I

UNIVERSIDAD DE CONCEPCIN FACULTADA DE INGENIERA AGRCOLA

NATURAL FRECUENCIES ANALYSIS OF A WIND TURBINE BLADE

RODRIGO FUENTES CCERES

TESIS PRESENTADA A LA ESCUELA DE GRADUADOS DE LA UNIVERSIDAD DE CONCEPCION, PARA OPTAR AL GRADO DE MAGISTER EN INGENIERA AGRCOLA, MENCIN MECANIZACIN Y ENERGA.

CHILLAN CHILE 2009

II

NATURAL FRECUENCIES ANALYSIS OF A WIND TURBINE BLADE

Aprobado por:

Gabriel Merino Coria Licenciado en Fsica, Ph. D. Profesor Asociado

Profesor Gua

Cristian Rodrguez Godoy Ingeniero Civil Mecnico, Doctor Profesor Asistente

Profesor Asesor Externo

Emilio Dufeu Delarze Ingeniero Civil Mecnico, Doctor Profesor Asociado

Profesor Asesor Externo

Octavio Lagos Roa Ingeniero Civil Agrcola, Ph. D. Profesor Asistente

Director de Programa

Eduardo Holzapfel Hoces Ingeniero Agrnomo, Ph. D. Profesor Titular

Decano

III

INDEX

Page Summary.................................................................................................. Resumen.................................................................................................. Introduction.............................................................................................. Finite Elements Model...................................................................... Experimental Modal Test.............................. Results............................................................................ Conclusions........................................................................ References............................................................................................... Figures and Tables.................................................................................. 1 3 5 8 9 11 13 14 16

NATURAL FRECUENCIES ANALYSIS OF A WIND TURBINE BLADE

Keywords: Wind Turbine, FEM, Natural Frecuencies.

ABSTRACT This paper studies the influence of the rotor angular speed on the natural vibration frequencies of the blades for a low power wind turbine, due to the stiffness effect produced by the centrifuge force. The wind turbine under study corresponds to a 1 kW wind turbine, which consists of three blades of compound material of fiber glass with epoxy resin, and whose range of angular speed in operation fluctuates from 0 to 600 rpm. The research has considered a finite elements model (FEM) of the blades calculate their natural frequencies at several operational rotating speed, and an experimental modal test for validating the numerical model at zero rotating speed. For the FEM development, SAMCEF software was used, using volume elements for the discretization of the blade. The impact test done on the turbine blade was carried out reproducing the clampling condition of the blade on the turbines rotor in the laboratory, using a Bruel & Kjaer, 8206 Model impact hammer, and a data collector-analysis system PXI, National Instruments, 4472B Card was used to experimentally obtain the natural vibration frequencies of the blade at a zero rotating speed. Researches related to the natural frequencies about wind turbine blades have considered a beam model to simulate the blades on the wind turbine. This

research has focused on high power turbines. Results presented about the stiffness effect of the blades on large turbines due to rotation speed, show a non-significative stiffness effect, which means that the natural frequencies do not significantly change in operation, regarding its resting position. By using the FEM model developed in this study, graphs of the first, second and third natural vibration frequencies of the blades in flapwise mode were obtained, versus the turbines rotating speed where the blades stiffness effect appears, due of the centrifugal force. Results indicate that the first natural frequency, in flapwise bending, varies from 4 to 20 Hz, in a range of speeds from 0 to 600 rpm, that is, an increase of five times the value of natural frequency of the resting blade, regarding its value in operation at 600 rpm.

ANALISIS DE LAS FRECUENCIAS NATURALES DE VIBRAR DE LAS ASPAS DE UNA TURBINA EOLICA

Palabras ndice adicionales: Fotovoltaico, bombeo PV, bombeo de acople directo, monitoreo.

RESUMEN En este trabajo se ha estudiado el grado de incidencia que produce la velocidad de rotacin del rotor sobre las frecuencias naturales de vibrar de las aspas, en un aerogenerador de baja potencia, debido al efecto de rigidizacin que produce la fuerza centrfuga. El modelo bajo estudio, corresponde a una turbina de 1 kW de potencia, que consta de tres aspas de material compuesto de fibra de vidrio con resina epxica y cuyo rango de velocidad de rotacin, en operacin, flucta entre 0 y 600 rpm. La investigacin ha contemplado la generacin de un modelo numrico en elementos finitos de las aspas, para calcular sus frecuencias naturales de vibrar a distintas velocidades angulares de operacin, previo un ensayo modal experimental para validar el modelo numrico, a velocidad de rotacin cero. Para el desarrollo del modelo numrico, se utiliza el programa SAMCEF, para modelacin por elementos finitos, utilizando elementos de volumen para la discretizacin del aspa. El ensayo de impacto realizado al aspa de la turbina, se ejecuta replicando en banco de pruebas, la condicin mecnica de empotramiento del aspa montada sobre el rotor de la turbina y se utiliza un martillo de impacto Bruel & Kjaer, Modelo 8206 y un sistema de

