8th International DAAAM Baltic Conference INDUSTRIAL ENGINEERING 19-21 April 2012, Tallinn, Estonia MODEL VALIDATION AND STRUCTURAL ANALYSIS OF A SMALL WIND TURBINE BLADE Pabut, O.; Allikas, G.; Herranen, H.; Talalaev, R. & Vene, K. Abstract: The goal of this study was to develop and validate a simplified finite element analysis (FEA) model for a glass fibre reinforced plastic (GFRP) wind turbine blade. A 3D virtual model of the blade was built up in ANSYS Workbench software with corresponding load cases and boundary conditions. Data for the FEA was obtained from tensile tests of GFRP laminates. Experimental validation of the virtual model was performed on manufactured blade subject via bending tests and modal analysis. Key words: wind turbine blade, GFRP, finite element analysis, modal analysis 1. INTRODUCTION One of the most important components of a wind energy converter is the rotor together with blades. Structural design and performance analysis of wind turbine blade is an important part of the design theory and application of wind turbines [1-2]. Manufacturing costs of a small horizontal axis wind turbine (SHAWT) blade can reach about 20% of the turbine productions costs. Therefore, possible profits resulting from a better structural model and use of suitable composite materials refer to a need of multi-criteria optimization and refined modeling techniques [3]. These statements are furthermore reinforced by the fact that for a cost effective wind turbine solution, the blades must achieve a very long operating life of 20-30 years [4]. For SHAWT blades, only few full scale blades are normally physically tested during the design process as a confirmation of general structural integrity. By utilizing computational technology and various design approaches, information from these tests can be taken into account in a rational way in order to speed up the design process and increase structural reliability of blades [5]. In this paper, validation of a full scale single layer lay-up SHAWT blade 3D FEA model is performed. The virtual model is validated through experimental bending test and modal analysis. The proposed model allows effective consideration of GRFP material properties, large dimensions of the structure and rapidly changing geometrical variables [6-7]. 2. PROBLEM FORMULATION 2.1 Blade structure The SHAWT blade subjected to analysis consist of three main structural components that are presented in the Figure 1. A C Fig. 1. GFRP blade: geometrical layout GFRP skin material is denoted by A, resin based connection part by B and metallic root tube by C. The blade has an overall B
length of 2850 mm and a maximum width, at the widest chord length, of 465 mm. The manufacture skin laminate consists of following E-Glass fiber materials: - one layer of Gelcoat GS with 0,4 mm thickness. - one layer of 600 g/m 2 Chopped Strand Mat (CSM) with 2 mm thickness; - four layers of 600 g/m 2 balanced stitched biaxial roving mat 0º/90º with 4 0,5 mm thickness. The fibers were impregnated via hand layup with polyester resin (413-568). This resulted in a 4,4 mm thick shell structure, after being post-cured at a room temperature. The skin in this case is the main load bearing element as it is subjected to aerodynamic thrust loads. The connection part B consists only of polyester resin which is formed in a separate mold. It is mainly used as a load transferring unit from the skin to the root tube element. The root tube C is made of conventional S355 steel and acts as a fixing unit. 2.2 Goals of model validation In order to achieve preliminary safety of the blade structure, it has to possess properties that do not lead to catastrophic failures under ultimate loads. Therefore, it becomes necessary to be able to predict following properties of the blade: maximum allowed loads, bending stiffness and natural frequencies [8, 9]. The natural frequencies are important as the blade is of an elastic structure and the load on it is of stress alternation and random variability. This could lead to coupling modes and often to direct malignant structure damage and failure [9]. 3. EXPERIMENTAL STUDY 3.1 Stiffness and strength analysis For an experimental study, the blade subjects were manufactured according to previously mentioned parameters. Stiffness and strength analysis was performed in a manner where calibrated weights were attached to a designated point at the blade tip. The deformation was acquired with a laser measurement device as a vertical tip displacement from initial position. The test was carried out until a full loss of blade structural integrity occurred. Results of the experimental test are presented in Table 1. The applied force is weighed against vertical deformation of the blade tip. Nr. Load [kg] Load [N] Def. [mm] 1 20 196,2 86 2 25 245,3 116 3 30 294,3 141 4 35 343,4 169 5 40 392,4 196 6 45 441,5 234 7 50 490,5 261 8 55 539,6 293 9 60 588,6 322 10 65 637,7 364 11 70 686,7 402 12 75 735,8 483 13 80 784,8 failure Table 1. Experimental results of the stiffness and strength analysis Fig. 2. Broken test subject At a load of 784,8 N the test specimen experienced a full loss of structural integrity as the bottom side of the skin collapsed under compressive load (Figure 2). Also opening of the two skin halves due to shear forces was noted during the failure incident.
