Effect of surface contamination on the wind turbine performance

Abstract

The performance of wind turbines installed in the Middle East and North Africa region is affected by dust accumulation on airfoil surface leading to change in the geometry. The objective of this work is to study the effect of particulate deposition on the aerodynamic performance of wind turbines. Flows past clean and fouled airfoil sections were simulated using a computational fluid dynamics model at various angles of attack at a constant wind speed. Reynolds-averaged Navier–Stokes equations are used along with shear stress transport k–ω model to investigate the flow. The sliding ratio of clean and fouled airfoils was calculated to examine the influence of deposits on the aerodynamics of the airfoil. It has been found that the percentage drop in sliding ratios ranges from 17% to 75%, and NACA 63-215 is the least sensitive to particulate accumulation and is recommended to be used in the blade design of wind turbines operating in dusty environment.

Keywords

Airfoil fouling dust accumulation wind turbines computational fluid dynamics

Introduction

Energy generation is a major challenge facing the world during the current and coming years. Fossil fuels are one of the main sources of energy, providing up to 78% of the total energy consumption (REN21, 2019). However, with the current emergence of renewable and sustainable energy sources, dependence on fossil fuels begins to decrease due to their depletion and their negative impact on the environment, as well as their effect on the health of humans. Renewables have grown fast in recent years (Adams et al., 2018), followed by sharp cost reductions for solar photovoltaics and wind power. A study by the International Energy Agency (IEA) expects renewable electricity generation to increase by more than one-third by 2022 (IEA, 2017). Wind turbines have evolved a lot since its beginning. Global wind power capacity reached 591 GW in 2018, with an increase of 51 GW as of 2017 as stated in a study by Renewable Energy Policy Network for the 21st Century (REN21, 2019).

The wind turbine blade uses the same lift concept as the wing of a plane. The cross-sectional areas of both blades are made of airfoils. However, the types of airfoils differ with the application. The airfoil is designed to maximize the lift over drag ratio L/D, which is called the sliding ratio. Two of the most important parameters to determine the performance of an airfoil are the lift coefficient $C_{L}$ and the drag coefficient $C_{D}$ per unit length, which are calculated using the following equations (Anderson, 2016)

C_{L} = \frac{L}{\frac{1}{2} ρ V_{\infty}^{2} c}

(1)

C_{D} = \frac{D}{\frac{1}{2} ρ V_{\infty}^{2} c}

(2)

where $L$ and $D$ are the lift and drag forces, respectively, $ρ$ is the air density, $V_{\infty}$ is the free stream velocity and c is the chord length. Reynolds number, $Re$ , is calculated to determine whether the flow is laminar or turbulent. It is one of the main governing parameters in all viscous flows where a numerical model is selected according to pre-calculated Reynolds number. It is defined as

Re = \frac{c V_{\infty}}{v}

(3)

where $v$ is the kinematic viscosity of air. During operation, wind turbines are subjected to different weather conditions, such as hail, rain and dust, where each condition has its distinct effect on the turbine output. The wind turbine performance changes greatly according to modifications in the airfoil geometry. These modifications depend on the surrounding weather at the location where the turbine is installed. Unwanted materials accumulate on the surface of the airfoil leading to alteration in the geometry. In Egypt, dust has a major effect on wind turbines located in Hurghada and Zaafarana sites (Khalfallah and Koliub, 2007). Therefore, it is important to study the effect of dusty flow and airfoil surface contamination on the aerodynamic performance of a wind turbine.

Output power and structural stability are two of the main factors affecting the design process of the wind turbine blade. The selection of airfoils with different types and thicknesses at different radial distances is a vital matter when creating the optimal blade design (Khalil et al., 2019), as it will be having direct impact on the wind turbine strength and output power. Thin airfoils with high-lift coefficients and sliding ratios are placed in the blade tip area to maximize the output power performance, whereas thick airfoils are placed near the hub to carry the load. The annual energy production of the wind turbine can be adversely affected by leading edge contamination of the blade. Han et al. (2018) studied the effect of contamination and erosion on the aerodynamic performance of NACA 64-618 airfoil, and consequently on the annual energy production of the wind turbine. Simulations were performed for a 5-MW National Renewable Energy Laboratory (NREL) wind turbine. It was found that the lift and drag coefficients reduced and increased, respectively, as a result of contamination which led to a drop in annual energy production ranging from 2% to 3.7%, thus showing the effect of the change in aerodynamic performance of the airfoil on the wind turbine output power.

Fouling affects other types of equipment such as heat exchangers, where work has been done by Abd-Elhady et al. (2009a, 2009b) to investigate these phenomena and its minimization and removal. Also, a number of studies on airfoil surface contamination roughness have been published in the last few years. In the work by Diab et al. (2015), the change in performance of a wind turbine blade due to dust contamination on different airfoil profiles was studied. A computational fluid dynamics (CFD) model was developed to predict the performance and the drop in power outcome due to dust buildup. Safety precautions were suggested to prevent this problem, such as a leading edge slat. Also, in the research work by Ren and Ou (2009), a simulation was carried out on NACA 63-430 airfoil which is widely used in wind turbines using full 2D Navier–Stokes algorithm and shear stress transport (SST) k–ω turbulence model. The airfoil was first tested at clean surface conditions and was having good consistency with experimental data. Then, it was tested in rough conditions at different locations of the chord length. The theory that roughness induces the premature flow separation was verified by the numerical results. It was concluded that suitable roughness at the trailing edge can be of benefit to lift coefficient and it was proposed that a period of 3 months without any rain would be a proper time to clean the blade’s surface. In a similar work by Khalfallah and Koliub (2007), an experimental study was carried out on blade surface roughness due to dust accumulation in a Hurghada site in Egypt on a 100 kW horizontal axis wind turbine. The mechanism of dust buildup was investigated and the roughness area was varied from 5% to 20% from the chord line to the leading edge. The results showed that the effect of dust accumulation on the performance of the turbine depends on several factors such as speed of the rotor, and altitude of the nacelle from the ground. It was concluded that surface roughness of the blades is increasing in desert and sand sites due to dust accumulation. Computational studies were also performed by Salem et al. (2013) on the performance of wind turbines in dusty environments. The aim of the study was to assess the performance degradation of wind turbines operating in the Middle East and North Africa (MENA) region. A CFD model was developed with a particle deposition model to estimate the effect of roughness in inducing premature flow separation. The model predictions were compared with data from Khalfallah and Koliub (2007) and Ren and Ou (2009) for validation and were found to be reconcilable. The full two-dimensional Navier–Stokes equations with the SST k–ω turbulence model were used to study the decline in the aerodynamic performance of a NACA 63-215 airfoil, and the results showed the loss of power due to dust buildup on the blades. It was recommended to clean the blades every 3 months, rather than a year, to avoid huge power losses. In a related study, Srinivasan and Surasani (2015) also analyzed the effects of surface fouling using CFD. The research aims to study the effect on two specific airfoil profiles: the NREL S814 and NREL S826 at two different Reynolds numbers, and the effect of changing airfoil thickness on the performance degradation of a fouled airfoil. The two airfoils were tested during fouled conditions. It was observed that the transition model predicted flow over smooth airfoils better. However, fully turbulent model had better accuracy when predicting flow over fouled airfoils. Also, it was found that the S826 airfoil was more resistant to the fouling phenomenon. Moreover, the location of transition point was tested in relation to angle of attack. Both airfoils were not able to prevent the fully turbulent flow from occurring over the entire airfoil. Surface roughness effect on wind turbines was also studied by Sagol et al. (2013). Different contamination agents were studied such as dust, ice, dirt and insects with the concentration on how they affect the degree of roughness of the blade. The effect of roughness was investigated on the flow field as well as the performance. It was found that the contamination provokes early transition to turbulent flow which agrees with what Ren and Ou (2009) and Salem et al. (2013) found. It was found that roughness properties such as size, density and location are the most important parameters affecting the performance of the turbine. The SST k–ω turbulence model produced the most accurate results when simulating different models. Solutions were suggested to use airfoil designs with low sensitivity to roughness as stated by Darbandi et al. (2014).

