Effects of surface curvature on rain erosion of wind turbine blades under high-velocity impact Authors: Wenping Zhou, Dongyou Zhang, Maoli Yang Published: 2024 ABSTRACT AbstractRain erosion induced by raindrops impacting wind turbine blades at high velocity can change the aerodynamic characteristics of the blades and increase maintenance costs. Previous numerical studies on rain erosion have not considered the curvature of the blade leading-edge surfaces and assumed them to be flat surfaces. This study established a fluid-solid coupled numerical model combining the finite element method and smooth particle hydrodynamics. It models a water droplet with a diameter of 2.74 mm impacting the curved leading-edge surface of wind turbine blades with radii of curvature of 1.35 mm, 6.75 mm, 67.5 mm, and infinite at 110 m/s, and the effects of the radius of curvature on the impact response were analyzed. The results show that as the radius of curvature of the leading-edge surface increases, the surface obstructs the water droplet more significantly, and the lateral jetting of the water droplet is enhanced. A larger radius of curvature causes more droplet impact energy to be transferred to the curved surface, increasing the contact force between the water droplet and the surface. The increased transferred impact energy results in higher stress and plastic strain values. The decrease in the radius of curvature of a curved surface increases the error in the stress and strain results obtained by assuming it to be a flat surface.Keywords: Wind turbine blade, Water droplet, High-velocity impact, Surface curvature, Dynamic response Nomenclature Index of space Stress tensor Contact force vector Vector normal to surface Coordinate vector Time step Thickness of the sample mm Maximum effective plastic strain Particle density kg/m3 Maximum stress Mpa Computational volume Face area of the element Relative distance Field function Scale factor of contact stiffness External force Smoothing length Water droplet height mm Index of particles Stiffness factor Bulk modulus Length of the sample mm Mass of particle or node Total number of particles Radius of curvature mm Water droplet radius mm Horizontal distance between TP and impact center Penetration distance Impact velocity m/s Volume of the element Width of the sample mm Interpolation function Introduction The global demand for sustainable energy has constantly increased in recent years, with wind power playing an increasingly prominent role in energy supply decarbonization efforts [1]. Wind turbines are critical for converting wind energy into electrical power. To generate more power from wind energy, the unit capacity and blade geometry of wind turbines have been significantly enhanced over the past few decades [2], [3], [4]. For instance, the length of large-scale wind turbine blades can reach up to 120 m, with blade tip speeds exceeding 100 m/s [5], [6], [7]. When wind turbine blades operate at such tip speeds under severe environmental conditions such as heavy rainfall, the effects of raindrops on the blade surface become essential. One of the main challenges caused by raindrops is known as rain erosion. The physics of rain erosion is complex and is generally believed to result primarily from a combination of one or more of the following mechanisms [8], [9], [10]. First, when a water droplet impacts the blade surface, the water droplet is compressed, and an intense water hammer pressure is generated, which can overload the blade surface and generate compression, shear, and Rayleigh waves. These stress waves with high amplitude and long duration may exceed the blade surface material's dynamic fracture strength, resulting in material damage. Secondly, shortly after impacting the blade surface, the high pressure inside the water droplet is released, and a lateral jetting with a velocity up to several times the impact velocity occurs, potentially damaging the irregular blade surface material. Thirdly, due to the repetitive nature of water droplet impacts during rainfall, repeated loading may lead to surface fatigue. Lastly, repeated water droplet impacts during rainfall may prevent the material from recovering from the previous impacts. Currently, the most widespread measure in industry is the use of thermoset coating systems on blades to resist rain erosion. However, this cannot completely prevent rain erosion. When a water droplet impacts a coating system, shock waves propagate toward the coating-substrate interface depending on the structure and material properties. Some waves are reflected in the coating, and the remainder are transmitted to the substrate. The interaction between the shock and reflected waves may still cause different degrees of damage to the coating material. Previous studies have found that rain erosion on wind turbine blades may increase the drag coefficient and decrease the lift coefficient significantly, thereby reducing the wind turbine's annual energy production [11], [12], [13]. In addition, the maintenance cost of checking and repairing the blades after the occurrence of rain erosion increases significantly [14], [15], [16]. Therefore, the study of rain erosion on wind turbines is