Estudio del comportamiento no gaussiano de las cargas aerodinámicas en aerogeneradores

Luna González, Ana (2021). Estudio del comportamiento no gaussiano de las cargas aerodinámicas en aerogeneradores. Thesis (Master thesis), E.T.S.I. Industriales (UPM).


Title: Estudio del comportamiento no gaussiano de las cargas aerodinámicas en aerogeneradores
  • Luna González, Ana
  • Muñoz Paniagua, Jorge
Item Type: Thesis (Master thesis)
Masters title: Ingeniería Industrial
Date: June 2021
Freetext Keywords: Wind, Gaussian distribution, Non Gaussian distribution, wind profile, speed, direction, turbulence intensity, wind turbine loads, atmospheric stability, statistical moments
Faculty: E.T.S.I. Industriales (UPM)
Department: Ingeniería Energética
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (21MB) | Preview


The design of wind turbine is a highly developed technology since over the years there has been a great progress in materials science, aerodynamics and rotor designs. However, less importance has been given to other aspects related to wind behavior, thus, it has been assumed that the wind turbulence responds statistically to a Gaussian distribution. This assumption of the wind behavior allowed to simplify the mathematical modelling of the design of wind turbine components, which were a priori complex, making easier their handling. However, this mathematical simplification is of great importance when designing a wind turbine, since it is the basis for estimating the loads of the structure, in particular, it influences the fatigue that the turbine faces. The development process is characterized with a high degree of maturity, therefore, it is more difficult to advance technologically. For this reason, it is necessary to open new fields of research that until now had been considered less relevant for the design of wind turbines, such as the study of the statistical model to which wind turbulence responds. Consequently, this project does not focus its purpose on innovating the aerodynamic surface, improving the composites of the structure or designing a new rotor, but it improves the knowledge about the behavior of the wind and how it affects the structure.This will improve the prediction process and therefore the efficiency of the energy that can be obtained from the wind and the efficiency of the turbine itself. Thus, the main objective of this project is to study the real distribution that follows the wind turbulence, since up to now, when designing a wind turbine it is assumed that its behavior responds to a Gaussian distribution. This mathematical simplification has important consequences in the estimation of loads of the structure, in particular on fatigue loads since it constrains its design and thus its optimization. As this research has an impact on such primary elements of the design of wind turbines, but highly constraining elements of the capacity to harness wind energy, it can be of great importance in the development of new generations of wind turbines, more efficient and capable of optimizing the use of these energies, both in places where they are already installed, as well as in those locations that until now were considered useless. The project presented here begins with a brief introduction to the subject, where recent studies are summarized and an initial introduction to wind and its characteristics is given. It also introduces and characterizes the turbine that has been used as a base for the study, as well as the place where it is located. In addition, since the data has been collected with two different anemometers, their characteristics and differences are presented, and the delay between both devices is calculated so that in later studies the measurements can be accurately compared. These two devices are a meteorological mast located 130 meters from the turbine in the direction of 289◦ and a fourbeam lidar mounted on top of the nacelle. It should be recalled that all the data that has been used, has been downloaded from the server of MySQL from the data base denominated licoreim. Here, depending on the variables that wanted to be analyzed, the restrictions were different, although some filters remains always constant in order to assure the well function of the device and avoid wrong data. Once an introduction to situate the study as well as the previous considerations and the conditions for its realization were clarified, the theory needed for understanding the results is presented. In addition, the process that has been followed to analyze the samples and the theore tical background is also exposed. The objective was to study the turbulence and the fluctuations that composed the wind, where the variable denominated as increment is defined as the difference between two points separated a specific time. With this time series, the probability density function is calculated and the distribution of the wind is represented and examined. Due to the fact that wind depends on many parameters, several analysis were performed, where each time a different variable is analyzed independently. Here, the speed, direction, shear, and turbulence intensity has been analyzed first as independent factors, then as a whole. Their analysis has been presented on the discussion and results section. Moreover, on this section, a comparison between the lidar and the mast is carried out. The main purpose is to study the advantages and disadvantages that each device present, as well as to determine which one is more accurate to conduct the turbulence study. Furthermore, a deep analysis of the loads is realized. The main purpose is to demonstrate that those samples which fit into a Gaussian distribution are characterized with lower loads if compare with the winds that do not follow it. For that, all the parameters previously studied individually, are now studied as a whole, where the turbulence intensity that the samples had, played an important role. Finally, once it was demonstrated that the wind is characterized with a non Gaussian distribution, a search is conducted for other types of functions that could fit better with the actual data. For doing so, first the Gaussian type is analyzed along with the characteristics that it should have. Once several functions are chosen, the analysis is carried out both graphically and statistically and a conclusion is achieved. To end up the project, all the conclusions are collected and summarized on the conclusions section and several ideas for future projects are proposed

More information

Item ID: 67968
DC Identifier:
OAI Identifier:
Deposited by: Biblioteca ETSI Industriales
Deposited on: 24 Jul 2021 16:16
Last Modified: 25 Sep 2021 22:30
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM