Finite element simulations for investigating the strength characteristics of a 5 m composite wind turbine blade

Full-scale structural tests enable us to monitor the mechanical response of the blades under various loading scenarios. Yet, these tests must be accompanied by numerical simulations so that the physical basis of the progressive damage development can be better interpreted and understood. In this work, finite element analysis is utilized to investigate the strength characteristics of an existing 5 m RÜZGEM composite wind turbine blade under extreme flapwise, edgewise and combined flapwise plus edgewise loading conditions. For this purpose, in addition to a linear buckling analysis, Puck’s (2D) physically based phenomenological model is used for the progressive damage analysis of the blade. The 5 m RÜZGEM blade is found to exhibit sufficient resistance against buckling. However, Puck’s damage model indicates that laminate failure plays a major role in the ultimate blade failure. Under extreme flapwise and combined load cases, the internal flange at the leading edge and the trailing edge are identified as the main damaged regions. Under edgewise loading, the leading edge close to the root is the failure region. When extreme load case is applied as a combination of edgewise and flapwise loading cases, less damage is observed compared to the pure flapwise loading case.

In the literature, many studies on the structural behavior of composite turbine blades have been published in the past two decades with most of the studies conducted for large wind turbine blades. Jensen et al. (2006) carried out full-scale tests and 30 nonlinear FE simulations of a 34-meter composite wind turbine blade under flap-wise loading. In both the tests and the simulations, they noted that spar cap (suction side) deflected nonlinearly and observed cross-sectional ovalisation, i.e. Brazier effect, arguing that Brazier effect is the main reason for the blade collapse. Overgaard et al. (2010) reported on testing and numerical analysis of the collapse of a 25-meter composite wind turbine blade subjected to static flap-wise loading. In the study, in contrast to Jensen et al. (2006), the main failure mechanism which leads to blade collapse is argued to be delamination 35 and its interaction with local buckling. It is stated that nonlinear strain behavior, i.e. local buckling is triggered by geometrical imperfections. According to Overgaard et al. (2010) delamination and local buckling occur prior to the Brazier effect observed in spar caps. Once the blade is weakened by the aforementioned failure mechanisms, blade collapses due to the compressive strains in fiber directions. Yang et al. (2013) investigated the structural collapse of a 40 m blade and based on full-scale test results showed that debonding 40 was the root cause of rotor blade collapse. Kim et al. (2014) studied the structural response of a Multi-MW class wind turbine blade using Puck's 3D damage model and linear buckling analysis. According to the results, blade shows sufficient structural strength and resistance against buckling. When laminate failure is of concern, the major weak point of the blade is located on the skin at the maximum chord. In another study, Chen et al. (2014) conducted a full-scale bending test of a 52.3 m wind turbine blade. Delamination in the spar cap and shear web failure at the root transition region were found to be the main failure 45 mechanisms for the blade collapse. Local buckling contributed to the main failure mechanism by facilitating local out-of-plane deformation. They conclude that for large blades, through-the-thickness stresses which cause debonding and delamination at the blade root transition region need to be considered in the FEA. In another study, Chen et al. (2017) stated that 3D stresses/strains are important in the failure of a 52.3-meter blade and recommended use of solid elements in the FE simulation when debonding failure is of concern. 3-D strains and Yeh-Stratton failure criterion were used to calculate delamination and 50 debonding failures in the blade utilizing sub-modeling technique. Haselbach and Branner (2016)  blades. They point out that the superposition of stresses for the fatigue may be misleading for modern, flexible rotor blades 60 where geometric nonlinearities must be considered. In addition to this, they propose a novel methodology for calculating stresses with a new load application method that reduces geometric nonlinear behavior of the blade. Montesano et al. (2016), state that progressive failure models incorporating failure criteria do not consider the progressive nature of sub-critical microscopic intra-laminar damage, which is vital for predicting the onset of macroscopic failure modes. Therefore, in the study a physically based multi-scale damage model, which can account for failures of a 33.25 m rotor blade 65 under quasi-static and fatigue loading is introduced. The simulation results show the capability of the model to predict the evolution of sub-critical ply cracks between spar webs located near blade root at maximum chord length. The capability of the model to show damage evolution in the early stages of the progressive damage analysis is important for increasing damage tolerance accuracy and structural health monitoring. In the follow up work by Zu et al. (2018), the multi-scale model is further expanded to include cohesive zone elements to predict structural debonding failure at the spar/skin interface located near blade 70 root at maximum chord.