3.2 Modal analysis The experimental modal analysis was performed in a form of calibrated impulse hammer (Model AU01) test. The test specimen was mounted with soft restraints. Predefined measurement points were excited to acquire the frequency response functions (FRF). Results were recorded with SigLab Model 20-22A. Fig. 3. Experimental modal analysis 4. FINITE ELEMENT ANALYSIS 4.1 FEA model ANSYS Workbench software was utilized for numerical analysis. For the skin hexahedral shell elements (SHELL181) with four nodes were considered. These are well-suited for linear, large rotation, and/or large strain nonlinear applications [10]. Due to the complex and irregular geometry solid tetrahedral (SOLID187) elements were considered for the connection part. The root tube was meshed with solid hexahedral (SOLID186) elements. Three different element side lengths were considered according to the general importance of the objects. Skin was meshed with 10 mm element side length, connection part with 20 mm side length and root tube with 6 mm side length (Figure 5). The measurement data was processed in STAR Modal software. The obtained FRFs where iteratively curve fitted with polynomial method to identify the modal parameters (Figure 4). Results of first five natural frequencies are presented in Table 2. Mode Frequency, Hz 1 16,6 2 46,8 3 82,1 4 95,4 5 125,7 Table 2. Results of experimental analysis Fig 4. FRF of a 10 th measurement point Fig. 5. Mesh for FEA For the connector and root tube linear isotropic material models were used (defined by elastic modulus E and Poisson s ratio υ). Linear orthotropic material model was used to define the skin properties (defined by elastic modulus E, Poisson s ratio υ and shear modulus G). The skin lay-up was depicted as one layer to simplify the model and reduce calculation time. Properties of the 4,4 mm layer were determined via tensile test with a servo hydraulic test machine Instron 8800. Material properties were calculated according to methods mentioned in previous studies [11].
E-Glass singel layer lay-up, GPa E x 14500 E y 14500 E z 1800 υ xy 0.11 υ yz 0.30 υ xz 0.30 G xy 2800 G yz 1500 G xz 1500 Table 3. Mechanical properties of GFRP laminate 4.2 Stiffness and strength analysis Boundary conditions for the strength and stiffness analysis were selected according to the test setup. The principle scheme of the stiffness and strength analysis is presented in the Figure 6. Fig. 6. Boundary conditions for the stiffness and strength analysis In Figure 6 the fixed support of the root tube is denoted by A. The tube is supported along 200 mm area, starting from the free end of the tube. Acting force is denoted by B and applied to a 50 mm area, measured 2675 mm from the free end of the root tube. Standard earth gravity (9,806 m/s 2 ) is denoted by C. During the simulation force magnitude was incrementally raised to generate a forcedisplacement curve. Due to the relatively large deformations compared to the initial bounding box, ANSYS large deformation mode was used. This enabled to include non-linear material effects and utilized multiple time step iterations in order to achieve a more refined solution. However, it must be noted that the usage of large deformation mode increases considerably calculation time and effort. Results of the stiffness and strength analysis are presented in Table 4. The applied force is weighed against resulting vertical deformation of the blade tip and maximum stress on the skin. Each load case can be traced by corresponding number. Nr. Load [N] Stress [MPa] Def. [mm] 1 196,2 33 86 2 245,3 40 107 3 294,3 46 128 4 343,4 53 149 5 392,4 60 171 6 441,5 66 195 7 490,5 73 220 8 539,6 79 248 9 588,6 89 285 10 637,7 139 337 11 686,7 221 410 12 735,8 336 503 13 784,8 447 607 Table 4. FEA results of the strength and stiffness analysis 4.3 Modal analysis The essence of modal analysis is solving the vector of the modal equations with a finite number of degrees of freedom under the non-damping and non-external load condition. The impact of structural damping to the modal frequency and the vibration mode is so small that it is ignored [9]. Modal analysis with two different sets of boundary condition was carried out. Firstly, a free body analysis without supporting fixtures was conducted. This method allows to discard the stiffness of restraints and therefore exhibits fewer degrees of uncertainty. Secondly, a fixed body analysis was carried out to predict the natural frequencies of the blade in its working position. In this configuration the end of