The relation between surface roughness and the change in turbine performance has been addressed in various studies; however, the effect of fouling on the performance of airfoil profiles with different thickness ratios at different angles of attack has not been elucidated. The purpose of this research is to study the effect of surface contamination on the aerodynamic performance of airfoil cross-sections of wind turbine blades and the sensitivity of each of the investigated airfoils to dust accumulation at different angles of attack. The studied airfoils are examined for three different thickness-to-chord ratios (t/c ∼ 15%, 20% and 25%) for Delft University (DU), NACA and NREL sections. The CFD package ANSYS fluent is used in this study to simulate the air flow around the examined airfoils and to examine the sensitivity of each airfoil to dust accumulation.

Numerical model and validation

ANSYS Fluent chosen to simulate the air flows in this study uses Reynolds-averaged Navier–Stokes (RANS) equations, which are used to describe turbulent flows and can be written in Cartesian form as follows (ANSYS, 2012)

\frac{\partial ρ}{\partial t} + \frac{\partial}{\partial x_{i}} (ρ u_{i}) = 0

(4)

\frac{\partial}{\partial t} (ρ u_{i}) + \frac{\partial}{\partial x_{j}} (ρ u_{i} u_{j}) = - \frac{\partial p}{\partial x_{i}} + \frac{\partial}{\partial x_{j}} [μ (\frac{\partial u_{i}}{\partial x_{j}} + \frac{\partial u_{j}}{\partial x_{i}} - \frac{2}{3} δ_{ij} \frac{u_{l}}{x_{l}})] + \frac{\partial}{\partial x_{j}} (- ρ \bar{{\overset{\cdot}{u}}_{i} {\overset{\cdot}{u}}_{j}})

(5)

where $ρ$ is the density, $t$ is time, ${\bar{u}}_{i}$ and ${\overset{\cdot}{u}}_{i}$ are the mean and fluctuating velocity components, $μ$ is the dynamic viscosity, $δ_{ij}$ is Kronecker delta, and i, j are indicial notations denoting the x, y, z axes $(i, j = 1, 2, 3)$ . RANS equations are used to predict the time-averaged flow fields. However, it is not possible to purely average the equations and obtain equations obtaining only the desired averaged values. Therefore, turbulence models are used to determine these unknown variables and provide closure to the equations (Munson et al., 2012). The shear–stress transport (SST) $k - ω$ model is one of the most common and widely used turbulence models. The SST $k - ω$ model was created by Menter (1994) to combine robustness of the $k - ω$ to simulate the near-wall region with the free stream dependence of $k - ϵ$ model in the far field.

The modeling process is divided into three main elements: pre-processing, solving and post-processing. In the pre-processing phase, the geometry of the airfoil and computational domain is created, as shown in Figure 1, which shows the location of the boundary conditions in relation to the airfoil. A C-type computational domain was used, and the shape was made of a semi-circle with a size reaching radius of 10 chord lengths and a rectangle with an upper and lower side length of 40 chord lengths, to satisfy the independence of the results on the size.

Figure 1.

Computational domain around the airfoil.

Grids, or meshes, fall into two categories: structured and unstructured, depending on whether a pattern of connectivity of grid points with the neighboring ones occurs or not. Unstructured grids, shown in Figure 2, can be very useful when applied to complex geometries such as the fouled airfoils. Therefore, it was used when simulating the investigated airfoils. Several mesh controls were used to ensure that the requirements would be met for cells around the airfoil to produce accurate results. Body sizing, edge sizing and inflation around the airfoil were employed to create finer cells. The number of cells in the mesh ranged from 25,000 to 35,000 depending on the mesh settings.

