Int.
Journal of Renewable Energy Development 8 .
2019: 285-292 P a g e | 285 Contents list available at IJRED website Int.
Journal of Renewable Energy Development (IJRED) Journal homepage: http://ejournal.
id/index.
php/ijred Research Article Numerical Study of Effect of Blade Twist Modifications on the Aerodynamic Performance of Wind Turbine Wiroj Beabpimaia and Tawit Chitsomboonb* School of mechanical engineering.
Institute of engineering.
Suranaree University of Technology.
Nakhonratchasima.
Thailand Independent researcher.
Nakhonratchasima.
Thailand ABSTRACT.
This paper aims to investigate aerodynamic performance of a wind turbine blade with twist modifications using computational fluid dynamics (CFD).
The phenomenon of 3D stall-delay effect in relation to blade twist is the key feature to be investigated in order to improve efficiency of a wind turbine.
The NREL (National Renewable Energy Laborator.
Phase VI wind turbine rotor was used for validation and as the baseline rotor.
The baseline blade geometry was modified by increasing/decreasing the twist angles in the inboard, mid-board and outboard regions of the blade in the form of a symmetrical curve with maximum twist angle of 3A.
The steady incompressible Reynolds-averaged Navier-Stokes (RANS) equations with the k-O Shear Stress Transport (SST) turbulence closure model were used for the calculations at wind speeds ranging from 5-20 m/s.
The computational results for the baseline Phase VI rotor were validated against experimental data and a good agreement was found.
The computational results for the modified blades were compared against those of the baseline blade.
It was found that increase of annual energy production of up to 5.
1% could be achieved by this modification technique.
A2019.
CBIORE-IJRED.
All rights reserved Keywords: CFD, wind turbine, blade twist, stall delay.
NREL Phase VI Article History: Received: March 29, 2019.
Revised: August 17, 2019.
Accepted: August 19, 2019.
Available online: Oct 30,2019 How to Cite This Article: Beabpimai.
, and Chitsomboon.
Numerical Study of Effect of Blade Twist Modifications on the Aerodynamic Performance of Wind Turbine.
Internasional Journal of Renewable Energy Development, 8.
, 285-292.
https://doi.
org/10.
14710/ijred.
Introduction Renewable and sustainable energies are increasingly being researched to provide the world with clean and reliable energy sources.
Renewable energies can be used for various purposes with several applications such as electricity generation, desalination, cooling, heating, etc.
Some interesting papers on solar and nanofluid can be found in references (Ahmadi et al.
, 2018.
, and Ramezanizadeh et al.
, 2.
Wind energy is another option which has experienced rapid growth during recent decades in terms of technology and commercialization.
Research and development on making wind turbines to be more efficient and competitive are crucial.
The rotor blades are very important parts of a wind turbine since they convert kinetic energy of the wind into mechanical Designing of wind turbine rotor blade to yield a high efficiency is a very challenging task since it involves various complicated aerodynamic manipulations.
Accurate predictions of aerodynamic forces are necessary for the effective design of wind turbine blades.
Innovations are also needed to gain appreciable increase in efficiency.
Generally, two main theoretical approaches for the design of wind turbine are the Blade Element Momentum (BEM) theory, and the Computational Fluid Dynamics (CFD).
The BEM method is computationally fast and is easy to implement but has some limitations such as there is no inclusion of the three-dimensional .
D) effects and the unsteady effects (Hansen et al.
the other hand, the CFD method which employs the Navier-Stokes equations, does not have these limitations and thus can provide a deeper insight into the complex flow phenomena over wind turbine rotor.
CFD
methodology has been widely used in the literature to analyze the aerodynamic performance of horizontal-axis wind turbines .
ee, e.
Syrensen et al.
Johansen et Duque et al.
Hsu et al.
Li et al.
Twist angle of a blade is one of key parameters in design, having considerable effects on the mechanical power capture of a wind turbine rotor.
The twist angle is defined as the angle between the local airfoil chord and that at the blade tip.
The increase in incidence angle of relative flow velocity from the blade tip to the root suggests that a blade be twisted more toward the root in order to achieve the optimal flow angles over the entire length of the blade.
Most commercial wind turbines have nonlinear twist distributions along the blades.
The twist angles toward the root are greater than the angles toward the tip.
However, in the case of nonlinear blade twist distributions with increasing or decreasing twist angles at Corresponding author: tawit.
boon@gmail.
A IJRED Ae ISSN: 2252-4940.
All rights reserved Citation: Beabpimai.
and Chitsomboon.
Numerical Study of Effect of Blade Twist Modifications on the Aerodynamic Performance of Wind Turbine.
Int.
Journal of Renewable Energy Development, 8.
, 285-292, doi.
org/10.
14710/ijred.
P a g e .
some regions of the blade span, the aerodynamic performances of rotor blade have not been extensively researched and only a few studies exist.
Guntur .
numerically studied the MEXICO rotor with inboard twist modifications and found that the trailing vortices, which is caused by the change in the span-wise loading, results in an effect similar to that typically observed near the blade tip.
Chow and van Dam .
investigated the effects of twist modifications in the inner area of the 5 MW The study focused on wind speed of 11m/s, at the knee of the power curve.
The results showed that increasing inboard blade twist reduces rotor power capture and thrust.
With decreasing inboard blade twist, rotor power remains constant while thrust increases approximately linearly.
3-D rotational effects on wind turbine blade is a very interesting phenomenon.
It was found to delay stall at high angles of attack and increase lift coefficient significantly over that of 2-D airfoil, especially at the inboard region of the blade span.
This behaviour is also referred to as stall-delay.
it was first observed by Himmelskamp .
on propellers and was further investigated in several later studies, e.
Wood .
Ronsten .
Schreck and Robinson .
This research proposes to investigate and explore beneficial effects of the 3-D stall by altering blade twists from that of the baseline values.
In this study, the effects of blade twist modifications on the aerodynamic performance of wind turbine rotor have been investigated using CFD methodology.
The Phase VI wind turbine rotor of the National Renewable Energy Laboratory (NREL) was chosen as the baseline The baseline blade geometry was modified by increasing or decreasing the twist angles on three different regions along the blade span.
The computational results of the modified blades were compared against those of the baseline blade, in terms of rotor power, thrust and annual energy production (AEP).
Fig 1.
NREL Phase VI wind turbine in the wind tunnel of NASA-Ames (Hand et al.
Methods 1 Specification of wind turbine rotor The NREL Phase VI rotor shown in Figure 1 is a two-blade wind turbine with stall regulated design, which was tested comprehensively in the large wind tunnel of NASA Ames Research Center under extensive range of operating conditions (Hand et al.
This wind turbine has a rotor radius of R=5.
029 m and a power rating of 19.
W).
The blade geometry is based on the S809 airfoil profile, with a nonlinear twist and a linear taper (Giguere and Selig 1.
The twist and chord distributions, and the geometric model of the blade are shown in Figure 2.
The blade twists decrease from 20.
at r/R=0.
25 to -1.
815A at the tip.
The blade chords taper 737 m at r/R=0.
25 to 0.
355 m at the tip.
In the NREL Phase VI experimental campaign, an upwind configuration consisting of the cone angle of 0A, non-yaw, the tip pitch angle of 3A and the rotational rotor speed of 72 rpm, is considered for this computational study.
More details about the wind turbine and procedures of the experiment are available in Hand et al.
Fig 2.
Geometry of the NREL Phase VI blade.
chord and twist distributions, and top view of the blade Fig 3.
Designed blade twist distribution A IJRED Ae ISSN: 2252-4940.
All rights reserved Int.
Journal of Renewable Energy Development 8 .
2019: 285-292 P a g e | 287 Fig 4.
Spanwise twist distributions for the baseline and modified blades 2 Twist distribution modification The blade was identical to that of the baseline NREL Phase VI, only the twist angles over a specific region of the blade were modified.
An example of the twist angle distributions that were changed from the baseline blade was a symmetry curve with a maximum angle of 3A degree as shown in Fig.
The curve was constructed by a cubic spline function through five control points .
olid points in the figur.
The starting and the ending points were set to be zero while the other three control points were determined by the Gaussian function (Eq.
G ( x ) = a exp y -( x -b ) y Where a, b, and c are constants The numerical experiments were performed over the regions of blade spans: r=0.
60R (TW1 and TW.
, 75R (TW3 and TW.
, and 0.
90R (TW5 and TW.
, with increasing or decreasing twist angles in each blade span.
The final blade twist distributions along the blade span are shown in Figure 4.
3 Numerical method The computational mesh is illustrated in Fig.
The flow domain is semi-cylindrical geometry since the wind turbine contains two symmetrical blades.
Therefore, only one of the two blades is modelled to save computational resources, and the remaining blade is accounted for using periodic boundary conditions imposed on the 180A cyclic The computational domain extends the wind turbine six rotor radius upstream, five rotor radius in the span-wise direction, and eight rotor radius downstream to guarantee that the turbine flow is independent of the presence of the outer boundaries (Thumthae and Chitsomboon 2.
For the inlet boundary condition, uniform velocity normal to the inflow surface with turbulence intensity of 5% was imposed.
At the downstream outlet boundary, the pressure outlet condition was used.
The wind tunnel wall was treated as a symmetry boundary condition.
the rotor blade surface, the no-slip wall condition was The wind speeds of the simulation were 5, 7, 9, 10, 11, 13, 15, 17, and 20 m/s with 1.
23 kg/m3 air density 78Oe5 kg/.
viscosity, for all of the runs.
The computational mesh generated by ANSYS ICEM CFD software (ANSYS ICEM CFD, 2.
The mesh contained about 2.
74 million hexahedral cells based on grid independence tests.
The grid was an O-type topology along most of the blade surface with 185 nodes around the airfoils and 165 nodes in the spanwise direction.
The y of the first cell layer close to the blade surface was around 1 with a growth ratio of 1.
2 in order to ensure a well resolved boundary layer.
The CFD computation was performed with the commercial code ANSYS FLUENT (ANSYS FLUENT, 2.
which is based on the finite volume method.
The steady, incompressible.
Reynolds-averaged Navier-Stokes (RANS) equations were applied to solve the problem on a rotating reference frame.
The QUICK scheme was used for discretization of convective term, and SIMPLE algorithm was used for pressure-velocity coupling.
The iterative tolerances of convergence for all solved variables were set at 10Oe6.
For turbulence modeling, the shear stress transport (SST) k-O turbulence closure model (Menter et al.
was selected because of its good predictions of boundary layer flows under adverse pressure gradient and The formulation of the model is expressed A( r k ) A( rU j k ) ! = Pk - b * r k w At Ax j A y Ak y y( AA s k AAt ) Ax j yy Ax j yy A ( rw ) A ( rU jw ) = ar S 2 - brw 2 At Ax j rs w 2 Ak Aw A y Aw y y( AA s w AAt ) y 2 .
- F1 ) Ax j yy Ax j yy w Ax j Ax j A IJRED Ae ISSN: 2252-4940.
All rights reserved .
Citation: Beabpimai.
and Chitsomboon.
Numerical Study of Effect of Blade Twist Modifications on the Aerodynamic Performance of Wind Turbine.
Int.
Journal of Renewable Energy Development, 8.
, 285-292, doi.
org/10.
14710/ijred.
P a g e .
AAt = a1Ak max.
SF2 )
where ycI = 2ycI,- ycI,- is the invariant measure of the strain rate, and yca" is a constant.
The value of yca" = 0.
30 is used in present work, instead of yca" = 0.
The blending function ya3 is given by:
y 2 k 500AA y y y F2 = tanh y y max yy * yy y y y yy y y yy ya" and ya3 are blending functions which are designed to be one inside the boundary layer .
-O mode.
, and switches over to zero for free shear layers .
-A mode.
All constants in the SST model are calculated by a blend from the corresponding constants of the k-O and the k-A model via:
j = j1F1 j2 .
- F1 ) .
j = .
k , s w , b , a } .
The constants for this model are:
s k1 = 0.
85 , s w1 = 0.
5, b1 = 0.
075 , a1 = 5 / 9 , b * = 0.
09 , s k 2 = 1.
0 , s w 2 = 0.
856 , b 2 = 0.
a 2 = 0.
Fig 5.
Computational mesh.
Complete domain.
blade surface and cross section mesh 1 Numerical validation The production terms are:
y AU AU j y AU i Pk = AAt y i y Ax Axi yy Ax j P!k = min ( Pk ,10b * r kw ) .
The blending function ya" is given by:
y k 500AA y 4 rs w 2 k y y yy y y .
F1 = tanh yy min y max yy * , 2 yy , yy y b w y r y w y CDkw y yy yy yy 1 Ak Aw -10 y CDkw = max y 2rs w 2 Results and discussion .
yco is the turbulence kinetic energy.
O is the turbulent dissipation frequency, yc is the normal distance to the nearest wall, ycO is the flow velocity, yuU is the density, yuN is the dynamic viscosity The turbulent eddy viscosity .
uN( ) is given by:
The validation of CFD was performed by comparing the simulation results with the baseline blade .
NREL
Phase VI) experimental data (Simms et al.
Hand et As shown in Fig.
6, the predicted rotor power and axial thrust values are generally in good agreements with the experimental results.
Although, there is a slight overprediction of the rotor power at 11 m/s wind speed, and an underprediction of the rotor power for high wind speeds of 15-20 m/s where massive flow separations occur.