Adquisicin de datos PXI, National Instruments, Tarjeta 4472B para obtener en forma experimental las frecuencias naturales de vibrar del aspa, a velocidad de rotacin cero. Investigaciones relacionadas a frecuencias naturales sobre aspas de turbinas, han considerado un modelo de viga para simular las aspas en la turbina, aunque los trabajos desarrollados se centran en turbinas de grandes potencias. Resultados presentados sobre el efecto rigidizador de las aspas en grandes turbinas, debido a la velocidad de rotacin, muestran un no gran efecto rigidizador, esto significa que las frecuencias naturales no varan significativamente, en operacin, con respecto a su condicin de reposo. Al utilizar el modelo de elementos finitos desarrollado en este estudio, se obtienen grficos de las tres primeras frecuencias naturales de vibrar de las aspas, en modo flapwise, versus la velocidad de rotacin del rotor de la turbina, en los cuales se aprecia el efecto de rigidizamiento de las aspas, producto de la fuerza centrfuga. Se puede apreciar que la primera frecuencia natural, en flexin flapwise, vara desde 4 Hz hasta 20 Hz, en un rango de velocidades desde 0 rpm hasta 600 rpm, esto es, un aumento de cinco veces el valor de la frecuencia natural del aspa en reposo, con respecto a su valor en operacin a 600 rpm. .

INTRODUCTION Due to the Chilean growing energy demand, which increases with the countrys economic development, in addition to the exponential increase in prices of the oil, to obtain electric energy from non-conventional renewable sources is becoming more and more important. Wind power is a good alternative source of non-contaminating renewable energy. Its low power demand is growing, due to the increase of generation costs based on diesel in isolated areas. The big barriers for using wind power have been the high costs of the turbines and the lack of general knowledge in terms of the selection, installation, performance and maintenance of the equipment. Based on this fact, the objective of this study is to contribute with indispensable knowledge for the development of this type of technology at a national level. Wind turbines can be classified as small and large, related to their power. According to [12], we refer to small turbines as those with power less than 10 kW. Those with greater power are considered large wind turbines. According to [9], currently, the majority of the machines that are installed present nominal powers between 0.6 and 5 MW. The objective of small wind turbines is to supply electric energy to residents, farms, or small rural groups, isolated from the electric supply network. The power of these machines usually varies between 0.1 and 10 kW [14]. These wind turbines are much more simple than those used for large scale

generation. In terms of its operating conditions, low power wind turbines have operation speeds that can reach 1100 rpm, according to [14], and its blades length between 1 and 4.5 m, compared with high power turbines, in which the rotating speed does not exceed 60 rpm, according to [11] and [13], and whose blades measure length between 10 and 55 m. These operational characteristics produce a greater rigidity effect of the blades in the small turbines, due to the angular velocity. In the large turbines, a greater rigidity effect of the blades occurs due to their mass and radius. In general, researches done about wind turbines related to the natural vibration frequencies focuses on high power turbines, because of the

prioritary economic interest on the electric energy conversion. Tests done on high power turbines blades show a minimal blades rigidity effect due to the centrifugal force (see Figure 2). In [4], the vibration response of a turbine tower and blades assembling under the action of the stationary wind force was studied. In this case, the blades are modeled as beams fitted to a rotating mass and only consider the bending of one of its axis. Studies done about beams fitted to a rotating mass, as in [1] and [2], where the equations of motion of the beam are represented, considered as a Euler Bernoulli beam, under large rotation displacements (rigid body motion) and with small elastic deformations, like in [5], considering three variables: bending in the most flexible direction (flapwise), bending in the most rigid

direction (edgewise) and the lengthening of the beam (stretching). This way, three partial differential equations about the beam motion are obtained. In [2], these equations of motion are solved by the finite elements method and an analysis of the dimensionless natural vibration frequencies versus the beams dimensionless rotation speed was done on the three previously mentioned variables. As they are dimensionless, these results are applied to beams of any size. In [3], the blade is modeled considering its complete geometry, but only for zero angular speed, for which the influence of the centrifugal force due to rotational speed on the behavior of the blades natural vibration frequency was not considered. In this case, a mathematical model is presented regarding a high power turbine blade (19 m in length) mounted on a rigid mass (See Figure 1). The equations of motion describe small rotations of the mass (hub), small flapwise and edgewise bending, and considers the warp movement of the blade (warping). The mathematical model is solved as a eigenvalues problem and then compared with an experimental modal analysis, with excellent results. As previously mentioned, the stiffness increase that is result of the rotating speed of blades on large size turbines is not considerable. According to [11], it is worth asking what the influence of the rotating speed of the blades on small size turbines is, since the rotation speed are greater, considering its geometry and the material parameters it is made of.

The principal objective of this study is to investigate the degree of influence of the rotors rotating speed on the natural vibration frequencies of the blades, due to the stiffness increase that the centrifugal force produces on a low power wind turbine (1 kW). For this purpose, a numerical model of the turbines blade will be generated, with the SAMCEF software, for modeling by the finite elements method. In addition, the natural frequencies of the turbines resting blade will be determined through an experimental modal test. Then, the numerical model at rotation speed zero will be validated, comparing it with the natural frequencies experimentally obtained, to then determine through the numerical model, the natural frequencies of the wind turbines blade for different operation angular speeds.

FINITE ELEMENTS MODEL (FEM) The modal dynamical analysis, through the finite elements method, was done with the SAMCEF program. In this study the blade is discretized using volume elements, pentahedrons and hexahedrons of first order [6], using a total of 4646 nodes of 3 degrees of freedom, and 2600 elements, with a total of 200 pentahedron elements (7.7%) and 2400 hexahedrons (92.3%). The model was made considering the blade clamped in one of its extremes, as shown in Figure 4, and replicating the real profile of the blade, as observed in Figures 3 and 5. The blade is built of fiber glass with epoxy resin, for which, according to [10] and [8], an elasticity module (Young) of 38600 MPa and a Poissons model of 0.26 were considered. On the other hand, the measured density is

1.9 X 10-9 ton/mm3. This data was used in the simplified construction of the model, where the non-linear elastic orthotropic properties that compound materials generally present were not considered, but the isotropic properties with behavior within a linear elastic range of the materials were, as was estimated in [3], achieving good results. With the model in question, the modal study of the blade was carried out, at a rotating speed zero, where the natural vibration frequencies were obtained, which were then compared with those obtained in the experimental modal test in Table 1. Next, the results obtained from the numeric model are presented, in terms of the three first flapwise bending modes (regarding the Y axis) of the blades vibration. The first vibration mode has a natural frequency of 4.4 Hz and posseses a node on the extreme of the blades clamping, see Figure 6. The second vibration mode has a frequency of 27.7 Hz and posseses two nodes, as observed in Figure 7. The third vibration mode is observed in Figure 8, which has a frequency of 77.4 Hz, with its corresponding three nodes.

EXPERIMENTAL MODAL TEST In the impact test, a Bruel & Kjaer, 8206 Model impact hammer, a Bruel & Kjaer, 4513-001 model accelerometer sensor and a PXI, National Instruments, card 4472B data adquisition card were used, as well as LabView software.

10

The blade was clamped to a base in the laboratory, as obserevd in Figure 9, to simulate its behavior when mounted on the turbines rotor. The modal impact test consisted of hitting the blade, which is measured with the impact hammer, and then the vibratory amplitude of the blades vibration is measured by the accelerometer. This signal is captured by the data acquisition card, and with the LabView software, the frequencies answer function of the vibration was calculated, where the natural vibration frequencies of the blade were obtained [7]. In Figure 10, that answer in the time of the amplitude of the blades vibratory acceleration is observed, due to the impact given in its free extremity. The accelerometer was installed in the clamped side of the blade. The impact given and the blades response of the free - damped vibration can be seen in the waveform, which decreases in amplitude as time passes. In Figure 11, the spectrum in frequencies of the acceleration amplitude of the vibration is observed, in a range of frequencies from 0 to 800 Hz. The several frequencies of which the vibratory wave is composed of can be seen, among which the blades natural vibration frequencies are found. The following natural vibration frequencies were obtained by this impact test, which are compared with those obtained by the numerical model by finite elements, in Table 1. Comparing the blades natural frequencies obtained through the experimental test with those obtained through the numerical model (on SAMCEF), in Table 1, a minimum difference can be seen (less than 8%), in the results of both