Figure 2.

Unstructured mesh over an airfoil: (a) overall grid for the whole computational domain and (b) zoom in for the grid around the airfoil.

To correctly represent the boundary layer of the airfoil, the mesh was altered by creating cells near the airfoil surface with height as low as 5 × 10^–5 m to ensure that for a Reynolds number of 1 × 10⁶, the y-plus value would fully resolve the boundary layer. Y-plus is a parameter used to describe how fine the applied mesh is for a wall. Each turbulence model has its limiting y-plus value, for the SST $k - ω$ model, y-plus value should range from 1 to 5 through the airfoil (Diab et al., 2015). The y-plus values for clean and fouled versions of NREL S820 examined airfoil are shown in Figure 3, where the maximum value for the clean airfoil was approximately 1.67, and the maximum value for the fouled airfoil was about 3.5, due to profile irregularities, which is in the acceptable range for the chosen turbulence model.

Figure 3.

Y-plus values for NREL S820 Airfoil in case of (a) clean and (b) fouled airfoils.

The second part is the numerical solver setup, where the boundary conditions are specified as shown in Figure 1. The edge on the left (A) with the upper and lower edges (B) represents the inlet of the flow, while the far right edge (C) represents the outlet. The airfoil (D) is simulated as a wall. The velocity inlet boundary condition is set using a velocity of 14.6 m/s, air density of 1.225 kg/m³ and viscosity of 1.7894 × 10^–5 kg/m s to reach a Reynolds number of 1 × 10⁶. Pressure outlet boundary conditions are set at the domain outlet, and the airfoil surface has been set to no-slip-solid boundary.

The simulation results are compared with experimental data to validate the model. The first validation was performed on airfoil NREL S814 and was tested at a Reynolds Number of 1.5 × 10⁶ and zero angle of attack, which are the conditions of the experiment. Simulation results are shown in Table 1, along with the experimental data from Srinivasan and Surasani (2015).

Table 1.

Comparison between calculated and experimental C_L and C_D of NREL S814 airfoil.

	Computational results	Experimental measurements	Error (%)
C_L	0.415	0.40	3.75
C_D	0.01009	0.01	0.9

The second validation case was carried out on NREL S809 airfoil at a different Reynolds number. The NREL S809 airfoil was tested at Reynolds number of 2 × 10⁶ and the results were compared to experimental data from NREL wind tunnel measurements (NREL, 2019). The free stream velocity was 29.2 m/s. The numerical results are presented with the experimental data in Figure 4, where the lift coefficients are compared at different angles of attack.

Figure 4.

Lift coefficient validation of NREL S809 with experimental data.

In the first validation, the error percentages in lift (3.75%) and drag (0.9%) coefficients, as shown in Table 1. As for the second validation shown in Figure 4, the CFD results show good agreement with the experimental data, which proves the credibility of the model used.

Results and discussion

Dust accumulation over the tested airfoils over a period of 3 months at zero angle of attack and wind speed of 15 m/s was simulated using a particle deposition model created by El-Batsh (2001) using FLUENT’s user defined functions capability. According to the flow field, Fluent calculates particle paths, and several scenarios are considered as they approach the surface which are particle deposition, reflection or detachment. Geometries of the fouled airfoils are obtained from Diab et al. (2015) using the particle deposition model and are tested at a range of optimum angles of attack from −2° to 3°, as stated by Sayed et al. (2012). The airfoil profiles used in this research are NACA 63-215, NREL S820, DU84-132V3, NACA 64(4)421, NREL S819, NACA 4424 and NREL S815 (NREL, 2019; UIUC Applied Aerodynamics Group, 2019), which are shown in Figure 5.

Figure 5.

