mais promissora em termos econômicos ou de produção. Dado que em terrenos complexos as características do vento variam na área do parque eólico de forma não-linear, emprega-se a modelagem numérica do vento por CFD para extrapolar os dados medidos para toda a região. O presente trabalho consiste no desenvolvimento de uma metodologia de micrositing em terrenos complexos capaz de auxiliar no projeto do layout e seleção do tipo e altura de eixo do aerogerador que maximiza o fator de capacidade (FC) utilizando softwares livres de código aberto. A metodologia consiste na simulação do vento para várias direções de incidência, assimilação das medições, convers ão de velocidade em densidade de potência, ponderação por frequência de ocorrência de cada direção, sobreposição, seleção das coordenadas dos aerogeradores e cálculo do FC para diversas con gurações de tipo e altura de eixo dos aerogeradores. Veri cação, validação e seleção das constantes do modelo de turbulência é realizada anteriormente às simulações Veri cou-se que o modelo k- produziu um escoamento horizontalmente homog êneo e que o melhor desempenho na validação foi obtido com a escolha de constantes para escoamentos atmosféricos. A metodologia foi demonstrada em uma região de terreno complexo em que o FC do parque eólico proposto apresentou caráter convergente com o re no progressivo da malha, porém oscilatório em termos do número de direções simuladas. Por m, obteve-se FC brutos superiores a 40% para as cinco melhores con gurações e de aproximadamente 52% no melhor caso, indicando bom potencial eólico. A metodologia foi capaz de preencher uma lacuna na literatura cientí ca de micrositing ao possibilitar o planejamento do layout, tipo de aerogerador e altura de eixo, bem como a estimativa da produção e FC brutos da usina em terrenos complexos. Além disso, a estrutura de trabalho com o uso de recursos computacionais livres e de código aberto reforça o caráter de desenvolvimento contínuo, compartilhamento e transparência da metodologia. / Micrositing is the eld concerned with the strategic positioning of wind turbines in the wind farm area aimed at the most promissing con guration economically- or yield-wise. Given the wind characteristics vary non-linearly across the wind farm area in complex terrain, numerical wind modeling with CFD is employed to extrapolate the measured data to the whole site. The present work consists of the development of a micrositing methodology in complex terrain capable of assisting the layout project and selection of wind turbine type and hub height that maximizes the capacity factor (CF) using free and open-source software. The methodology consists of simulating the wind for a number of incoming directions, assimilation of measurements, conversion of wind speed into power density, weighing by frequency of occurrence of each direction, overlapping, selection of wind turbine coordinates and CF calculation for a number of wind turbine types and hub heights. Veri cation, validation and selection of turbulence model constants is performed previous to the simulations It was veri ed that the k- model is able to sustain horizontally-homogeneous ow and that the classic atmospheric ow constants performed best in the validation step. The methodology was demonstrated in a complex terrain region for which the proposed wind farm CF showed converging behavior with progressive mesh re nement, however oscillating with the number of wind directions simulated. Ultimately, CF greater than 40% were obtained with the ve best performing con gurations and approximately 52% in the best case scenario, suggesting good wind potential. The methodology was capable of lling a major gap in the scienti c literature of micrositing for allowing the layout planning, selection of wind turbine type and hub height, as well as gross production estimates and CF for the wind farm in complex terrain. Additionally, the free and open-source-based framework strengthens the continuous development, sharing and transparency of the methodology.

A comparative study of two different approaches in wind energy simulations has been made where the aim was to investigate the performance of two commercially available tools. The study includes the linear model by WAsP and the computational fluid dynamic model of WindSim (also featuring an additional forest module). The case studied is a small wind farm located in the inland of Sweden featuring a fairly complex and forested terrain. The results showed similar estimations from both tools and in some cases an advantage for WindSim. The site terrain is however deemed not complex enough to manifest the potential benefits of using the CFD model. It can be concluded that estimating the energy output in this kind of terrain is done satisfyingly with both tools. WindSim does however show a significant improvement in consistency when estimating the energy output from different measurement heights when using the forest module compared to only using the standardized roughness length.