11

methods regarding the three first natural frequencies. The dispersion of the values of high frequencies could be caused by a lack of refinement of the mesh (4646 nodes), and by the using elements of first order. However, this does not detriment the study objectives, as they require analyzing the dynamic behavior of the blades for a range of the turbines operational speed, for which, the high frequencies are very outside the range. Therefore, it is not necessary to make any adjustments to the numerical model. With this, the numerical model is validated at zero angular speed and will be used to simulate the behavior of the blades natural frequencies for the turbines several operational speeds, and study its stiffness effect on the blade and the influence on the behavior of its natural frequencies under operational conditions.

RESULTS OF NATURAL VIBRATION FREQUENCIES OF THE BLADES VERSUS ROTATING SPEED The range of rotating speed in normal operation of the turbine is from 0 to 600 rpm. The modal analysis was simulated for rotating speeds of 10, 20, 30, 40, 50 and 60 rad/s, that is, between 95 and 668 rpm. With this data, a graph was created with the behavior of the three first natural vibration frequencies for flapwise mode, versus the rotating speed in rpm (see Figure 12). In this first graph, it can be seen that the behavior of the second and third modes natural frequency are out of the range of turbines frecuencies, 1X, 2X and 3X.

12

For better observation about if there is at some time a resonant zone of one of the two first natural vibration frequencies, in flapwise bending, a graph was made of these two first natural frequencies (in Hz) versus the rotating speed of the turbine (in Hz) in its range of operational speed. It can be observed that the rotating frequency 1X, is out of the range of the second natural vibration frequency. In terms of the first natural vibration frecuency, which at a zero rotating speed is of 4.92 Hz, it can be observed that as the rotating speed increases, it generates an increase of the value of the natural frequency, due to the stiffness effect, unable the 1X to reach this first natural frequency, for which there are no problems of resonance. In terms of the blades frequency, it means 3X, there is a problem around the 2 Hz (120 rpm) of rotation, as the 3X = 6 Hz coincide with the first natural frequency, for which there is a posible problem of structural resonance at this speed (see Figure 13). It must be considered that this condition is given for the blades frequency, when passing in front of the tower tube, for which the excitation force should be transmitted from the tower to the turbines rotor and then to the blades, therefore it is not a direct excitation, which is not very significative.

13

CONCLUSIONS The experimented by [3] is proven in this paper; by developing a modal dynamical model of the blades, the non-linear elastic orthotropic condition of the compound materials can be approximated by a simplified linear elastic isotropic model, with good results, at a zero rotating speed. Under the conditions of the simplifed model previously indicated, it is possible to simulate the behavior of the blades of a wind turbine with the method of finite elements, acheiving a very good approximation by comparing the obtained natural vibration frequencies with the finite elements model (FEM), versus those obtained through an experimental modal test, at a zero rotational speed. In this paper, it can be appreciated that the stiffness effect of the blades due to rotating speed, for small sized turbines, is considerably significative. Graphs are presented of the blades three first natural vibration frequencies, in flapwise mode, versus the turbines rotors rotating speed, in which it can be seen that the first natural flapwise frequency varies from 4 Hz to 20 Hz, in a range of speed from 0 to 600 rpm, that is, an increase of up five times the blades natural frequency while resting, versus its operational value at 600 rpm. Therefore, for the design of low power turbine blades, where rotational speed is high, it is not sufficient to only calculate the natural frequencies with an impact test of the resting blade, rather it also requires a study of their dynamical behavior, to avoid resonances and prolong its lifespan.

14

REFERENCES. [1] H. H. Yoo, R. R. Ryan and R. A. Scott 1995, Journal of Sound and Vibration 181, 261 278. DYNAMICS of FLEXIBLE BEAMS UNDERGOING OVERALL MOTIONS. [2] J. Chung and H. H. Yoo 2002, Journal of Sound and Vibration 249, 147 164. DYNAMICS ANALYSIS of a ROTATING CANTILEVER BEAM by

USING the FINITE ELEMENT METHOD. [3] A. Baumgart 2002, Journal of Sound and Vibration 251, 1 12. MATHEMATICAL MODEL FOR WIND TURBINE BLADES. [4] P.J. Murtagh, B. Basu, B.M. Broderick 2005, Engineering Structures 27, 1209 1219. ALONG WIND RESPONSE OF A WIND TURBINE TOWER WITH BLADE COUPLING SUBJECTED TO ROTATIONALLY SAMPLED WIND LOADING. [5] J. C. Simo 1985, Computer Methods in Applied Mechanics and A