Investigated airfoils: (a) NACA 63-215, Max thickness 15% at 34.9% chord; (b) NREL S820, Max thickness 16% at 45.1% chord; (c) DU84-132V3, Max thickness 13.6% at 33.9% chord; (d) NACA 64(4)421, Max thickness 20.9% at 34.8% chord; (e) NREL S819, Max thickness 21.1% at 28.4% chord; (f) NACA 4424, Max thickness 24% at 29.4% chord; and (g) NREL S815, Max thickness 26.2% at 25.7% chord.

Lift and drag coefficients are obtained for clean and fouled airfoils, then the sliding ratio is calculated to observe the change in the aerodynamic performance. Lift coefficients of clean and fouled versions of airfoils under study are shown in Figure 6 at different angles of attack. The lift coefficient for most fouled versions of the airfoils is lower than that of the clean versions. However, the percentage drop differs according to the airfoil profile. As shown in Figure 6, the airfoil with the greatest lift coefficient before fouling is the DU 84-132V3. Also, in some rare cases, fouling improves the lift coefficient of the NREL S815 at angles of attack of −2 and −1. Dust accumulation considerably affects the drag coefficient of the studied airfoils, as shown in Figure 7. The NREL S815 airfoil has the highest drag coefficient in case of clean and fouled airfoils and it has the greatest sensitivity when it comes to changing the angle of attack. Also, the graph shows the increase in drag coefficient of all airfoils when they are subjected to surface contamination. Fouling caused an early transition to turbulent boundary layer causing an increase in drag at the whole range of angles of attack. However, the change in drag depends on the thickness-to-chord ratio of the investigated airfoil. The thickness-to-chord ratio compares the maximum vertical thickness of a wing to its chord, and it varies from approximately 15% in case of the NACA 63-215, NREL S820 and DU84-132V3 to 26% in case of the NREL S815. The increase of the thickness-to-chord ratio follows the increase with drag coefficient across the investigated airfoils as shown in Figure 7. However, it does not affect the increase in the drag coefficient which depends on the fouled shape of the airfoil, which results in different pressure distribution around the airfoil.

Figure 6.

Lift coefficient versus AOA for clean and fouled airfoil profiles.

Figure 7.

Drag coefficient versus AOA for clean and fouled airfoil profiles.

All of the airfoils were affected by surface contamination; however, the extent of this effect depended on the airfoil under study and the thickness of the fouling layer. The airfoils tested have different responses to dust accumulation. Although a certain airfoil can be having a high sliding ratio, it does not guarantee maintaining the same performance when dust is accumulated on the surface and creating a fouled profile. This change in performance is shown in Figure 8, where it clarifies the effect of fouling on airfoils and the sensitivity of each airfoil to the change in geometry. The percentage drop in the sliding ratio ranged from 17% to 75%, with NACA 63-215 having the lowest drop at different angles of attack, thus showing the least sensitivity compared to other airfoil profiles.

Figure 8.

Percentage drop in sliding ratio for fouled airfoils.

To understand this phenomenon, two airfoils, namely NACA 4424 and NACA 63-215, were studied at different angles of attack to determine the influence of fouling on the performance of the airfoils. NACA 4424 is one of the airfoils with deterioration in aerodynamic performance due to fouling at all angles of attack. The lift coefficient, shown in Figure 9, decreased due to fouling at the whole range of angles of attack, with the exception of angle of attack of −2°, where the change was close to zero. However, the drag coefficient largely increases after dust contamination, as shown in Figure 10.

Figure 9.

Effect of fouling on lift coefficient for NACA 4424.

Figure 10.

Effect of fouling on drag coefficient for NACA 4424.

These changes can be explained by the pressure contours and pressure coefficient shown in Figures 11 and 13, and the velocity contours shown in Figure 12, over the clean and fouled airfoil at angle of attack of 0°. The difference in pressure over the upper and lower surface of the clean airfoil is greater than that of the fouled airfoil. Fouling creates points with smaller pressure and larger velocity in the lower surface more than the upper surface, as shown in Figures 11 and 12. These points contribute to the decrease in lift coefficient of the airfoil after contamination. Also, points with high pressure were created at the leading edge of the airfoil, which can be further observed in Figure 13, creating larger pressure difference and increasing drag coefficient. Fouling created an area of inconsistent irregularities at the leading edge of the airfoil. This explains the fluctuations in pressure coefficient of the fouled airfoil shown in Figure 13, which is due to the abrupt changes in geometry in this region.