In contrast to large wind turbine blades, fewer number of investigations on the structural analysis of small wind turbine blades exists. Chen et al. (2015) focused on the local buckling resistance of 10.3 m wind turbine blades. FE analysis showed that configurations with sharp edges are susceptible to local buckling. During testing of the 10.3 m blade, although local buckling of shear web and flatback airfoil was observed, composite laminate failure in these locations was not observed. These results 75 show the possibility of different mechanisms for different blade sizes. In another study, to improve the structural strength of a small size 9-m wind turbine blade, Paquette and Veers (2007) carried out a blade system design study (BSDS), where structural innovations such as flatback airfoils, thick root diameter and carbon spar caps were introduced and their advantages were demonstrated. The static strength of the blades was determined by measuring strains to failure by tests and using finite element analysis. Moreover, linear buckling analysis of the blades were conducted. 80 Fagan et al. (2016) presented a new iterative design process and utilized failure criteria to check the structural strength of different blade designs under various load cases. Besides this, Fagan et al. (2017) utilized failure criteria for the structural design optimization of a 13m glass-fibre epoxy composite wind turbine blade. In the study, experimental testing was used to calibrate finite element models. In these studies, Puck's damage model is used to determine the most suitable composite turbine blade design in terms of its structural behavior. 85 Within the framework of this study, the authors of this study refer back to their previous paper concerning the strength analysis of an existing 5-meter GFRP turbine blade using Puck failure criteria (Ozyildiz et al., 2018). As a part of the previous study, the linear Puck material model was compared with the progressive damage model (Puck). They concluded that progressive failure analysis is necessary to capture a more realistic simulation of failure mechanisms prior to testing.
The scope of this work is limited to the investigation of the structural response of an existing 5-meter wind turbine blade using 90 global finite element modeling approach and progressive composite failure analysis. Hence, using the current modeling technique with shell elements, critical locations for failure and worst load case scenario are identified. Puck's 2-D damage model demonstrates the direction to proceed for a complete and comprehensive modeling of the failure mechanisms. Furthermore, differences between edgewise, flap-wise and combined flap-wise/edgewise loading conditions are discussed. In combined edge-wise and flap-wise loading, less damage is observed compared to the pure flap-wise loading case. 95 The existing blade investigated in this work was designed as part of a joint-project between Core Team of the University of Patras and METUWIND (RUZGEM) -METU Center for Wind Energy. The blade was designed for a wind turbine that has 30 kW nominal power capacity at 10 m/s wind speed. According to the wind turbine characteristics, optimized aerodynamic blade design was finalized by the blade manufacturer. The existing blade consists of five main parts: suction side, pressure side, internal flange, "hat-shaped" chassis/spar, and flange, as seen in Figure 1. 100 Figure 1. Blade assembly for the 5-meter RUZGEM turbine blade (Philippidis and Roukis, 2013).

Methodology
For the progressive failure analysis of the blade, Puck criteria explained briefly in the following paragraphs, are used. Puck failure criteria (Puck and Schuermann, 1998) are one the most commonly used and well-established criteria for the assessment 105 of composite laminate strength. In this study, Puck's failure criteria are implemented for the evaluation of stress results of unidirectional and tri-axial composite laminates.
For fiber failure, Puck's criteria are as follows: (1) where ( ) and ( ) are stress exposures for fiber failure under tension and compression loading cases. 1 is the stress 110 value in fiber direction, and are tensile and compressive strengths in fiber direction, respectively. Puck's inter-fiber failure uses different equations depending on the failure mode, which is detected. Under two dimensional (2-D) biaxial loading, the failure modes, which can be detected, are summarized in Figure 2 below. In Figure 2 the transition point from failure mode B to failure mode C is denoted by the point ( ⊥⊥ , 21 ) and is calculated by the ratio ⊥⊥ / 21 . Their values are calculated by the expressions below: 115  (Knops, 2008).
Depending on the region of the failure envelope, the following inter-fiber failure expressions are written: 120 In the equations above ⊥ (+) , ⊥ (−) and ⊥⊥ (−) represent inclination parameters that control the shape of the failure envelope.
According to Puck and Schuermann (1998), ⊥ (+) = 0.3 and ⊥ (−) = 0.25 are chosen for the GFRP material. 2 is the stress value in the transverse fiber direction, and are tensile and compressive strengths in the transverse fiber direction. Shear stress and shear strength are represented by 6 and S, respectively. If the value of stress exposure (fE) exceeds 1, failure initiation occurs. Mode A is caused by tensile and shear stresses. Modes B occurs under compressive and shear stresses. Mode 125 C is a dangerous failure mode in compressive shearing, which may lead to delamination.
Degradation rules are applied to the elements which fail according to the specific Puck's failure criteria that are inter-fiber failure (IFF) mode A, B, or C (Eq. (6), (7) and Eq. (8), respectively). As presented by Passipoularidis et al. (2011), based on degradation rules in Table 1, transverse elasticity and shear moduli of the damaged elements are reduced accordingly.  In Eq. (8) is known as the degradation factor and can be expressed according to the equation below: 135 The summary of the algorithm of the FE analysis based progressive failure analysis of a composite laminate using Puck failure criteria is shown in Figure 3. The complete algorithm is implemented using ANSYS Parametric Design Language (APDL).
First, using ANSYS APDL script, different material numbers are given to each lamina, which constitutes a layer of a composite laminate. This step is necessary because, during the execution of the progressive damage propagation, each lamina is subjected to different degradation rules. Then, an extreme load case is applied incrementally to the model, and static analysis is run. 140 Afterward, in the post-processing module, stresses are read. From Puck failure criteria for FF (Eq. (1) and (2)) and IFF (Eq. (6), (7) and (8)), stress exposures are calculated. Depending on the rules presented in Table 1whether ply failure happens is checked. If ply failure occurs due to First-Fiber-Failure (FFF) or if IFF(C) is observed in at least one third of the plies of a laminate, laminate failure is assumed to take place. If IFF A or B or IFF C in any ply is seen, gradual degradation rules are applied, and after the assembly of the new constitutive material model, the load is increased, and the analysis is re-run. This 145 calculation procedure is run until the solution does not converge. If the solution does not converge, the analysis aborted. As seen from the flow chart, as long as no FF or IFF failure occurs, without updating the constitutive material model, load is incremented, and the analysis is executed.

Finite Element Model of the RUZGEM Blade
The 2D blade technical drawings, which include the blade aerodynamic design details such as chord length and twist angle along the blade, were provided by the blade manufacturer Compblades. By using these given 2D blade drawings, the 3D CAD 160 model of the blade is prepared in NX 12.0 environment. In Table 2, the material properties and design allowables of the blade materials for static analysis is listed (Philippidis and Roukis, 2013). Referring to Germanischer Lyod (2010), design allowables are obtained from the knockdown of the experimental strength values by the material safety factor 2.406.

Table 2. Material properties and design allowables of RUZGEM blade for static analysis
The skin of the blade is composed of unidirectional and tri-axial laminates, whereas only tri-axial laminates are used for the spar. The lay-up sequence for the pressure and suction side differs only in the area from 1.25m to 2.0m, where an extra 170 unidirectional glass fabric was placed in the suction side of the blade. The root part of the blade is composed of unidirectional laminate, tri-axial laminates, and steel. The outer surface of the blade is covered with transparent Gel Coat and a layer of chopped strand mat, 300 g/m 2 CSM 300. In addition to this, the Divinycell H45 foam used in the trailing edge is of 10 mm thickness in the area from 0.7m to 2.0m and 5mm thickness from 2.0m to 3.0m. Since the gel coat, CSM 300, and foam do not have a significant contribution to the strength of the blade; these materials are not included in the finite element model. The 175 details of the material lay-up and geometry are given in the supplement.
After geometric modeling of the blade, the material model of the blade structure is prepared in Ansys ACP/Pre module. Plane stress SHELL 181 quadrilateral elements are used to mesh the blade entirely in ANSYS Workbench, as seen in Figure 5 For the design of the RUZGEM blade, the turbine specifications are obtained from the meteorological data in Ankara, Turkey (Weinzierl and Pechlivanoglu, 2013). These were analyzed so that average wind speed, the occurrence of gusts, and wind speeds are determined. Based on this information, the turbine specifications were selected according to British Standard IEC 200 61400-2 standard (2006). Loads for the structural design were selected as the worst load case scenario for the complete set of IEC extreme loads provided from aero-elastic simulations performed at Smart Blade GmbH (Weinzierl and Pechlivanoglu, 2013).
Mixture of different Design Load Cases (DLCs) and time instances are used for this study and moments are extreme for all positions along the blade. The extreme loads were computed using the wind turbine aero-hydro-servo-elastic software tool 205 FAST version v7.01.00a-bjj. During the simulations, the turbine is simulated as a stall regulated constant speed turbine at 83 rpm with a gearbox and simple induction generator. Using this input, the blade is analyzed under extreme loads in the flapwise and edge-wise directions. Loads are calculated at 28 stations along the blade span direction.
According to the recommendations of IEC 61400-23:2002 (2015)    In order to decide for the reasonable mesh density mesh convergence study is conducted. Mesh convergence is shown for total displacement at blade tip under extreme flap-wise loading and first eigenfrequency as seen in Figure 12   This section begins with a linear buckling analysis, followed by the progressive failure analysis of the blade under extreme flap-wise, edgewise and combined loading conditions. This section is followed by a discussion regarding the comparative study of these three loading scenarios.

Buckling Analysis 265
Linear buckling analysis of the blade is performed in order to investigate its buckling resistance and location of buckling eigenmodes. The results are depicted in Figure 13, which shows the buckling modes of the blade under (a) 100% flapwise(max) loading case (Figure 12a) 100% edgewise(min) loading case (Figure 12b) and 100% combined edgewise(min) and flap-wise(max) loading (Figure 12c). Negative buckling factors correspond to the loads applied in the opposite direction, because no critical eigenvalue could be found in the load application direction. In other words, the blade exhibits sufficient 270 buckling resistance in the load application direction for flap-wise, edgewise and combined edgewise plus flap-wise loading.
According to GL 2010 the load factor should be greater than 1.25, which is fulfilled for all the load cases studied. We note that in other cases in the literature such as Paquette and Veers (2007) and Chen et al. (2015), local buckling of small size wind turbine blades is found to be a major design concern.

Progressive Damage Analysis
In this subsection results from the progressive damage analysis of the RUZGEM blade subjected to flap-wise, edgewise and combined loading conditions are presented.

Progressive damage analysis under flap-wise (max) loading
The total deformation of the nonlinear blade model versus the undeformed model under 100% extreme flap-wise loading is 290 displayed in Figure 14. The maximum blade deflection at the blade tip is 121 mm. Load displacement curves of the blade in the range between 10% -120% of extreme flap-wise loading are displayed for the 295 linear elastic model and progressive damage model (Puck) in Figure 15. Load application and displacement measurement is done in the flap-wise direction. Based on this load displacement curve, element failure progression in the damaged blade components; pressure side, internal flange and suction side at 75%, 100%, 105% and 166% is depicted in Figure 17 below.
According to the implemented methodology, an element fails if FF or IFF(C) in at least one third of the plies of a laminate is detected. It can be observed from the figure that up to 75% of the extreme flap-wise loading, the stiffness for both linear elastic 300 and progressive damage models remains almost the same. However, the analysis output data show that degradations in the transverse elasticity, shear moduli and Poisson's ratio starting from 13% loading occur, but this does not play a significant role in the deflection of the blade as can be seen from Figure 15. According to the analysis results, element failure is observed in the internal flange at 75% loading. This is the first slope drop of the load displacement curve and can be considered as the first turning point. As seen from Figure 17, failure in the internal flange grows further as the load is increased to 100%. At 105% 305 loading in addition to the damaged region in the internal flange, damage grows along the trailing edge and leading edge. This turning point can be regarded as the second slope drop in the load displacement curve. Due to the element failure, deformation in the form of local buckling at the trailing edge is observed in Figure 16 below.
Shortly before collapse at 166% loading, damaged regions at the leading and trailing edges evolve further and damage close to the blade tip occurs. The reasons for damage initiation at the blade tip at the most extreme load level can be explained by 310 the fact that there is although low, some loading on the blade tip as seen in Figure 8. At 166%, 1.16 times the extreme flapwise loading, which read is from Figure 8 is applied to the blade and the blade collapses afterwards. Moreover, blade tip structure is rather thin and less stiff compared to other regions of the blade. On top of this, as seen in Figure 17 (d) at 166% load level trailing edge and the internal flange which is used to bond the pressure and suction sides of the blade are already damaged. As a consequence, towards the blade tip the pressure and suction sides of the blade are detached at this load level. 315 Under these circumstances blade tip structure is weaker and can be damaged more easily. Likewise, debonding of suction and pressure sides from the adhesive joints was reported as the main failure mechanism causing a progressive collapse of the blade structure in Yang et al.(2013). Finally, the blade collapses after 116% of extreme flap-wise loading    to bottom in a row) at 50% of extreme flap-wise loading.

Progressive damage analysis under edgewise (min) loading
The total deformation of the nonlinear blade model versus the undeformed model under 100% extreme edgewise loading is displayed in Figure 19. The maximum blade deflection at the blade tip is 31 mm and much less than the deformation compared 350 to pure flap-wise loading. Load displacement curves in the range between 10% -320% of extreme flap-wise loading of the blade are displayed for the linear elastic model and progressive damage model (Puck) in Figure 20. In the figure, the load is applied and displacement at the spar tip is measured in the edgewise direction. Figure 22 shows the element failure evolution in the blade components; pressure side, internal flange and suction side of the blade at 100%, 210%, 309%, and 314% load levels. Stress exposures 360 fiber-failure (FF) and inter-fiber failure mode C (IFF(C)) distribution are shown on the same plot. Since element failure in spar is not detected within above mentioned loading range it is not shown in Figure 22. The regions where the stress exposure is equal to or greater than one is depicted in red. The stiffness of these failed elements is set to zero and they do not contribute to blade strength anymore. It is seen that at 100% loading element failure is not detected. It can further be observed from the figure that up to 210% of the extreme flap-wise loading, the stiffness for both models remains almost the same. After 210% 365 loading, the stiffness is reduced slightly due to the element failures at the internal flange tip (See Figure . At %309 loading element failure at the leading in blade root is observed. Due to edgewise (min) loading, compressive stresses are dominant in this region and the blade material is weaker in compressive than in tension. This failure causes the more significant slope drop in the curve and at 314% loading the blade is close to collapse. At the turning point which corresponds to 309% edgewise loading, the stiffness of the failed elements in the leading edge towards blade root are set to zero and total deformation due 370 element failure in the form of local buckling is observed as depicted Figure 21 below.