becoming important in reducing maintenance costs and increasing annual energy production. Many experimental and numerical efforts have been put into the fundamental aspects of rain erosion in the academic field. In terms of experimental studies, high-velocity raindrop scouring can be physically modeled using whirling arm [17] or jet erosion [18], [19] facilities to test the rain erosion resistance of coating systems. However, experimental studies cannot provide helpful information on the rain erosion resistance of protective materials and coatings due to the differences between the actual environmental conditions of wind turbines and the artificially simulated rainfall in test facilities [20], [21]. In terms of numerical studies, the development of rain erosion models to predict the lifetime of coating materials has been a subject of research for several decades. The Springer model is most commonly used for predicting the fatigue life of coating systems; however, Hoksbergen et al. [22] argued that the model assumptions were not physically representative and required more modern methods to predict the life of coating systems. In subsequent studies, Hoksbergen et al. [23], [24] proposed a numerical model to determine the dynamic contact pressure between the water droplet and the target and combined it with a two-dimensional (2D) axisymmetric finite element model (FEM) to calculate the stress field of the coating system. Kuthe et al. [25] used a coupled Eulerian-Lagrangian (CEL) method to analyze the local stress distribution on the coating. Keegan et al. [26] also used the CEL method to perform water droplet impact simulations and obtained the impact stresses in epoxy panels. Hu et al. [27] adopted the smooth particle hydrodynamics (SPH) method to evaluate the impact stresses of water droplets on coating systems and developed a computational framework for analyzing the fatigue of wind turbine blade coatings. Doagou-Rad et al. [28] analyzed the effects of factors in three domains–environment, design, and manufacturing–on the rain erosion resistance of leading-edge coating systems based on the SPH and CEL methods. Verma et al. [29] developed a coupled computational model based on SPH and FEM to evaluate the effect of water droplet size, impact velocity, and impact angle on the leading-edge coating system. However, the existing numerical studies of the impact of water droplets on wind turbine blades and the development of rain erosion models were conducted assuming the blade surfaces were flat. Rain erosion occurs mainly on the leading-edge surfaces of blades, which are curved solid surfaces. Although a large number of studies have been conducted to investigate the effects of curved solid surfaces on the post-impact diffusion process of water droplets, these studies were mainly focused on relatively low ratios of curved surface size to water droplet size [30], [31], [32], [33]], and limited attentions were paid to the solid surfaces. The water droplet impact events of wind turbine blades involve large ratios of curved surface size to water droplet size. Jin et al. [34] used a many-body dissipative particle dynamics approach to simulate the effects of cylindrical surfaces on the post-impact dynamics of water droplets at relatively low velocities. They showed that the post-impact dynamics are highly correlated with the size factor. Verma et al. [35] employed the effects of the leading-edge surface curvature to analyze the kinematics of water droplet impacts on wind turbine blades, and the results showed that the blade leading-edge curvature affects the water droplet impact kinematics and that the effect increases as the radius of curvature decreases. Fujisawa et al. [36] used a spray-jet erosion device to measure the erosion volume loss of a half cylinder to analyze the effect of leading-edge curvature on the erosion of wind turbine blades, and the results showed that the curvature has a significant influence on the erosion rate and erosion initiation. Zhou et al. [37] investigated the erosion of steel with different curvatures under a water jet, and the results showed that the surface damage of the material decreases with the increase of curvature. Based on the above studies, it can be inferred that in the field of wind turbine rain erosion, the curvature of the blade surface can affect the hydrodynamics of water droplets and the erosion of the coating material. Therefore, conducting an exhaustive analysis of the curved surface's effects on the coating system's transient dynamic properties during high-velocity water droplet impact is necessary, which can help understand the water droplet-induced coating surface damage mechanism. In this study, considering the lack of studies on the impact of water droplets on the curved surfaces of wind turbine blades, a computational model coupling the SPH and FEM methods was established in LS-DYNA to simulate the effects of high-velocity water droplets on the curved leading-edge surface of a wind turbine blade and the impact responses that may occur in the gelcoat material of the blade. The computational results for different surface radii of curvature were compared with those assuming a flat surface to analyze