Engineering 49, 55 70. A FINITE STRAIN BEAM FORMULATION. THE THREE DIMENSIONAL DYNAMIC PROBLEM. PART I. [6] T. Chandrupatla, A. Belegundu 1999, INTRUDUCCION AL ESTUDIO DEL ELEMENTO FINITO EN INGENIERIA. Prentice Hall. [7] Peter Avitabile 2001, Sound and Vibration 35, 20 31. EXPERIMENTAL MODAL ANALYSIS. A SIMPLE NON MATHEMATICAL PRESENTATION. [8] C.Kong, J. Bang, Y. Sugiyama 2005, Energy 30, 2101 2114. STRUCTURAL INVESTIGATION OF COMPOSITE WIND TURBINE BLADE CONSIDERING VARIOUS LOAD CASES AND FATIGUE LIFE.

15

[9] J. W. Larsen 2005, Tesis for the degree of Doctor of Philosophy. Department of Civil Engineering, Faculty of Engineering and Science, Aalborg University. NONLINEAR DYNAMICS OF WIND TURBINE WINGS. [10] R. F. Gibson 1994: PRINCIPLE OF COMPOSITE MATERIALS MECHANICS. Mc Graw-Hill Ed. p. 1. [11] R. Gasch, J.Twele 2002, WIND POWER PLANTS. Fundamentals, Design, Construction and Operation. [12] J.F. Sanz 2001. GUIA DE LAS ENERGIAS RENOVABLES APLICADAS A LAS PYMES. [13] J. Wiley and Sons Ltda. 2001, WIND ENERGY HANDBOOK. [14] M. Sagrillo 2002, APPLES AND ORANGES CATALOGUE

16

Figure 1. Motion Sequence of the Blade in its Second Flapwise Mode. Three times in the same figure are shown.

17

Figure 2. Graph of the Primary Natural Vibration Frequencies versus Rotation Velocity of the Blade of a Turbine. Turbine Test DEBRA (DLR) [15].

18

Figure 3. Profile of the Bergey XL1 turbine blade, of 1 kW. Its clamped condition to the rotor of the turbine can be seen.

19

Figure 4. Model of turbine blade done on SAMCEF. The fitted condition of the model is marked the red oval.

20

Figure 5. Profile of the Bergey XL1 turbine blade, model done on SAMCEF.

21

nodo

FIGURE 6. First vibration mode of the blade in flapwise bending.

22

nodo 1

nodo 2

FIGURE 7. Second vibration mode of the blade in flapwise bending.

23

nodo 1

nodo 2

nodo 3

FIGURE 8. Third vibration mode of the blade in flapwise bending.

24

Figure 9. Impact test with the blade bolted to its base, such as the turbines rotor.

25

Figure 10. Response of the vibratory acceleration amplitude (g) in time, of the turbines blade, when is hit with the hammer.

26

Figure 11. Response in frequencies (from 0 to 800 Hz) of the vibratory acceleration amplitude (g), of the turbines blade, when is hit with the hammer.

27

120

100

80 fn 1er modo flapwise fnatural, Hz. fn 2o modo flapwise 60 fn 3er modo flapwise 1X 2X 3X 40

20

0 0 100 200 300 400 500 600 700 800 velocidad de rotacin, rpm.

Figure 12. Graph of the three primary natural vibratuon frequencies, in flapwise mode versus rotation velocity in rpm of the turbines blade.

28

45

40

35

30 fn 1er modo flapwise fn 2o modo flapwise 1X 20 2X 3X 15

fnatural, Hz.

25

10

0 0,00

2,00

4,00

6,00

8,00

10,00

12,00

velocidad de rotacin, Hz.

Figure 13. Graph of the two primary natural vibration frequencies, inflapwisemode, versus rotating speed, in Hz, of the turbines blade

29

Modal Impact Test FRF Flapwise Bending Mode 1 2 3 4 5 6 7 Frequency (Hz) 4.2 26.2 72.0 138.4 214.6 315.6 402.4

Numeric Model FEM Deviation Percentage Frequency (Hz) 4.4 27.7 77.4 151.1 248.4 368.6 510.6 % 4.8 5.7 7.5 9.2 15.8 16.8 26.9

Table 1. Comparison of the natural frequencies of the first seven modes of flapwise bending, obtained with the numerical model FEM and the modal test FRF.