The computed rotor power value at 17 m/s wind speed is approximately 18% low compared to measured data.
Figure 7 presents comparisons of predicted and measured pressure coefficient distributions on the bladeAos five spanwise sections .
30, 0.
47, 0.
63, 0.
80 and 0.
95R) for the case of 7 m/s, 10 m/s, 15 m/s, and 20 m/s wind speeds.
The pressure coefficient .
a6 ) is obtained by Equation .
, where ycE8 is the free stream pressure, yu is the rotational speed of the blade, ycO8 is the free stream wind speed, and yc is radius of section.
Cp = p - pAu 5r (U Au2 (Wr ) 2 ) .
At low wind speed .
, the predicted pressure coefficient distributions at all sections of the blade agree well with the measured values.
A IJRED Ae ISSN: 2252-4940.
All rights reserved Int.
Journal of Renewable Energy Development 8 .
2019: 285-292
P a g e | 289
Experiment
CFD
Experiment
CFD
Thrust .
N) Power .
W) Wind speed .
Wind speed .
Fig 6.
The calculated rotor power and thrust of the baseline blade Fig 7.
Comparison of measured and computed pressure distributions at 7 m/s, 10 m/s, 15 m/s and 20 m/s wind speeds A IJRED Ae ISSN: 2252-4940.
All rights reserved Citation: Beabpimai.
and Chitsomboon.
Numerical Study of Effect of Blade Twist Modifications on the Aerodynamic Performance of Wind Turbine.
Int.
Journal of Renewable Energy Development, 8.
, 285-292, doi.
org/10.
14710/ijred.
P a g e .
In the case of 10 m/s, there is a slight difference between the predicted and measured pressure distributions on the suction side for x/c < 0.
4 at the span sections r/R = 0.
47 and 0.
At 15 m/s wind speed, a large discrepancy in pressure distributions is noticed at r/R=0.
30 section, which is believed to be due to the large separation and strong vortices due to high blade twist near the root.
At 20 m/s wind speed, where the flow is fully separated over the entire blade span, the computed pressure distributions regain good agreements with the experimental results since the effects of the turbulence model are no longer 2 Power and thrust In this section, comparisons of CFD simulation results of the baseline blade and the twist modified blades are The rotor power and thrust comparisons and their percentage changes from the baseline blade are shown in Fig.
8 and Fig.
9, respectively.
From Figure 8, the power captures of TW1.
TW3 and TW5 blades .
, the cases that increase the twist angles from the baseline twist distributio.
are less than the baseline blade at low wind speeds and become higher than the baseline blade at medium and high wind speeds.
Compared with the baseline case, the TW1 blade has the maximum increase in rotor power around 11.
2% at 10 m/s wind speed, the TW3 blade has the maximum increase in rotor power about 27% at 13 m/s, and the TW5 blade has the maximum percentage increase in rotor power about 9% at the wind speed 15 m/s.
For the TW2.
TW4 and TW6 blades .
, the cases that decrease the twist angle.
, the opposite is true wherein the power captures are lower than the baseline blade at medium and high wind speeds but are higher than the baseline blade at low wind speeds.
Compared to the baseline blade, the rotor power of TW2 blade is reduced around 17.
5% at the wind speed 9 m/s, and TW6 blade has the maximum percentage decrease in rotor power about 26.
4% at 13 m/s.
Figure 9 shows axial thrust comparisons.
In contrast to the rotor power curves, there are small difference in the thrust among the blades when wind speed is higher than 9 m/s.
In comparison with the baseline blade.
TW1.
TW3 and TW5 blades exhibit lower thrusts at low wind speeds and slightly higher at medium and high wind speeds with maximum percentage increase in thrust less than 2%.
For the TW2 blade, the thrust as well as the power are quite lower than the baseline blade at 9 m/s wind speeds.
This is due to the separated vortex on the suction surface of the Percentage of power increase (%) Power .
W) Baseline TW1
TW2
TW3
TW4
TW5
TW6
TW1
TW2
TW3
TW4
TW5
TW6
Wind speed .
Wind speed .
Fig 8.
Comparison of rotor powers .
and percentage changes from the baseline blade .
TW1
TW2
Thrust .
N) Percentage of thrust increase (%) Baseline TW3
TW4
TW5
TW6
TW1
TW2
TW3
TW4
TW5
TW6
Wind speed .
Wind speed .
Fig 9.
Comparison of axial thrusts .
and percentage changes from the baseline blade .