Based on the results, blade design exhibits excessive safety in edgewise direction and is considered to be over-conservative for this type of loading.  In Figure 23, inter-fiber failure mode A (IFF(A)) and inter-fiber failure mode B (IFF(B)) stress exposure distribution in pressure side, internal flange and suction side of the blade at 140% edgewise loading are shown. Based on the output data from the FE analysis of the blade, the damage initiation begins at 45% of extreme flap-wise loading due to IFF(A). Stress exposures greater than or equal to one indicate damage, and damaged regions are shown in red. The figure shows that inter-fiber failure 390 begins in the internal flange of the blade. IFF(A) and IFF(B) do not lead to the element failure, but causes degradation in the transverse, shear moduli, and Poisson's ratio. The stress exposure at the leading edge near root and trailing are higher compared to other blade regions at this load level, which indicates damage growth at higher load levels in these areas. Under edgewise (min) loading condition leading edge close to blade root is subjected to compressive and trailing edge tensile stresses, respectively. Consequently IFF(A) caused by combination of tensile and shear stresses and IFF(B) caused by the combination 395 of tensile and shear stresses is seen at the trailing edge and leading edge close to blade root, respectively. It is noted that FF and IFF(C) initiate also in the internal flange first but at higher load levels; 210% load. At 309% load FF and IFF(C) is seen at leading edge close to root, where compressive stresses are dominant. As the material has less strength under compression compared to tension, failure initiation begins at the leading edge close to root. These failures are in the same location as IFF (A) or IFF(B) which can be considered as subcritical cracks. As it was the case for flap-wise loading, this observation is similar to 400 the statement (referring to Lambert et al. (2012) and Sorensen et al. (2004) studies) in Montesano et al. (2016) that subcritical cracks can act as precursor to more critical damage modes such as delamination or adhesive debonding. to bottom in a row) at 140% of extreme edgewise loading.

Progressive damage analysis under combined edgewise (min) and flap-wise (max) loading
The total deformation of the nonlinear blade model versus the undeformed model under 100% combined extreme flap-wise and edgewise loading is displayed in Figure 24. The maximum blade deflection at the blade tip is 104 mm and less than the 410 deformation in pure flap-wise loading. We further note that the deflection of 8 mm in the edgewise direction is much less than the deflection of 103 mm in the flap-wise direction.  Figure 27. The regions where the 420 stress exposure IFF(C) or FF is equal to or greater than one is depicted in red. The stiffness of these elements is set to zero and these elements do not contribute to blade strength anymore. In general, under combined loading element failure evolution is similar to pure flap-wise loading case, but failure occurs at higher load levels. Figure 25 shows that up to 85% extreme flapwise loading, the stiffness for both linear elastic and progressive damage models remains almost the same. After 85% loading, slope reduction starts in the nonlinear progressive Puck model due to the element failures at the internal flange, suction, 425 pressure side leading edges. This point is the first turning point in the load displacement curve. As the loading is increased to 100% element failure at the internal flange, suction, pressure side leading edges grows further. Second slope reduction is detected at 118% loading. At this load level, in addition to the damaged region in the internal flange and leading edges, element failure initiates at the trailing edge. This is a more obvious turning point compared to the first turning point at 85% loading. At this load level Figure 27 (c) shows laminate failure in the internal flange, leading and trailing edges. Due to element failure, 430 deformation in the form of local buckling at the leading trailing edge is observed as depicted in Figure 26 below. In their study regarding the full-scale testing of a 34-m wind turbine blade, under combined loading Haselbach and Branner (2016) also observed laminate failure along the trailing edge. As the loading is further increased to 126%, element failure at the leading and trailing edges covers a larger area. As a consequence, suction and pressure panels are detached and the blade integrity is lost. Finally, after %126 loading blade is close to collapse. When the results in Figure 22 are compared with Figure 27, it can 435 be concluded that under same load level the number of failed elements under combined loading is less than pure flap-wise loading case.  Discussion 450 In this section the progressive damage behavior of the blade under flap-wise (max), edgewise (min) and combined flapwise (max) and edgewise loading (min) is discussed. From Figure 15, Figure 20 and Figure 25 it is seen that the slope of the load-displacement curve is highest for edgewise loading followed by edgewise plus flap-wise loading. For flap-wise loading, the slope of the load-displacement curve is the lowest. The structural response of the blade under pure flap-wise versus combined loading is depicted in Figure 28. In the figure, instead of the resultant components of the combined loading, its flap-455 wise components, i.e., load and displacement in flap-wise direction, are plotted. It is seen that the blade exhibits a stiffer behavior in flap-wise direction under combined loading condition compared to pure loading case. As a consequence, flap-wise deflection component under combined loading is approximately 75% of the flap-wise deflection under pure flap-wise loading.
