You are on page 1of 18

Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Contents lists available at ScienceDirect

Journal of Wind Engineering & Industrial Aerodynamics


journal homepage: www.elsevier.com/locate/jweia

The effect of the integral length scale of turbulence on a wind


turbine aerofoil
Giulio Vita a, b, *, Hassan Hemida b, Thomas Andrianne c, Charalampos Baniotopoulos b
a
Department of Industrial Engineering – DIEF, School of Engineering, University of Florence, Italy
b
Department of Civil Engineering, School of Engineering, University of Birmingham, UK
c
Wind Tunnel Lab, Department of Aerospace and Mechanical Engineering, University of Liege Belgium

A R T I C L E I N F O A B S T R A C T

Keywords: The effect of free stream turbulence on a DU96w180 wind turbine aerofoil is investigated through wind tunnel
Wind turbine aerofoil experiments. Wind turbine blades experience large scale, high intensity turbulent inflow during their service life.
Integral length scale of turbulence However, the effect of turbulence is normally neglected in the assessment of their aerodynamic performance. This
Atmospheric turbulence
is normally justified based on common assumptions on the effect of the integral length scale of turbulence, which
Aerodynamics
Wind tunnel
supposedly only acts in the low-frequency range of the energy spectrum, hence affecting the angle of attack
Passive grid instead of the aerodynamic behaviour. In this study, an experimental setup implementing passive grids is
developed to vary independently turbulence intensity and integral length scale in wind tunnel testing, with a
range spanning I~5–15% and Le8  33 cm respectively. Results show that turbulence effects are not negligible
even at the largest integral length scales, provided that a critical value for the turbulence intensity is achieved.
Turbulence is found to increase mean and fluctuating Lift and delay separation in stalled conditions.

1. Introduction the fatigue limit state (Mouzakis et al., 1999; Lindeboom, 2010; Mücke
et al., 2011; Toft et al., 2016) all agree that turbulence has a major in-
Two decades since the first applications, urban wind energy remains a fluence on performance, bringing the IEC standard on small wind tur-
bizarre niche of wind energy, with a poor track record and too much bines (IEC 61400-2:2013 Wind turbines - Part 2: Small wind turbines,
room for improvement (Micallef and van Bussel, 2018; Stathopoulos 2013) to specifically forbid the installation of a small wind turbine in
et al., 2018). The small wind turbine technology is largely borrowed from locations where the turbulence intensity exceeds Iu > 15%. The appro-
large wind farms. As a result not much results are available on the urban priateness of such a limitation is debated (Tabrizi et al., 2015; Evans
wind resource, or the response of devices to the turbulence conditions et al., 2017; KC, Whale and Urmee, 2019), as more research on the tur-
found therein (Balduzzi et al., 2012; Ishugah et al., 2014; Battisti et al., bulence characteristics found at installation sites is necessary to advance
2018). If some progress has been made in the understanding of the wind knowledge (Ishugah et al., 2014; Micallef and van Bussel, 2018; Sta-

resource (Emeis, 2014; Sarki c Glumac, Hemida and H€ offer, 2018), the thopoulos et al., 2018). As regards the power output of small wind tur-
positioning of a wind turbines in a built environment (Balduzzi et al., bines, most studies suggest that ambient turbulence improves
2012; Toja-Silva et al., 2016), or the design optimisation of small wind performance at low wind speed while impairing it at high wind speed
turbines (Rezaeiha et al., 2018), the aerodynamic interaction between a (Milan et al., 2013; St Martin et al., 2016; Carbo Molina, Bartoli and De
rotor and ambient turbulence is a lack in the knowledge that prevents the Troyer, 2017; Tang et al., 2019). While the vast majority of works rely on
accurate assessment of the performance (Arnfield, 2003; Bak, 2007; van turbulent flow conditions recreated in wind tunnel or numerical testing
Kuik et al., 2016). to reach these conclusions (Walker, 2011; Al-Abadi et al., 2016), large
Although much research has been devoted to wind turbine perfor- scale experimental evidence of improved power performance seems to
mance in turbulence, a clear understanding and consensus on the phys- confirm the beneficial effect of turbulence at low speed is present in in-
ical mechanism affecting the aerodynamic interaction is yet to be dustrial scale wind farms (Albers et al., 2007; Wagner, 2010). The limi-
achieved. Studies on the structural loading (Frandsen, 2007; Kelly et al., tation in the scope of results published so far might explain the different
2014), the development of aeroelastic instabilities (Hansen, 2007), and trend observed in a recent study, which found that turbulence has a

* Corresponding author. Department of Industrial Engineering – DIEF, School of Engineering, University of Florence, Italy.
E-mail address: g.vita@bham.ac.uk (G. Vita).

https://doi.org/10.1016/j.jweia.2020.104235
Received 31 January 2020; Received in revised form 8 May 2020; Accepted 8 May 2020
Available online 31 July 2020
0167-6105/© 2020 Elsevier Ltd. All rights reserved.
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

blade span-wise flow and utilising 2D aerofoil data (Hau, 2013). These
are obtained in controlled laminar flow conditions (Tangler, 2002;
Sørensen, 2011). BEM theory accounts for ambient turbulence as an
unsteady fluctuation of the angle of attack (Hansen et al., 2006). The
justification for this approach resides on the fact that the integral length
scale of atmospheric flows is normally much larger than the aerofoil
chord (Miley, 1982; Hand et al., 2001; Buresti, 2012). However, this
assertion is based on knowledge on atmospheric turbulence borrowed
from large scale wind farms, hence not representative of conditions found
in urban winds. In fact, there is no facility specifically designed to
reproduce urban wind inflow, i.e. having large integral length scales
Lu ≫ c in conjunction with high turbulence intensity Iu e15 % (Vita et al.,
2018). First investigations on the aerodynamic performance of an aero-
foil in ambient turbulence were undertaken to improve wind tunnel
testing accuracy, i.e. measuring the effect of low turbulence intensities
1–2% at small angles of attack (Mueller et al., 1983; Hoffmann, 1991;
Huang and Lee, 1999). These early studies found turbulence to enhance
the performance of aerofoils, causing a delay in stall, and increasing
Fig. 1. Normalised power curve of a HAWT as measured frome the sea and city maximum lift and aerodynamic performance (or lift-to-drag ratio). The
sectors, predicted using the method described in Sunderland et al. (2013), and physical mechanism is analogous to that found in bluff bodies, with two
the design power curve (after data published in Pagnini et al., 2015).