A IJRED Ae ISSN: 2252-4940.
All rights reserved Int.
Journal of Renewable Energy Development 8 .
2019: 285-292 3 Annual Energy Production (AEP) AEP is a measure of economic feasibility of a wind turbine.
To calculate AEP it is necessary to multiply the rotor power curve with the probability density function of the In this study.
Rayleigh wind speed distribution was For a given average .
wind speed U , the probability of wind speed f (U ) is obtained by the following formulation:
f (U ) = y p yU y y exp y - y y y y 4 yU y y Wind power density (W/m.
P a g e | 291 AMWS = 6.
2 m/s AMWS = 7.
2 m/s AMWS = 8.
2 m/s Wind speed .
The annual mean wind speeds (AMWS) in the range of 2 to 8.
2 m/s were used, which correspond to the average wind speed that used in designing the original rotor (Giguere and Selig 1.
Figure 10 shows the Rayleigh wind speed distributions for three annual mean wind speeds.
AMWS = 6.
2 m/s, 7.
m/s and 8.
2 m/s.
Figure 11 shows the distribution of annual mean wind power densities which are calculated from the Rayleigh distributions.
The wind power density indicates how much wind energy is available at a site.
From Figure11, it is noticed that the highest wind power densities are found at wind speeds around 10 m/s, 11.
m/s, and 13 m/s for annual mean wind speed 6.
2 m/s, 7.
m/s, and 8.
2 m/s, respectively.
Figure 12 and Table 1 show the percentage increases in annual energy production (AEP) of the modified blades compared with the baseline blade.
It can be seen that TW1 and TW3 blades can produce higher AEP than the baseline blade for all annual mean wind speeds from 6.
2 to 8.
2 m/s.
Comparison to the baseline blade, the AEP of blades TW1 and TW3 increase by 2.
4% to 3.
9% and 1.
8% to 5.
This is mainly due to the blades having high percentage of rotor power increases at wind speeds in the range of 10-13 m/s which contain high wind power The TW5 blade tends to produce higher AEP than the baseline blade at high annual mean wind speeds, due to the blade having higher rotor power than the baseline blade at high wind speeds.
In the cases of TW2.
TW4, and TW6 blades, their AEPs are lower than the baseline blade at all annual mean wind speeds, except at AMWS = 6.
2 m/s for the TW6 blade.
Fig 10.
Rayleigh distributions for annual mean wind speeds of 2, 7.
2 and 8.
2 m/s Fig 11.
Wind power densities for the mean wind speeds of 6.
2 and 8.
2 m/s Fig 12.
Increase/decrease in AEP compared to the baseline Conclusion In this paper, the effects of blade twist modifications of the NREL Phase VI wind turbine were studied using CFD.
The validation results for the baseline blade obtained using RANS with the k-O SST turbulence closure model were generally in good agreement with the experimental data, though discrepancy existed at high wind speed The baseline blade was modified by increasing or decreasing the twist angles in the inboard, mid-board and outboard regions of the blade in the form of a symmetrical curve.
The results show that the rotor power of the increasing blade twist cases have significant improvements over the baseline values under medium and high wind speed conditions, while those of the decreasing blade twist cases have little improvements under low wind For annual mean wind speeds in the range of 6.
2 m/s, it is found that the cases of increasing twist angles at blade regions 0.
60R (TW1 blad.
, and 0.
75R (TW3 blad.
show best results with increase of AEP 4% to 3.
9% and 1.
8% to 5.
1%, respectively.
This study thus has confirmed that, under proper design, the blade twist modification in conjunction with 3D stall delay phenomenon can help improves efficiency of wind turbine A IJRED Ae ISSN: 2252-4940.
All rights reserved Citation: Beabpimai.
and Chitsomboon.
Numerical Study of Effect of Blade Twist Modifications on the Aerodynamic Performance of Wind Turbine.
Int.
Journal of Renewable Energy Development, 8.
, 285-292, doi.
org/10.
14710/ijred.
P a g e .
Table 1
Annual Energy Production Comparison
AMWS
Baseline AEP (MWh/.
TW1 blade
AEP
(MWh/.
Increase rate (%) AEP (MWh/.
Increase rate (%) AEP (MWh/.
Increase rate (%) AMWS .
Baseline AEP (MWh/.
AEP (MWh/.
Increase rate (%) AEP (MWh/.
Increase rate (%) AEP (MWh/.
Increase rate (%) TW4 blade Acknowledgments This research is supported by the Royal Golden Jubilee Ph.
Program of the Thailand Research Fund.
References