the root tube had all degrees of freedom removed (Figure 7). Fig. 8. First natural mode of the blade 5. RESULTS Fig. 7. Boundary conditions for fixed body modal analysis Results of the modal analysis are presented in Table 5. It has to be noted that for a free body analysis the first six natural frequency modes are always 0. These represent the unfixed 6 degrees of freedom of a rigid body. Free body anal. Fixed body anal. Mode Freq., Hz Mode Freq., Hz 1 0,0 1 7,2 2 0,0 2 21,4 3 0,7E-03 3 27,1 4 0,9E-03 4 54,8 5 0,2E-02 5 77,3 6 0,2E-02 7 15,5 8 43,2 9 78,2 10 82,3 11 91,1 Table 5. FEA results of the modal analysis For the fixed body analysis the first natural mode is found at 8,3 Hz. It is a bending mode of the blade tip and is depicted in Figure 8. The obtained FEA deformations are in good correlations with the experimental results. The maximum difference of 16,8% occurs during load case 6 (Figure 9). The stress figures predict structural failure of the blade already at load case 12, as the obtained stress is above the yield strength of the skin element (Rm=250 MPa). This can be explained by a rapid change of the blade geometry that leads to a singularity effect which results in a higher stress concentration factor. 500 400 Displacement, mm 300 200 100 Exp FEA 0 150 350 550 750 Force, N Fig. 9. Deformation comparison of FEA and experimental anlysis The obtained FEA modal frequencies provide satisfactory correlation of the result for first three frequencies with a differences of 7%, 8% and 5% respectively. However, for higher frequencies the deviation also increases. This can be explained by the fact that higher frequencies are more influenced by structural deviations. In current study the higher frequencies are not of concerne as their influence for structural vibrations is
relatively low and usually for turbine blades only first and second natural frequencies are of interest. Therefore, in general it can be concluded that the FEA model is validated against experimental results. 6. CONCLUSION FEA model validation of a single layer layup GFRP wind turbine blade has been performed. Experimental bending test and modal analysis have confirmed the results achieved in the ANSYS model. It can be concluded that: a) Simplified representation of the skin layout in FEA can be used to study the stiffness, and strength characteristic of the blade. b) Simplified representation is suitable to study the lower natural frequencies of the blade. In future studies different connections interfaces have to be examined in greater detail. It is planned to develop design principles for the root tube-resin and skinglue interfaces. Also a study of fatigue properties of the blade shall be carried out. 7. REFERENCES [1] Song, F., Ni, Y., Tan, Z. Optimization Design, Modeling and Dynamic Analysis for Composite Wind Turbine Blade. International Workshop on Automobile, Power and Energy Engineering, 2011, 16, 369 375. [2] Ronold, K.O., Cristensen, C.J. Optimization of a Design Code for Wind-Turbine Rotor Blades in Fatigue. Engineering Structures, 2001, 23(8), 993-1004. [3] Jureczko, M., Pawlak, M., Mežyk, A. Optimization of a Wind Turbine Blades. 2005 Int. For. on the Adv. in Mat. Proc. Tech, 2005, 167(2-3), 463-471. [4] Kong, C., Kim, T., Han, D., Sugiyama Y. Investigation of Fatigue Life for a Medium Scale Composite Wind Turbine Blade. The Third Int. Conf. on Fat. of Comp., 2006, 28(10), 1382-88. [5] Toft, H.S., Sorensen, J.D. Reliabilitybased Design of Wind Turbine Blades. Structural Safety, 2011, 33(6), 333-342. [6] Pohlak, M., Majak, J., Küttner, R. Multicriteria Optimization of Large Composite Parts. Composite Structures, 2010, 92, 2146-152. [7] Kers, J., Majak, J. Modeling a New Composite From a Recycled GFRP. Mechanics of Composite Materials, 2008, 44(6), 623-632. [8] Chen, C.P., Kam, T.Y. Failure Analysis of Small Composite Sandwich Turbine Subjected to Extreme Wind Load. The Proc. of the 20 th East Asia-Pacific Conf. on Str. Eng. and Const., 2011, 14, 1973-981. [9] Yanbin, C., Lei, S., Feng, Z. Modal Analysis of Wind Turbine Blade Made of Composite Laminated Parts. Pow. and En. Eng. Conf., 2010 Asia-Pacific, 2010, 1-4. [10] ANSYS, Inc. ANSYS 14.0, Manual, 2011 [11] Herranen, H., Pabut, O., Eerme, M., Majak, J., Kers, J., Saarna, M., Allikas, G., Aruniit, A. Design and Testing of Sandwich Structures with Different Core Materials. Jour. of Mat. Scn. of Kaun. Un. Of Tech, 2011, 17(3), 276-281. 8. ADDITIONAL DATA ABOUT AUTHORS MSc. Ott Pabut TUT, Department of Machinery Ehitajate tee 5, 19086 Tallinn, Estonia Phone: 372+51 644 57, E-mail: ottpabut@hotmail.com http://www.ttu.ee