detrimental effect on power output regardless of the wind speed (Pagnini


et al., 2015). Fig. 1 shows the power curve of a horizontal axis wind
turbine (HAWT) installed in an urban seafront location alongside a ver-
tical axis wind turbine (VAWT). Power measurements were conditionally
split depending on the wind direction, i.e. undisturbed conditions from
the sea or turbulence from the city. The higher turbulence intensity
(20–40%) from the city sector was linked with a drop in performance of
e80 %. While some other technical issues might contribute to the drop in
power output (although observed for two different wind turbine, in
absence of any damage), this work clearly shows a correlation between a
lower power performance and enhanced turbulence in the installation
site (Pagnini et al., 2018).
The aerodynamic forces experienced by a wind turbine are calculated
using the blade element momentum (BEM) theory, i.e. neglecting the Fig. 3. The DU 96w180 aerofoil and the position of the pressure taps.

Table 1
Experiments on effect of turbulent inflow on WT blades. PG/AG Active/Passive grid, PT pressure taps.
Authors/year chord c [m] grid type t. int. TI [-] len. sc. L/c measures Notes

Present Study 0.125 PG 0.05–0.15 0.50–2.80 40 PT independently varied


Amandol ese and Szechenyi (2004) 0.5 PG 0.075 – 25 PT oscillating aerofoil
Devinant et al. (2002) 0.3 PG 0.16 – 43 PT –
Kosasih and Saleh Hudin (2016) 0.15 PG 0.29* – power Rot. Blade *close to grid
Li et al. (2016) 0.14 PG 0.139 – 46 PT Low Re custom. blade
Seddighi and Soltani (2007) 0.25 PG 0.4* – 64 PT pt placed at 20 angle
(Sicot and Devinant, 2006) 0.3–0.07 PG 0.16 – 43 PT; PIV* *small blade for PIV
Sicot et al. (2008) 0.07 PG 0.12 0.88 26 PT Rotating blade
Swalwell and Sheridan (2001) 0.125 PG 0.07 0.56 28 PT thin aerofoil
Swalwell et al. (2004) 0.125 PG 0.13 1.30 28 PT thick aerofoil
Maldonado et al. (2015) 0.25 AG 0.061 0.60 32 PT Rough blade
Wang et al. (2014) 0.1 AG 0.06 0.25 LDA Low Re

Fig. 2. Experimental setup mounted in the test section of the wind tunnel.

2
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Table 2
Independently varied turbulence intensity cases 1, 2 and 3. Relevant grid number and axial distance are re-
ported together with the Power Spectral Density of velocity and calculated flow statistics Iu and Lu.

effects on the flow pattern (Bearman and Morel, 1983; Zdravkovich, have in common the generation technique for ambient turbulence, i.e. a
1997; Buresti, 2012): passive or active grid upstream to the aerofoil. Most of cited works do not
show or discuss the integral length scale obtained in the study. Of those
1) The triggering of laminar-to-turbulent transition in boundary or shear studies, the large majority reports Lu < c, which is clearly not represen-
layers; tative of urban wind conditions. Furthermore, most of the studies listed
2) Enhanced mixing and entrainment, increasing the thickness of the in Table 1 do not provide a detailed description of the flow field used in
detached shear layer, and resulting in a delayed separation or early the experiments, being turbulence intensity the parameter of reference in
reattachment. all studies, with the integral length scale disregarded in most cases.
Therefore, consensus on the response of aerofoils under atmospheric
Small scales are supposedly responsible for the mechanism 2), i.e. turbulence has not yet been achieved, and authors of mentioned studies
those scales in the flow with a size comparable to the thickness δ of the in Table 1 all auspicate further experiments.
boundary or shear layer (Haan et al., 1998). Largest length scales around Sicot, Aubrun, et al. (2006) and Amandolese and Szechenyi (2004)
the integral length scale affect the overall features of the flow pattern, show that turbulence affects the stall mechanism in a stronger way than
such as the wake recovery or the vortex shedding if present (Bearman and respectively rotation or oscillations. However, while the latter are
Morel, 1983; Nakamura et al., 1988). Some works on bluff bodies argued accounted for with a corrective factor to the angle of attack (Hansen
that large scales may also act as a constraint to the effect of turbulence, et al., 2006), the former is neglected in BEM calculations. Maldonado
preventing it (Ohya, 2004; Kistler and Vrebalovich, 2006). However, et al. (2015) acknowledge the difficulty in studying the effect of turbu-
authors agree that this is more a suggested tendency of results, rather lence having a length scale larger than the chord length, while stressing
than proven knowledge (Bearman and Morel, 1983; Nakamura, 1993; how indeed the performance of the rotor increases due to turbulence at
Nakamura and Ohya, 2006; Nakamura et al., 2012). the expense of the increased demand to the blade structure due to an
In more recent years, the renewed interest in wind energy in un- increase in drag. Nevertheless, it is unclear whether these conclusions
conventional locations has prompted a surge of studies on wind turbine hold in more realistic atmospheric turbulence. Swalwell, Sheridan and
aerofoils in ambient turbulence, i.e. with higher turbulence intensities Melbourne (2004) and Swalwell and Sheridan (2001) notice a monotonic
than previously investigated. Relevant works are given in Table 1, where trend between the delay in stall and increase in performance with the
the main features of the experiments are briefly summarised. All works turbulence intensity, which is present for different thicknesses of