how the contact forces and surface stress patterns applied to the curved surface vary according to the surface curvature and ultimately affect the strains in the gelcoat material. Computational methods and models The widely used nonlinear finite element modeling software LS-DYNA was used to model and solve the solution. The SPH method and the fluid-solid coupling method were first presented. Following that, a three-dimensional (3D) finite element model of a water droplet impacting the curved leading-edge surface of a wind turbine blade was established, and the finite element model was validated using results from the literature. Finally, the effects of the radius of curvature of the leading-edge surface on the water droplet impact loads and stresses induced in the blade coating were discussed. SPH method As water droplets are subjected to large deformation during impact, they are modeled using the SPH method. The SPH method is a pure Lagrangian mesh-free technique whose implementation can be divided into two steps: integral representation of the field function and particle approximation [38], [39], [40]. Firstly, the field function f is defined as where Ω denotes the computational volume, x and x⁎ are the coordinate vectors of the integral and neighboring points, respectively, h is the smoothing length, W is the interpolation function represented by a cubic B-spline curve [41], and d is the relative distance between the integral point and its neighbor. Secondly, in the particle approximation step, the computational domain Ω is described by a collection of particles with physical properties, as shown in Fig. 1. Subsequently, Equation (1) can be converted into the discretized form of the superposition of adjacent particles in the support domain: where i and j are the considered and adjacent particles, respectively, m is the particle mass, ρ is the particle density, and N is the total number of particles in the support domain with smoothing length h. By applying the SPH method to the Navier–Stokes equations of the fluid, we obtain where α and β are used to index the space, v is the particle velocity vector, and σ and G are the stress tensor and external force, respectively. Schematic of the particle approximation. Coupled method Owing to the low deformation rate of the blade surface during impact, it was modeled using the FEM method. The treatment of the fluid-solid coupling is one of the most critical steps when using a combined SPH and FEM approach to model the behavior of water droplets impacting the blade surface. In this study, the coupling of the solid and fluid parts was implemented using a contact algorithm. The node-to-surface contact type was used, where the solid and fluid parts are specified as the master and slave parts, respectively. The interface algorithm adopts the soft constraint penalty method (i.e., SOFT is set to 1 in the LS-DYNA *CONTACT card [42]) because the parts in contact have a significantly different stiffness. When applying this penalty method, the first step is to determine whether the slave particles penetrate the master surface. In the case of penetration between the particles and the surface, a contact force is imposed between them. The contact force vector Fi is given by where ni is the vector normal to the master surface, s is the penetration distance, and k is the stiffness factor, which is given by where fs0 and fs1 are the scale factors for the contact stiffness when SOFT = 0 and 1, respectively; Ki, Ai and Vi are the bulk modulus, face area, and volume of the master surface element, respectively; m is the slave node mass; and Δt is the solution time step during the simulation. Simulation setup A 3D finite element model of a water droplet impacting a wind turbine blade was developed using LS-DYNA and applied to the high-velocity water droplet impact on the curved blade surface with gelcoat material. The model configuration and the physical parameters are described next. The water droplet diameter selection can affect impact characteristics. According to [29], the expected water droplet radius r for a very heavy rainfall intensity of 50 mm/h is approximately 2.74 mm; a water droplet radius of 1.37 mm was selected in this study. Considering that the tip speed of large-scale wind turbine blades typically exceeds 100 m/s, the impact velocity v was selected as 110 m/s. A surface coating system for wind turbine blades based on in-mold technology was selected. The coating system has a complex composition, and in this study, only the outermost layer of the coating system was modeled. Because of the relatively low impact energy of water droplets, according to Keegan [8], the outermost layer of the coating system is mainly affected when the surface material is relatively stiff, and the substrate is only used to represent the coating system and boundary conditions realistically. The response of the gelcoat material was of interest in this study; therefore, it was sufficient to model only the outer layer of the coating system. The gelcoat material profile is represented by a curved sample, as shown in Fig. 2a. For a wind turbine blade, the radius of curvature R of the leading-edge surface varies with blade