The blade is stronger under combined loading and at 166% loading and the damaged region is less in combined loading compared to flap-wise loading. as shown in Figure 29. 460 In Figure 29, element failure progression is compared for edgewise, flap-wise, and combined loading scenarios under %166 loading. It is observed that the degree of the failed region is highest for the flap-wise loading case, followed by combined loading, and there is no failed region in the edgewise loading case. This failure development can be explained by the superposition of loads and stress tensors as depicted in Figure 30. For this study points A, B and C are picked from the heavily 465 damaged blade regions at 166% flap-wise loading. Since stresses cannot be read in regions where element failures are present, the study is carried out at 100% loading. Point A is picked at the pressure side, and at this point axial tensile stresses are present for both flap-wise and edgewise loading conditions. Hence, when these two load cases are combined there is an increase in the axial tensile stress level. However, stress magnitude is not high enough to cause laminate failure. Points B and C are picked from internal flange and suction side, respectively. Under flap-wise loading point B is subjected to tensile axial stress whereas 470 at point C is subjected to compressive axial stresses. Under edgewise loading point B is subjected to compressive axial stress whereas at point C is under tensile axial stresses. When combined loading is applied, stress components under flap-wise and edgewise loading are superimposed, causing an overall reduction in stress levels at these two points. The superposed stresses decrease the extent of damage at the specified locations.

475
To summarize, finite element analysis of the RUZGEM 5-m blade using Puck's damage model indicates that laminate failure plays a major role for the ultimate blade failure. Laminate failure progression observed under flap-wise, edgewise and combined loading conditions fall into type 4 (internal damage formation and growth in laminates in skin) and type 5 (splitting and fracture of separate fibers in laminates of the skin) wind blade damages as categorized in Sorensen et al. (2014). Although local buckling of small size wind turbine blades is found to be the major design concern in Paquette and Veers (2007) and Chen 480 et al. (2015), the RUZGEM 5m blade is found to exhibit sufficient resistance against buckling in our investigation.  100% extreme loading cases. 495

Conclusions
In this work, strength characteristics of an existing 5-meter RUZGEM composite wind turbine blade under extreme flap-wise, edgewise and combined flap-wise and edgewise loading conditions is investigated. For this purpose, in addition to a linear 500 buckling analysis, progressive damage analysis of the blade using Puck's (2D) physically based phenomenological model is performed. The main conclusions are as follows: • Linear buckling analysis show that blade shows sufficient strength against buckling.
• Failure of elements due to IFF(C) or FF are observed, and a slope reduction in the load displacement is detected after 505 the application of 75% extreme flap-wise loading and %85 combined loading cases. In contrast, under 100% edgewise loading element failures are not observed.
• For flap-wise and combined loading scenarios a similar damage pattern is observed; laminate failure due to IFF (C) or FF in the internal flange causes the first slope reduction in load displacement curve. As the load is increased, damage grows along the trailing edge, which causes second slope reduction before collapse. 510 • For edgewise loading laminate failure observed in the internal flange is the first slight slope reduction in load displacement curve. As the load is further increased, due to compressive stresses, damage accumulates at the leading edge close to blade root, which leads to second slope reduction before collapse.
• FF and IFF(C) initiate in the same location as IFF(A) or IFF(B). IFF(A) or IFF(B) denote subcritical ply cracks which are precede more critical damage modes such as IFF(C) (which is an indicator of possible delamination), and FF. 515 • Less damage is observed under combined loading compared to pure flap-wise loading.
As a summary, internal flange located at the leading edge and trailing edge of the 5-m RUZGEM blade are found to be damaged primarily under flap-wise and combined loading conditions. For the edgewise loading, internal flange and leading edge close to blade root are the mainly damaged areas of the blade. The simulation results will be utilized prior to blade tests which are 520 planned to be conducted on the existing 5-m RUZGEM wind turbine blade.
Author contributions. CM implemented the failure analysis method, conducted the numerical simulations, and wrote the paper. DC is the supervisor and guided CM for the conception of the ideas and participated in writing, structuring, and review of the paper. 525 Data availability. Blade geometry and layup that supports the results of this research is uploaded to the supplement.