3
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Table 3
Independently varied integral length scale cases 4, 5 and 6. Relevant grid number and axial distance are re-
ported together with the Power Spectral Density of velocity and calculated flow statistics Iu and Lu.

aerofoils and Reynolds flow regimes. However, their experimental setup applicable in general to low Reynolds number aerofoils in pre- and post-
was designed to control turbulence intensity exclusively, with a limited stalled conditions under a variety of isotropic and gaussian turbulent
applicability to atmospheric flow conditions. Other studies (Seddighi and inflows which resemble turbulent atmospheric flows. Results might also
Soltani, 2007; Kosasih and Saleh Hudin, 2016) seem to implement pas- be useful to industrial scale wind farms to understand the possible role of
sive grids so that a correspondence to the ABL in terms of normality, turbulence from the wake of upstream turbines on the flow pattern of
isotropy, and turbulence cannot be verified, and therefore have limited downstream rotors (Vermeer et al., 2003). Nevertheless, this work re-
applicability to wind energy. mains limited from the fact that the expected turbulence characteristics
This work is designed to overcome the limitations of works listed in to be found at a generic unconventional installation site are still largely
Table 1, to understand the role of the integral length scale on the aero- unknown.
dynamics of a wind turbine aerofoil. The aim is to understand whether This work also shares the limitations of the BEM theory as a static
turbulence effects are negligible based on the length scale utilising a aerofoil in turbulence is considered. Results by Sicot, Aubrun, et al.
wind tunnel setup implementing passive grids designed to generate a (2006) seem to reassure on the negligibility of rotation effects on the
variety of turbulent inflows having turbulence intensities Iu and integral blade when in turbulence, but obviously their setup would need to be
length scales Lu , varied independently. The setup was optimised to obtain extended to include a length scale compatible with the urban environ-
a combination of high turbulence intensity Iu e15 % having a sufficiently ment, which is yet to be constructed, and this study might prompt the
large length scale Lu e3c based on the chord of the aerofoil model (or development of innovative experimental setups for the investigation of
Lu e30 cm). The experiment also provides a variability of length scales, turbulence effects on a more comprehensive setup involving a wind
namely Lu/c ~ [0.50, 2.80], which is normally difficult to reach and turbine in operation.
control in wind tunnel testing with an acceptable range of turbulence Section 2 reports on the experimental setup and calculation strategy.
intensities Iu ~ [5%, 15%]. The quality of the grid generated turbulence In Section 3, results are presented and discussed in terms of pressure
is extensively studied in Vita et al. (2018). coefficients and related force coefficients. Conclusions are presented in
The obtained turbulence statistics are relevant to possible conditions Section 4.

a small wind turbine in the built environment might experience (Sarki c
Glumac, Hemida and H€ offer, 2018). However, results in this work are

4
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 4. Stall Mechanism in terms of (left) Lift coefficient CL and (right) Drag coefficient CD, with comparison to experimental data from literature (Timmer and van
Rooij, 2003; Devenport et al., 2010; Lindeboom, 2010; Sareen et al., 2012; Suryadi and Herr, 2015; Matyushenko et al., 2017). Markers represent measurements under
turbulent (red) and laminar (grey and blue) inflow conditions, at various Reynolds regimes, listed in the table below the figure. (For interpretation of the references to
colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5. Undisturbed case. (Top) Mean surface Pressure Coefficient for 3 Angles of attack used in this study, compared with results from (Timmer and van Rooij, 2003;
Timmer, 2010). (Bottom) Standard deviation of surface pressure coefficient.

5
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 6. Mean pressure coefficient cp for the three angles of attack and Cases 1 to 3.

2. Methodology (FDM). Fig. 2 shows the model in position in the wind tunnel. The model
was fitted with a steel bar to hold it together, and fixed to an aluminium
2.1. Experimental setup: the DU96w180 aerofoil model frame screwed to the wind tunnel floor.
The monolithic centrepiece hosted the PTs and a ~2 cm section of the
The tests have been carried out in the Wind Tunnel Lab of the Uni- tubing, which was then fitted with recesses to glue the 1 m long pressure
versity of Li
ege (Belgium) at the same time with the grid turbulence tests tubing in order to limit disturbance to the signal. The tubing was con-
reported in Vita et al. (2018). The wind turbine aerofoil of choice for the nected to a Dynamic Pressure Measurement System (DPMS, from the
test is the DU 96w180, designed at the Delft University of Technology company Turbulent Flow Instrumentation Ltd.), able to measure up to
(Timmer and van Rooij, 2003), shown in Fig. 2 along with the position of 128 PTs in the range 0  2500 Pa at 500 Hz.
the pressure taps. This aerofoil has been specifically designed to have a Fig. 3 shows the position of the PTs along the aerofoil. The upper
smooth stall mechanism and optimised against noise, being suitable for section of the aerofoil, called ‘suction’ side, presents 24 taps, while the
small wind turbines which are normally stall-regulated. A sufficient lower section is called ‘pressure’ side, with the remaining 16 taps at a
structural stiffness is guaranteed bdue to its thickness of 0.18c. Wind larger spacing. A pressure tap was placed in correspondence of the
turbine blades might be fitted with trip strips to impose leading edge, while the value of the pressure at the trailing edge,
laminar-to-turbulent transition (Lyon et al., 1997; Traub, 2011), a ne- necessary to calculate force coefficients, was estimated by extrapolating
cessity which was not deemed necessary in this study. The model is fitted from the closest taps, following the procedure implemented in Mary and
with 40 pressure taps (PT) having a diameter of 0.6 mm. In order to Sagaut (2002).
minimise the chord length c, while retaining a sufficiently high resolu- The chord of the model is c ¼ 0:125 m. This value has been chosen
tion, the PTs have been embedded in the 3D printing of the mid-span in order to achieve a compromise between the possibility of fitting the
section. Selective Laser Sintering (SLS) was chosen for the mid-span model with pressure taps and the ability to test large length scale tur-
section of the aerofoil, while the rest of the 1.25 m total span consisted bulence Lu/c~2. In literature, Lu~20–25 cm is considered a reachable
of mountable sections printed with the Fused Deposition Modelling range of values using passive grid generated turbulence (Kurian and

6
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 7. Standard deviation for the three angles of attack and Cases 1 to 3.

Fransson, 2009). This chord length allows for Reynolds numbers up to measured, andU ¼ uðtÞ is the mean velocity. The standard deviation
Re~1200 000 being reachable, which is consistent with a small wind pffiffiffiffiffi pffiffiffiffiffi
u2 , is used to calculate Iu ¼ u2 =U. The integral length scale Lu is
turbine operating in the urban environment. The main parameter to be
varied throughout the experiment is the angle of attack α. The angles are estimated from the autocorrelation coefficientρðτÞ ¼ Ruu ðτÞ=u2 , where
chosen so they can be representative of the stall mechanism: pre-stalled Ruu ðτÞ ¼ uðtÞuðt þ τÞ is the autocorrelation function, and τ is the time
4 ; post-stalled 14 ; and full-stalled 24 . In the pre-stalled condition, the lag. In this case Lu ¼ UTu where Tu is the integral time scale, obtained
flow is attached to the aerofoil and a partial limited separation can be from the area subtended by the ρðτÞ curve, which can be approximated
Z τ0
observed as α grows. 4 has been specifically chosen as the flow is fully with Lu ¼ U ρðτÞdτ, where ρðτ0 Þ ¼ 0. In this work, Tu is estimated
attached until the trailing edge (Timmer and van Rooij, 2003). As sep- 0
aration occurs, the lift coefficient cL increases up to a maximum value using a simplified relation, where ρðTu Þ ¼ 1=e (Conan, 2012).
cL;max , which occurs at αe10 . cL does not suddenly drop for this kind of Table 2 and Table 3 show the passive grid configurations and relevant
aerofoils, and a post-stall region can be defined 10 < α < 20 , where cL turbulence statistics respectively for setups where turbulence intensity
slowly decays. For α > 20 , cL starts to increase again, following a flat was kept constant to assess the effect of varying the integral length scale
plate-like pattern. (cases 1 to 3), and vice versa where the integral length scale of turbulence
is kept constant to quantify turbulence intensity effects (cases 4 to 6).
Tables 2 and 3 also report the power spectral density for the turbulent
2.2. Experimental setup: the inflow
inflows. Cases 1 to 3 see the low-frequency ends of the spectra over-
lapping, while cases 4 to 6 have the maximum occurring at the same
The experimental setup was studied in order to vary Iu and Lu statistics
frequency, consistent with the target turbulence statistics. More details
separately. The turbulence intensity Iu is calculated using the Reynolds
are given in (Vita et al., 2018), where a slightly different grouping of
decomposition u ¼ uðtÞ – U, where uðtÞ is the velocity realisation as

7
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 8. 99th percentile of cp for the three angles of attack and Cases 1 to 3.

statistics is performed. turbines in full scale and therefore relevant to urban wind energy
Tests have been conducted setting a wind speed for the wind tunnel (Rezaeiha et al., 2018; Toja-Silva et al., 2018). Fig. 4 compares the pre-
rotor of Ur ¼ 15 m/s. However, due to the presence of the grids and the sent dataset with results available from literature, only published for
conformation of the wind tunnel, the actual wind speed as measured higher Reynolds ranges varying from 1 to 3  106 (Timmer and van
using a Pitot Tube, varies. More details can be found in the following and Rooij, 2003; Devenport et al., 2010; Lindeboom, 2010; Sareen et al.,
in Vita et al. (2018). 2012; Suryadi and Herr, 2015; Matyushenko et al., 2017). A strong
Reynolds effect is noticeable in the absence of turbulence when α < 20 .
3. Reynolds effects and stall mechanism However, when Ree1  105 , such an effect disappears as cL is analogous
to the experimental data from literature (Freudenreich et al., 2004).
One of the turbulent setups (using grid #3 at distance 4 m, with Iu ¼ However, in turbulence, Reynolds effects are negligible, as the data
8:50 %) has been adapted to assess the presence of possible Reynolds aligns to undisturbed configuration at the highest wind speed. Therefore,
effects. As mentioned, the mean wind speed downstream of grids varies it seems sensible not to adjust the mean wind speed to match exactly the
depending on the solidity of the grid, the distance to it, and the confor- whole dataset: turbulence characteristics are insensitive to the mean
mation of the wind tunnel, as the one in Liege has an expansion which wind speed (Kurian and Fransson, 2009).
was useful to obtain a varied range of statistics, as specified in Vita et al. The Lift coefficient is linear for α < 10 , then it gently drops to then
(2018). increase again at α > 26 following a flat-plate-like behaviour. The Drag
Fig. 4 shows the mean Lift and Drag coefficients against the angle of coefficient increases quickly until αe18 , and then continues to grow
attack, for a variety of configurations as specified in the legend, calcu- linearly for higher angles of attack, analogously to flat-plates.
lated from the surface pressure. The Reynolds number varies from 21,000 Fig. 4 might lead to think that in turbulence, an aerofoil behaves
to 120,000. Such range might resemble Reynolds regimes of small wind