Figure 11.

Pressure contours for NACA 4424 airfoil at AOA = 0°: (a) clean and (b) fouled.

Figure 12.

Velocity contours for NACA 4424 airfoil at AOA = 0°: (a) clean and (b) fouled.

Figure 13.

Pressure coefficients of clean and fouled NACA 4424 airfoil at AOA = 0°.

One of the special cases is the NACA 63-215, which showed a very small effect on the lift coefficient due to surface contamination. The lift coefficient had a slight change when subjected to contamination at all angles of attack, as shown in Figure 14. However, the drag coefficient, shown in Figure 15, was largely affected by surface contamination.

Figure 14.

Effect of fouling on lift coefficient for NACA 63-215.

Figure 15.

Effect of fouling on drag coefficient for NACA 63-215.

The change in lift and drag can be explained by the pressure contours and pressure coefficients at different angles of attack. At angle of attack of −2°, the pressure on the lower surface of the clean profile was lower than that on the upper surface causing a negative lift. In the case of the fouled profile, the airfoil had a negative lift also, but the differential pressure was lower than that of the clean version. The pressure contour at this angle of attack, as shown in Figure 16, shows close pressure values. However, the change in pressure can be more clarified in Figure 17, showing the small changes in pressure at the upper and lower surface, as well as the change in pressure at the leading edge of the airfoil due to fouling. Pressure fluctuations due to fouling were smoother in case of NACA 63-215 than in case of NACA 4424 as shown in Figures 13 and 17, which proves that NACA 63-215 was less sensitive to dust contamination and its performance was less affected.

Figure 16.

Pressure contours for NACA 63-215 airfoil at AOA = −2°: (a) clean and (b) fouled.

Figure 17.

Pressure coefficients of clean and fouled NACA 63-215 airfoil at AOA = −2°.

As for angle of attack of 3°, shown in Figure 18, the airfoil was subjected to positive pressure in both cases. However, the difference in pressure was lower in case of the fouled version, causing a lower lift coefficient and fouling caused a premature transition to turbulent flow causing an increase in drag coefficient. This change can also be explained by the change in velocity over the airfoil, as shown in Figure 19. The fouled area in the leading edge created more points with low velocity, thus higher pressure causing an increase of the drag coefficient of the studied airfoil. However, the change in lift force was relatively low due to the small difference in pressure over the upper and lower surfaces at the same angle of attack.

Figure 18.

Pressure contours for NACA 63-215 airfoil at AOA = 3°: (a) clean and (b) fouled.

Figure 19.

Velocity contours for NACA 63-215 airfoil at AOA = 3°: (a) clean and (b) fouled.

Conclusion

Each airfoil showed different aerodynamic sensitivity to surface contamination leading to a change its geometry. For most airfoils, lift coefficient decreased while the drag coefficient increased for fouled airfoils relative to their clean versions, resulting in a drop in sliding ratio, thus affecting the output power of the wind turbine. The percentage drop in sliding ratio ranged from 17% to 75%. The average drop in sliding ratio for most airfoils was about 40%. Another important aspect was the increase in the sliding ratio of all clean and fouled versions with the increase in angle of attack. The best airfoil operating in clean conditions was the DU 84-132V3 because of its high sliding ratio at different angles of attack. However, this does not ensure the same performance after leaving the blade in dusty environments resulting in a change in the geometrical shape of the airfoil. Therefore, studying the sensitivity of each airfoil to particle deposition was as important as gaining the maximum power when operating in clean conditions. The NACA 63-215 airfoil was the least sensitive to dust accumulation and showed exceptional performance at different angles of attack and had a small drop in sliding ratio in comparison with the percentage drop of other airfoils.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Youssef ElMessiry

References

Abd-Elhady

Rindt

CCM

, et al. (2009a) Removal of gas-side particulate fouling layers by foreign particles as a function of flow direction. Applied Thermal Engineering 29: 2335–2343.