size, span position, and airfoil thickness, among other variables. In general, R values are larger near the root of the blade, which can be up to 80 mm, and decrease gradually along the outward direction of the blade to a minimum of less than 5 mm near the tip of the blade [35]. This study considered four samples with radii of curvature of 1.35, 6.75, and 67.5 mm, as well as infinite (i.e., flat sample). Considering the mesh quantity and computational cost, the arc length l and width w of the samples' lower surface were set to 8 mm, and the thickness δ was set to 0.3 mm, as recommended in the design guidelines [43]. The calculation cases are presented in Table 1. Schematic of water droplet impact on a curved sample: (a) water droplet and curved sample geometry; (b) SPH particles and curved sample elements. Table 1Calculation cases.Table 1Water droplet radius,Velocity,Radius of curvature,r (mm)v (m/s)R (mm)1.371101.35, 6.75, 67.5, infinite Calculation cases. The lower surface of the gelcoat sample was set as a fully fixed boundary, and the lateral surfaces were set as non-reflecting boundaries to eliminate stress wave reflection. The sample was modeled using finite elements, and the water droplet was discretized with SPH particles, as shown in Fig. 2b. The element type was a single-integration-point constant-stress element with hourglass control added to avoid zero-energy deformation. The element size was set to 0.04 mm, and a finite element total of 140,000 was generated. The SPH particle spacing of the water droplet was set to 0.04 mm, and 113,000 total SPH particles were generated. The particle spacing of the SPH method used in the model was based on independence validation, which is presented in Section 2.4. Owing to the different creation and curing methods, there are many types of gelcoat materials; however, the data on material properties lack fullness. Little et al. [44] conducted extensive experiments on the properties of E862 epoxy and obtained its compressive and tensile material properties at various strain rates. Although E862 epoxy is primarily used in aerospace composites, it was selected as the gelcoat material in this study considering the fullness of the compressive and tensile test data. The engineering stress-strain curves of E862 epoxy were converted to obtain the true stress-strain curves, as shown in Fig. 3. It can be observed that the properties of E862 epoxy material under compression and tension are significantly different and are sensitive to the strain rate. Therefore, the material model *MAT_124 was adopted to model the behavior of E862 epoxy because of its ability to model stress-strain curves in compression and tension individually, and the effect of strain rate on the yield stress can be accounted for by using two load curves to scale the yield stress values under compression and tension. Additionally, the model is capable of simulating failure based on the plastic strain or minimum time step. However, in this study, the impact of a single water droplet was insufficient to damage the material; therefore, it was not considered. The behavior of the water droplet was modeled using the material model *MAT_009 because it allows the equation of state (EOS) to be considered without computing the deviatoric stresses. The Grunissen EOS was used to define the material compressibility. The detailed parameters of the water droplet and EOS can be found in [45]. True stress-strain curves for E862 epoxy for different strain rates: (a) Compression; (b) Tension. Model validation To the best of our knowledge, there is a lack of detailed experimental data on the high-velocity impact of single droplets with velocities greater than 100 m/s [46]. Therefore, the experimental results at a relatively low velocity obtained in [47] were chosen to validate the established numerical model. In the experiment, the impact velocity v was set to 2.99 m/s, the water droplet radius r was set to 1.35 mm, and the target body was a 15-mm-diameter 1-mm-thick circular aluminum panel. The computational parameters and conditions used during the validation calculations were consistent with those in the experimental setup. A model sensitivity study was performed to evaluate the effects of discrete SPH particle spacing. Five SPH particle spacing sizes were investigated: 0.1, 0.06, 0.05, 0.04, and 0.03 mm. Fig. 4a shows the calculated maximum contact force between the water droplet and the aluminum panel as a function of the total number of SPH particles. The maximum contact force converged at a particle spacing of 0.04 mm. By further reducing the particle spacing to 0.03 mm, the total number of particles increases from 172,000 to 395,000, with an error of approximately 0.50%. In this study, considering the computational efficiency and accuracy, the particle spacing used in the SPH representation of the water droplet was set to 0.04 mm. (a) Maximum contact force for different total number of SPH particles; (b) Maximum contact force for different contact stiffness scale factors; (c) Contact force comparison between the experimental and numerical results for r = 1.35 mm and v = 2.99 m/s; (d) Contact force comparison between the proposed and Verma's models for r = 1.37 mm and r = 107.6 m/s. According to Equation (7), the contact stiffness scale factor fs1 affects the magnitude of the contact force when SOFT = 1. We performed a contact-stiffness scale-factor sensitivity study. Fig. 4b compares the calculated maximum contact force with the experimental values for various scaling factors. As can be observed, the calculated value is close to the experimental value at a fs1 of 0.2. Therefore, the contact stiffness scale factor fs1 was set to 0.2 in this study. Fig. 4c shows the numerical results for the contact force for 172,000 SPH particles, and the experimental results are also shown for comparison. It can be observed that both curves achieve good agreement. The maximum contact force calculated by the numerical simulation was approximately 0.0542 N, whereas that measured by the experimental test was approximately 0.0536 N; the error between them was only 1.12%. It is worth noting that the above validation process was performed under low-velocity conditions, and the physical phenomena represented are different from the high-velocity impact of water droplets inducing rain erosion on wind turbine blades. Furthermore, the validation was performed using an aluminum material with linear elastic properties. In contrast, the E862 epoxy is strain-rate sensitive and exhibits complex post-yield behavior, which may alter the contact behavior. To enhance the reliability of the model, the established numerical model was compared with a high-velocity impact model described in the literature. Verma et al. [29] used a coupled fluid-solid interaction computational model to calculate the impact of a single water droplet on a coating system. Fig. 4d shows the contact force results for both models. It can be observed that the two curves present a reasonable fit to each other, except for the passive spreading phase of the droplet after 10 μs. Based on the results, the numerical model established in this study is considered valid and can be used for the subsequent numerical analysis of gelcoat material on wind turbine blades impacted by water droplets at high velocities. Results and discussion Water droplet results on a curved gelcoat surface In this section, the kinematic and dynamic characteristics of water droplets impacting the curved gelcoat samples at high velocities are discussed. As described in Section 2.3, the water droplet radius r was fixed at 1.37 mm, the impact velocity v was fixed at 110 m/s, and different radii of curvature of the samples (R = 1.35, 6.75, 67.5 mm, and flat) were considered. Water droplet morphological evolution Fig. 5 shows the morphological evolution of the water droplet impact on the curved gelcoat samples with varying R by sectioning the water droplet through its center and plotting the velocity contours in the geometry. After impact, the water droplet profile features are similar to the results reported in [29], where the water droplet experienced both an initial compressible stage and a lateral jetting stage. Nevertheless, some observable differences were still observed. First, the size of the low-velocity region within the water droplet increased as R increased, indicating that the water droplet compression was more severe at larger R. It can be inferred from the work of Wu et al. [48] that this is caused by the fact that the larger the R of the sample, the larger the contact area and the more compression wavelets are generated. The lateral jetting stage is the first step in the depressurization of the water droplet, and its strength can be characterized by the radial velocity within the narrow lamella. Fig. 5 shows that the radial velocity within the narrow lamella increased as R increased. To better understand the water droplet morphology, the water droplet height hw and horizontal distance rTP between the turning point (TP) and impact center are defined (Fig. 6 inset). hw is the distance between the highest point of the water droplet and the impact center of the sample, and TP is the connection point between the water droplet body and the lamella. Fig. 6 shows the hw and rTP time history curves for a water droplet after lateral jetting. It can be observed that hw and rTP gradually increase as R increases. For instance, hw and rTP are approximately 1.28 and 1.53 mm, respectively, at 14 μs after the water droplet first touches the surface at R = 1.35 mm; hw and rTP increase to approximately 1.32 and 1.72 mm, respectively, at R = 67.5 mm. The above analysis shows that when a water droplet impacts a sample with larger R, the water droplet's lateral jetting is more severe. This difference in lateral jetting can be caused by the larger contact area between the sample with larger R and the spreading water droplet, which obstructs the falling of the water droplet more significantly. Schematics of the morphological evolution of water droplet impact on the samples: (a) R = 1.35; (b) 6.75; (c) 67.5 mm; and (d) flat. hw and rTP time history curves for the water droplet. Effect of surface curvature on contact force Fig. 7 qualitatively compares the contact force histories of the water droplet impact on the curved samples with varying R. Regardless of the variation in R, there is a distinct contact force peak owing to the initial compression of the water droplet once it touches the sample, and a higher peak occurs owing to the lateral jetting of the water droplet within a short time, after which the contact force curve declines steadily and approaches zero. A maximum contact force of 34.55 N was obtained for R = 1.35 mm, occurring at approximately 3.0 μs after the initial contact. However, for the sample with R = 67.5 mm, a larger R means a larger contact area and more intense water droplet impact, which leads to more impact energy transfer. Consequently, the maximum contact force in our simulation changed to 75.83 N, an increase of 119.47%, and the occurrence time changed to 4.0 μs, a delay of 33.33%. The trends in the duration of the compression stage and the onset of lateral jetting agree with previous findings using equations describing the kinematics of water droplet impacts on curved surfaces [31], [35]. The contact force curves simulated using both a flat sample and a sample with a large radius of curvature (R = 67.5 mm) exhibited good agreement across the entire range. This is consistent with the fact that the larger the R of a curved sample, the closer it is to a flat sample. These results indicate that the maximum contact force generated by the water droplet follows a pattern in which the maximum contact force and occurrence time increase with increasing R. Therefore, for a curved sample with a smaller R, assuming it to be flat without considering the surface curvature can significantly overestimate the maximum contact force, and the lateral jetting of the water droplet can be delayed. Calculated force histories for different R. Effect of surface curvature on the gelcoat impact response Fig. 8 shows 45°top views of the propagation of stress waves in the gelcoat material by plotting the stress contours on the top and central sections of the samples. It can be observed that the impact energy carried by the water droplet is transferred to the samples after the initial contact, which generates a high-stress region around the center of the impact and induces local deformation on the surfaces. For the flat sample, the high-stress region exhibits a ring pattern. As R decreases, the high-stress region gradually shrinks along the arc length to form an elliptical pattern due to the anisotropic nature of the sample and the mitigated obstruction of the water droplet by the sample along the arc length direction. The impact energy and the initial compression stage have significant implications for the generation and propagation of stresses. As previously described, a larger R of the sample will lead to a more severe compression stage of the water droplet, producing a stronger confined water hammer shockwave within the water droplet. Meanwhile, due to the large contact area between the sample and the water droplet, the generated water hammer shock wave cannot be detached from the sample surface quickly, which results in more shock energy transfer, significantly increasing the surface pressure and developing a greater disturbance of the mechanical equilibrium of the sample material. Therefore, a larger R of the sample produces stress waves with higher energies, and the sample material is subject to higher stresses. As can be observed in Fig. 8, the Rayleigh waves become more apparent on the surface of a sample with a larger R. It is well known that Rayleigh waves carry more energy, which, when interacting with the local surface deformation, induces a higher effective stress in the sample. For instance, the maximum stress σmax was approximately 114 MPa for R = 1.35 mm, which increased to approximately 147 MPa for the flat sample. After the occurrence of the maximum stress on the sample, it propagates and dissipates along the surface and thickness directions; for samples with larger R, stresses in the gelcoat material dissipate more slowly after impacting the sample. 45°top views of the propagation of stress waves in the material of curved samples for different radii of curvature: (a) R = 1.35; (b) 6.75; (c) 67.5 mm; (d) and flat. The maximum effective stress σmax generated in the gelcoat sample considerably exceeded the yield strength of the E862 epoxy material (as shown in Fig. 3), resulting in plastic strain in the sample. Fig. 9 shows top views of the plastic strain pattern on the top surface of the samples by plotting the effective plastic strain contours. After the water droplet impacted the flat sample, a ring-like plastic strain distribution pattern with a diameter of approximately 2.14 mm was formed around the contact center. The maximum effective plastic strain (εe)max induced on the sample was approximately 0.060. As R decreased, both the plastic strain pattern and (εe)max decreased gradually. For instance, when R decreased to 6.75 mm, an elliptical plastic strain pattern with a maximum diameter of approximately 2.05 mm was formed around the contact center, and (εe)max decreased to approximately 0.030. As R further decreased to 1.35 mm, the maximum diameter of the elliptical plastic strain pattern was approximately 1.90 mm, and (εe)max was approximately 0.024. Keegan et al. [8] concluded that the combination of high