8
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 9. Mean pressure coefficient on the aerofoil. Compared with undisturbed case and results from Timmer.

similarly to a bluff body with respect to Reynolds effects even in the separation. The local maximum at x/c~0.55 might occur due to a sepa-
attached flow region. This means that under FST the aerodynamic per- ration bubble which spans the distance between the two local maxima of
formance could be investigated regardless of Reynolds effects. Unfortu- c’p , i.e. 0.4 < x/c < 0.6. The formation of separation bubbles might occur
nately, this study has not been designed to shed light on Reynolds effects at low angles of attack, if the boundary layer is not tripped as in this case,
in turbulence, and an interesting question opens whether turbulence and this pattern might explain the slight deviation of the stall mechanism
might sometimes even benefit wind energy (and not harm it, as at low angles of attack shown in Fig. 4. When α ¼ 14 , a more pro-
commonly thought). nounced local maximum can be observed at x/c~0.35. As α increases, the
possible separation bubble which might occur moves upstream as the
4. Undisturbed case adverse pressure gradients of the flow become more pronounced. This is
the condition when the maximum lift is reached. For x/c > 0.5, cp shows a
Fig. 5 shows the effect of the angle of attack for the DU96w180 plateau which might indicate a region of separated flow. Such a plateau
aerofoil in undisturbed flow conditions. The present experimental data- runs throughout the chord length in the case of α ¼ 24 , with no local
set is compared with pressure coefficient distributions as available in maxima for c’p , indicating the post-stall behaviour. A peak in the pressure
literature for Ree3  106 in undisturbed flow conditions in static wind coefficient can be observed at x/c~0.02 for 14 and 24 , close to the
tunnel tests (Timmer and van Rooij, 2003; Timmer, 2010). The mean stagnation point. Also c’p shows a peak for all cases, but at the next
pressure cp is shown together with its standard deviation cp ’. Results in pressure tap at x/c~0.05. This is due to the fluctuating flow downstream
the present dataset compare quite well to results in literature obtained for of the stagnation region.
slightly different angles of attack (Timmer and van Rooij, 2003). For α ¼
4 , the cp shows a monotonic trend which indicates the absence of

9
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 10. Standard deviation of pressure coefficient.

4.1. Effect of turbulence in absence of adverse pressure gradients 5. Effect of turbulence intensity

In literature, aerofoils are considered immune from turbulence effects This Section reports on statistical moments for the pressure time
as they are designed to work mostly in attached flow conditions. When histories for the cases 1 to 3, where the integral length scale Lu is kept
the flow is attached turbulence is more likely to trigger early transition, constant to study the effect of turbulence intensity Iu (Table 2). As
rather than enhance mixing with the boundary/shear layer. This holds mentioned earlier, reference is only made to the ‘suction’ side, with re-
especially when an adverse pressure gradient is absent, i.e. at α ¼ 4 . sults on the ‘pressure’ side only included for the mean pressure coeffi-
Fig. 6 and Fig. 9 both show the chord-wise mean pressure coefficient cient cp . Fig. 10 shows the standard deviation c’p , while Fig. 11 shows the
distribution in turbulence. In fact, the ‘pressure’ side of the aerofoil does 99th percentile of the instantaneous cp . It is found that the behaviour is
not suggest that turbulence is affecting the pressure distribution in any of strongly influenced by on the ratio Lu/c, in particular if Lu/c < 0.9, 0.9 <
the angles of attack or configurations. As for α ¼ 4 Figs. 6 and 9 also Lu/c < 1.5, and Lu/c > 2.5. Each case refers to the relevant row of Figs. 6,
show that at the ‘suction’ side of the aerofoil the variability of the pres- Figs. 7 and 8, to ease the consultation of data. Each column refers instead
sure coefficient is limited, with only a mismatch of ~20% when L/c~1 to the angle of attack.
and I > 7.5%. This matches results from literature for aerofoils in
attached flow conditions (Hoffmann, 1991; Amandolese and Szechenyi,
2004; Maldonado et al., 2015). Results about the ‘pressure’ side of the 5.1. Case 1 - Small length scale Lu/c < 0.9
aerofoil are omitted in the following, although they are used to calculate
the force coefficients. In literature, there is agreement upon the fact that a flow with a small
integral length scale is more likely to enhance mixing and be entrained

10
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 11. 99th percentile of cp for the three angles of attack and Cases 4 to 6.

into the boundary or shear layer of a bluff body (Zdravkovich, 1997). In 5.2. Case 2 - Length scale close to chord length 0.9<Lu/c < 1.5
fact, this does not seem the case for the present dataset. Fig. 6 shows no
evident effect of turbulence for cp at 4 and 14 angle. As Lu/c~0.9 cp As Lu ~ c a fairly clearer pattern on the effect of Iu is noticeable.
increases in the attached flow region, however the overall behaviour is Turbulence significantly delays separation, as cp shows an increased peak
not modified, as also the ‘pressure’ side cp are higher in magnitude. At at the leading edge at all angles of attack. The amplitude of this peak is
α ¼ 24 , stall delay takes place and an increased peak in cp is noticeable directly correlated with Iu. For Iu < 5%, the effect of turbulence seems
close to the leading edge for Iu~15%. As for the standard deviation c’p , negligible, as no significant peak close to the leading edge is noticeable in
the effect is more evident. Turbulence increases fluctuations deopending th
cp , c’p , or c99
p . A critical turbulence intensity Icr might be defined under
on Iu. However, a smaller length scale damps this effect as fluctuations which any effect of turbulence is negligible, regardless of α or Lu/c. A
are reduced for all angles of attack. The only exception are Lu/c~0.9, that significant effect is instead noticeable for Iu > 5%, and a direct coinci-
sees a strong increase in c’p , and Iu~15%, where the smaller Lu/c~0.64 th
dence between turbulence intensity and c’p and c99 p can be stipulated.
limits c’p . However, the different separation pattern is only present for
For x/c > 0.5 all values tend to collapse on the same curve, regardless of
Iu~15%, highlighting that stall delay is governed by turbulence intensity,
the actual value of Iu. At α ¼ 24 stall delay takes place for Iu > Icr. As Lu/
provided than the integral length scale is not smaller than a critical value,
c~1.4, fluctuations seem to be slightly reduced regardless of Iu. This is
which in this case seems Lcr/c~0.7.
consistent with the definition of a critical range for Lu/c around unity.
Fig. 8 leads to analogous conclusions as for Fig. 7, where the role of
Lu/c as a trigger to the effect of turbulence is evident. When Lu/c < 0.9
5.3. Case 3 - Large length scale Lu/c > 2.5
and Iu < 5% turbulence actually damps fluctuations, with lower values
than the undisturbed case.
Contrary to indications in literature, the effect of turbulence is still

11
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 12. Effect of turbulence on Separation Point xS for 14 (top row) and 24 (bottom row). Results are marked per the integral length scale of the inflow, as specified.

present for Lu/c~3. In fact, such a ‘large’ length scale has not been and results are closer to the undisturbed case for larger length scales.
achieved in previous studies, and the possible role of Lu/c might have
been misinterpreted. Fig. 6 shows two distinct behaviours depending on 6.2. Case 5 - Medium turbulence intensity 7<Iu<10%
Iu at 24 , while for other α the effect seems negligible. For Iu~7% no
significant deviation is noticeable, suggesting that the interaction with As Iu~7–8%, Lu/c does not seem to affect cp appreciably. In particular
turbulence occurs at the shear layer, while for Iu~15% separation is still for Lu/c < 0.9 and Lu/c > 2.5, turbulence intensity does not modify the
th
delayed, with enhanced c’p and c99 p . In the presence of a strong adverse aerodynamic behaviour. This reassures on the possibility of defining a Icr
pressure gradient, these results might indicate that Lu affects the critical in dependence of Lu/c. When Lu/c~1, a mild effect on cp can be noticed,
intensity Icr under which effects of turbulence become negligible. From with a peak close to the leading edge at 24 . c’p confirms the trend
the probability density functions (PDF) of pressure coefficients (not observed for cp , with Lu/c~1 enhancing fluctuations.
plotted for the sake of brevity), it might be argued that the entrainment of
the flow in the boundary layer is less efficient for large Lu. In fact, tur-
6.3. Case 6 - High turbulence intensity Iu >14%
bulence normalises the PDF distributions and the match increases for Lu/
c~1.
For Iu > 14% the effect of turbulence does not seem to depend on Lu/c,
as evident in Fig. 9 at 24 . However, cp seems to indicate that separation
6. Effect of integral length scale of turbulence
is delayed most due to larger length scales. The observed effects of the
length scale holding for Iu~7–8% on c’p seems to be milder, with no
This Section reports on statistical moments for the pressure time
noticeable difference for Lu/c~1. However, for Lu/c < 0.6 the length
histories for the cases 4 to 6, where Iu is kept constant to study the effect
th scale acts as a damper to the effect of turbulence, with reduced fluctua-
of Lu/c (Table 3). Fig. 9 shows cp , Fig. 10 shows c’p , and Fig. 11 c99
p , as in tions. At 4 and 14 , Lu/c~1 causes higher fluctuations in both c’p and
the previous section. th
c99
p .

6.1. Case 4 - Low turbulence intensity Iu<5% 7. Position of the separation point

Fig. 9 confirms that at Iu < 5% no appreciable difference with the In absence of measurements about the boundary layer profile over the
th
undisturbed case is noticeable in cp , c’p or c99
p at all angles of attack. Lu/c aerofoil or its skin friction, as in the present case, it is not possible to give
acts as a damper to the fluctuations if Lu/c < 0.9 at 24 . Consistent to an accurate estimate of the separation position xS. However, a necessary
findings shown in the previous section, Lu does affect higher moments, and not sufficient condition for separation is the negligibility of the

12
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 13. Normalised mean (top row) and standard deviation (bottom row) of force coefficients for α ¼ 4 against the turbulence parameter I(L/c). Results are marked
per the integral length scale of the inflow, as specified.

chord-wise pressure derivative dcp =dxe0. Icr e7 % (3)