Abd-Elhady

Rindt

CCM

van Steenhoven

(2009b) Optimization of flow direction to minimize particulate fouling of heat exchangers. Heat Transfer Engineering 30: 895–902.

Adams

Mensah Klobodu

Apio

(2018) Renewable and non-renewable energy, regime type and economic. Renewable Energy 125: 755–767.

Anderson

(2016) Fundamentals of Aerodynamics (6th edn). New York: McGraw-Hill Education.

ANSYS (2012) ANSYS Fluent Theory Guide. Canonsburg, PA: ANSYS.

Darbandi

Mohajer

Behrouzifar

, et al. (2014) Evaluating the effect of blade surface roughness in megawatt wind turbine performance using analytical and numerical approaches. In: Proceedings of the 10th international conference on heat transfer, fluid mechanics and thermodynamics, Orlando, FL, 14–26 July.

Diab

Alaa

El-Din

, et al. (2015) Performance degradation of wind turbine airfoils due to dust contamination: A comparative numerical study. In: Proceedings of the ASME turbo expo 2015: Turbine technical conference and exposition, Montreal, QC, Canada, 15–19 June. New York: ASME.

El-Batsh

(2001) Modeling particle deposition on compressor and turbine blade surfaces. PhD Thesis, Vienna University of Technology, Vienna.

International Energy Agency (IEA) (2017) Renewables 2017. Available at: https://www.iea.org/reports/renewables-2017 (accessed March 2020).

10.

Khalfallah

Koliub

(2007) Effect of dust on the performance of wind turbines. Desalination 209: 209–220.

11.

Khalil

Tenghiri

Abdi

, et al. (2019) Improvement of aerodynamic performance of a small wind turbine. Wind Engineering 44(1): 21–32.

12.

Han

Kim

(2018) Effects of contamination and erosion at the leading edge of blade tip. Renewable Energy 115: 817–823.

13.

Menter

(1994) Two-equation eddy-viscosity turbulence models for engineering applications. AIAA Journal 32(8): 1598–1605.

14.

Munson

Rothmayer

Okiishi

, et al. (2012) Fundamentals of Fluid Mechanics (7th edn). Hoboken, NJ: John Wiley & Sons.

15.

National Renewable Energy Laboratory (NREL) (2019) Wind turbine airfoils. Available at: http://wind.nrel.gov/airfoils/

16.

REN21 (2019) Renewables 2019 global status report, 2019. Available at: http://ren21.net/gsr_2019_full_report_en (accessed February 2020).

17.

Ren

(2009) Dust effect on the performance of wind turbine airfoils. Journal of Electromagnetic Analysis and Applications 1: 102–107.

18.

Sagol

Reggio

Ilinca

(2013) Issues concerning roughness on wind turbine blades. Renewable and Sustainable Energy Reviews 23: 514–525.

19.

Salem

Diab

Ghoneim

(2013) CFD simulation and analysis of performance degradation of wind turbine blades in dusty environments. In: Proceedings of the 2013 international conference on renewable energy research and applications (ICRERA), Madrid, 20–23 October. New York: IEEE.

20.

Sayed

Kandil

Shaltot

(2012) Aerodynamic analysis of different wind-turbine-blade profiles using finite-volume method. Energy Conversion and Management 64: 541–550.

21.

Srinivasan

Surasani

(2015) CFD analysis of effects of surface fouling on wind turbine airfoil profiles. International Journal of Energy and Power Engineering 4: 1–11.

22.

UIUC Applied Aerodynamics Group (2019) UIUC airfoil coordinates database. Available at: https://m-selig.ae.illinois.edu/ads/coord_database.html (accessed August 2019).