stresses in the material and lateral jetting of the water droplet is responsible for the plastic damage of the material. For relatively large R, stronger water hammer impacts result in higher pressure around the contact periphery of the sample surface, increasing the stress state of the material while generating more severe lateral jetting of the water droplet; therefore, the plastic strains in the target sample are more severe. In addition, for curved samples with small R, the plastic strains in the target samples are significantly different from those assumed to be flat. As previously described, R values of surfaces are relatively small near the tip of wind turbine blades; thus, it is important to include the effects of surface curvature in the analysis of stresses and strains and the modeling of erosion near the tip of blades. Effective plastic strain induced on the curved samples with varying radii of curvature: (a) R = 1.35; (b) 6.75; (c) 67.5 mm; and (d) flat. Conclusion In this study, a fluid-solid coupled numerical model for simulating the behavior of a high-velocity water droplet impact on a wind turbine blade was established by combining the SPH and FEM methods. The effects of curved leading-edge surfaces with varying radii of curvature on the impact process were analyzed, and the results were compared with those of a flat surface. The main conclusions of this study can be summarized as follows: 1) Validation of the established fluid-solid coupled numerical model is challenging because of the difficulties in measuring forces and strains at high velocities. Compared with experimental results for low-velocity water droplet impact on an aluminum panel and those of a high-velocity impact model established in the literature, the contact forces obtained using the established model were in reasonable agreement. The numerical model established in this study was validated and could be used for the subsequent numerical analysis of gelcoat material on a wind turbine blade impacted by a water droplet at high velocity. 2) In the typical leading-edge radius of curvature of a wind turbine blade, the water droplet morphology and response of the gelcoat target sample were significantly affected by the radius of curvature, which further affected the plastic strain behavior of the sample. As the radius of curvature increases, the obstruction of the water droplet by the sample is more significant, leading to an increase in the gelcoat sample's stresses and a slower stress dissipation, which further results in more severe irreversible plastic strains on the gelcoat sample. 3) This study can also be helpful in providing guidelines for rain erosion modeling based on the actual curvature of wind turbine blade surfaces. When analyzing water droplet impacts on a curved surface, if the curved surface is assumed to be a flat surface, the computational error increases as the radius of curvature decreases. Therefore, it is reasonable not to consider the effects of surface curvature while analyzing rain erosion at the root and middle of the wind turbine blade because the radius of curvature of the blade leading-edge surface is relatively large in these parts. For the leading-edge surface near the blade tip, the radius of curvature is relatively small, and it is necessary to consider the effects of surface curvature. The present study contributes to understanding water droplet-induced damage to the surface coating of wind turbine blades. However, limitations and future work for this study may include: 1) The 3D model only models the outermost layer of coating system, and stress propagation in the coating-substrate system has not been investigated. Next, research efforts will look into developing coating system models with realistic curvature (including substrate, putty, etc.) and analyzing the effects of superposition and interactions of stress waves on the stress-strain fields when different coating materials are used. 2) Modeling of impact contact between water droplets and curved surfaces is one limited aspect of rain erosion. Possible future work includes the development of a rain erosion model based on the real blade surface curvature to predict the fatigue life of coatings due to rain erosion and to help develop more durable coatings. CRediT authorship contribution statement Wenping Zhou: Writing – original draft, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization. Dongyou Zhang: Visualization, Software, Formal analysis. Maoli Yang: Validation, Resources, Investigation, Data curation. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This work was supported in part by the Science and Technology Development Program of Liupanshui under Grant 5202020230216, the Liupanshui Normal University Scientific Research and Cultivation Projects under Grant LPSSY2023KJYBPY12, and the Guizhou Provincial Research Project (Youth) for Universities under Grant 2024156. Data availability Data supporting the findings of this study are available from the corresponding author upon reasonable request. References Associated Data Data Availability Statement Data supporting the findings of this study are available from the corresponding author upon reasonable request.