The Stratford separation criterion is based on dcp =dx and it imposes a
condition on the arrangement of the turbulent boundary layer around the when the effect of turbulence steadily delays separation, regardless of Lu/
body at separation by comparing it to that of a flat plate (Stratford, 1959; c. If Iu < 7 %, the effect of turbulence is also affected by the length scale
Cebeci et al., 1972). It states: and only present when Lu ~ c. At the largest length scales, stall delay is
reduced for Iu < Icr, while for Iu~15% the larger length scale does not
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  1=10
xS  cp x dcp dx 106 Rex e0:4  0:5 (1) seem to affect the enhanced behaviour. Fig. 12 seems to point at an
asymptotic behaviour for Iu ðLu =cÞn e0:2, where the effect of turbulence
seems to persist for larger length scales. This behaviour is noticeable for
where Rex ¼ xU=ν, and 0.4-0.5 is a threshold based on the von-Karman
both α ¼ 14 and 24 . This seems to reassure on the role of the length
constant κ ¼ 0:41.
scale as an enhancement or a damper of the effects of turbulence, pro-
Fig. 12 shows the separation point as found with equation (1) at 14
vided that a critical turbulence intensity is reached to trigger any effect.
and 24 . As the effect of turbulence might depend on a combined effect of
Lu and Iu, a turbulence parameter is introduced
8. Force coefficients
Iu ðLu =cÞn (2)
The force coefficients are calculated by integration over the aerofoil
where n is an exponent normally taken as unitary (Bearman and Morel, surface of the pressure coefficient. The coefficient of Lift cL , Drag cD and
1983). Moment cM are defined respectively perpendicular and parallel to the
At 14 separation varies monotonically from xS/c~0.7 to ~0.9 with incoming flow field, and the resultant around a pole taken at x ¼ c/4 as
both Iu and Lu/c, with the effect triggered by the turbulence intensity and customary in aeronautics. In this work, cD should be referred to as form
enhanced or damped depending on the length scale., In fact, as Lu/c > drag, as the skin friction coefficient is not included in its calculation.
2.3, the effect seems to reach an asymptotic behaviour. When α ¼ 24 , Pressure time histories are integrated to obtain force time histories,
separation is also delayed from ~0.35 up to ~0.6, however the effect whose statistical moments are considered in the following with reference
seems uncorrelated with the length scale, while it grows proportionally to the stall mechanism.
to the turbulence intensity. The turbulence parameter Iu ðLu =cÞn shows
that for Lu/c > 2.3, stall delay might not be influenced strictly by tur- 8.1. Pre-stall α ¼ 4
bulence intensity, as other mechanisms might occur with a growing
adverse pressure gradient. From Fig. 12, a critical turbulence intensity Fig. 13 shows the force coefficients referred to the undisturbed inflow
might be defined as ðunÞ ðunÞ
cL =cL and their standard deviation c’L =cL . Values are referred to the
undisturbed condition to stress the role of turbulence on the aerodynamic

13
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 14. Normalised force coefficients for α ¼ 14 against turbulence parameter I(L/c). Results are marked per the integral length scale of the inflow, as specified.

performance. At α ¼ 4 results seems to indicate a bifurcated behaviour, 8.3. Full-stall α ¼ 24


with a part of the dataset not influenced by turbulence, and a part
strongly influenced despite the lack of separation for this angle of attack. Fig. 15 shows the force coefficients at α ¼ 24 . The bifurcated
ðunÞ behaviour is now even more evident, with a consistent monotonic in-
The undisturbed force coefficients cL =cL ¼ 1 are plotted along turbu-
lent ones for comparison. Data points are coloured following the length crease with the parameter Iu ðLu =cÞn , showing the beneficial effect of
scale ranges indicated in each figure. turbulence in delaying stall. Part of the dataset at Iu ðLu =cÞn < 0:2 is more
In general, as Lu/c~1 a strong effect of turbulence is present with an insensitive to turbulence, depending on Lu/c. However, the behaviour
increase in Lift and Moment coefficient, and a reduction of Drag. The now suggests that turbulence is non-negligible in all configurations.
fluctuation of the force coefficients increases steadily, with data points Unlike in Fig. 14, the effect of turbulence on Drag is similar to that on Lift,
having Lu/c~1, deviating from this tendency, as discussed in the previ- with cD being weakly affected unless Lu/c~1. The same behaviour can be
ous. Points with 1.2 < Lu/c < 2 show a dual behaviour, either behaving observed for cM. As for the force coefficient fluctuations, turbulence
similarly to Lu/c~1 and enhancing greatly the effect or damping it causes an increase in fluctuations similar to that found at α ¼ 4 , while at
depending on their turbulence intensity. α ¼ 14 the increase was milder. This might depend on the dual role of
turbulence in either triggering transition or enhancing mixing in the
boundary or shear layers.
8.2. Post-stall α ¼ 14

Fig. 14 shows the variability of force coefficients against the turbu- 8.4. Probability density function
lence parameter for α ¼ 14 . As observed for α ¼ 4 , cL shows a clear
bifurcated behaviour. Depending on Lu/c, part of the data is not influ- Fig. 16 shows the PDFs of the zero-mean Lift coefficient cL =c’L nor-
enced by turbulence for Iu ðLu =cÞn < 0:2. Within this range a part of the malised by its standard deviation for α ¼ 24 . Colours of graphs in Fig. 16
data around Lu/c~1 shows an enhanced effect of turbulence. For are referred to respectively Table 2 for the upper row and Table 3 for the
Iu ðLu =cÞn > 0:2 no effect of turbulence is noticeable. cD seems instead to lower row of plots. The undisturbed case shows a marked negative
increase linearly with Iu ðLu =cÞn . In literature, turbulence is always asso- skewed behaviour that improves with ambient turbulence consistently
ciated with an increase in Drag (Milan et al., 2013). However, the form with all configurations of Iu and Lu/c. For Iu~15% seems to show a shift
drag drops for Iu ðLu =cÞn < 0:2. cM shows a similar behaviour to cL , with a due to the large length scale from a negative skewness to a positive
dual response to turbulence enhanced if Lu ~ c. As at α ¼ 4 , the fluc- skewness. This might depend on large-scale fluctuations, which affect the
tuation of the force coefficients monotonically increase with the turbu- angle of attack and are visible in the Lift coefficient. The strongly nega-
lence parameter, with an enhanced effect when Lu/c~1. tively skewed undisturbed signal presents positive fat-tails, which

14
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 15. Normalised mean and standard deviation of force coefficients against turbulence parameter I(L/c). Results are marked per the integral length scale of the
inflow, as specified.

disappear in turbulence. However, a mild effect of the integral length - Lu contributes together with Iu to the effect of turbulence, damping or
scale of turbulence can be noticed. In case 1 where the turbulence in- enhancing the effect, provided that a sufficiently high Icr is present;
tensity is low, the signal is still negatively skewed, and it seems that a - When Lu > c, the effect of turbulence is still present in full-stall at α ¼
trend is noticeable for the normalisation is larger at Lu/c~1. The con- 24 , and it shows a proportionality to the parameter Iu (Lu/c)n, taken
clusions made for α ¼ 24 also apply to other angles of attack. with the exponent n ¼ 1;
- As Lu < c, the effect of turbulence is much weaker even at highest
9. Conclusions turbulence intensities, seemingly refuting the generalised opinion
that a small length scale is more likely to influence the behaviour of
In this work, novel results on the effect of the integral length scale of the boundary or shear layer;
turbulence on the aerodynamic behaviour of a wind turbine aerofoil are - When Lu ~ c the effects of turbulence are magnified, triggering a
presented in terms of surface pressure, separation length variation, and bifurcated behaviour for a critical value of the turbulence parameter
force coefficients. In particular, the possibility of controlling the setup of Iu ðLu =cÞncr < 0:2;
the turbulent inflow to vary statistics independently has allowed iden- - The turbulence parameter Iu ðLu =cÞncr ¼ 0:2 defines a limit for the
tifying mutual effects and interactions between Lu and Iu. Free stream bifurcated effect of turbulence. For Iu ðLu =cÞn < 0:2, the effects of
turbulence is found to modify the aerodynamic performance even if the turbulence are either minimum if Lu/c < 1 or at the contrary maxi-
integral length scale is around three times larger than the chord of the mised if Lu/c~1. For Iu ðLu =cÞn > 0:2, the effect of turbulence is either
aerofoil, in full-stalled conditions. The definition of a turbulence absent or present consistently with the behaviour at small scale
parameter Iu ðLu =cÞn is found to be representative of the complex inter- depending on the angle of attack. The behaviour is analogous for all
action of Lu and Iu in defining the possible mechanism which activates. force coefficients cL, cD and cM, and separation point xS/c.
The main finding of this work is that wind turbine aerofoils are not - The Drag coefficient might be reduced due to turbulence effects if
insensitive to the integral length scale of turbulence as previously Iu ðLu =cÞn < 0:2 and Lu/c < 1;
thought. FST is able to modify the aerodynamic behaviour provided that - Turbulence effects might be neglected when Iu < Icr ¼ 5% indepen-
a critical turbulence intensity Icr is reached whose value depends on Lu , dently of the Lu/c value, if cp or c’p are looked at;
which can damp or enhance it depending on the ratio Lu/c. - Turbulence has the effect of normalising statistics of pressure and
This setup is interesting as most small wind turbine blades placed in force coefficients independently from Iu or Lu, suggesting that tur-
the built environment might experience a similar configuration for the bulence helps the stationarity of performance;
inflow, with Iu~15% and Lu/c~3–10 (Sarki  c Glumac, Hemida and
H€offer, 2018). Several conclusions can be drawn from the presented Further research might include aerofoils having different character-
results: istics (such as a different thickness) or boundary layer treatment such as

15
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Fig. 16. Probability Density Function of Lift coefficient for cases as in Tables 2 and 3.

tripping or leading edge roughness to understand how to limit or exploit the run of the experiment, and to Arnaud Fabbri for the realisation of the
the effect of turbulence and improve aerodynamic performance. Never- aerofoil model through 3D printing.
theless, higher Lu/c values need to be achieved in wind tunnel or nu-
merical testing to have a better comprehension of the behaviour and References
understand in what proportion turbulence contributes to the unsteady
variability of the angle-of-attack or the interaction with the boundary or Al-Abadi, A., et al., 2016. Turbulence impact on wind turbines: experimental
shear layer. investigations on a wind turbine model. J. Phys.: Con Series. IOP Pub 753 (3).
https://doi.org/10.1088/1742-6596/753/3/032046, 032046.
Albers, A., et al., 2007. Influence of meteorological variables on measured wind turbine
Author contribution power curves. Milan, Italy. In: Proceedings of the European Wind Energy Conference
(EWEC). Available at: https://www.researchgate.net/publication/229015286.
(Accessed 8 May 2020).
The manuscript has been read and approved by all named authors. Amandolese, X., Szechenyi, E., 2004. Experimental study of the effect of turbulence on a
The order of authors listed in the manuscript has been approved by all section model blade oscillating in stall. Wind Energy 7 (4), 267–282. https://doi.org/
10.1002/we.137.
named authors.
Arnfield, a.J., 2003. Two decades of urban climate research: a review of turbulence,
exchanges of energy and water, and the urban heat island. Int. J. Climatol. 23 (1),
1–26. https://doi.org/10.1002/joc.859.
Declaration of competing interest Bak, C., 2007. Sensitivity of key parameters in aerodynamic wind turbine rotor design on
power and energy performance. J. Phys. Conf.. IOP Publishing, 75 (1). https://
doi.org/10.1088/1742-6596/75/1/012008, 012008.
The authors declare that they have no known competing financial
Balduzzi, F., et al., 2012. Feasibility analysis of a Darrieus vertical-axis wind turbine
interests or personal relationships that could have appeared to influence installation in the rooftop of a building. Appl. Energy 97, 921–929. https://doi.org/
the work reported in this paper. 10.1016/j.apenergy.2011.12.008. Elsevier Ltd.
Battisti, L., et al., 2018. Small wind turbine effectiveness in the urban environment.
Renew. Energy 129, 102–113. https://doi.org/10.1016/j.renene.2018.05.062.
Acknowledgements Elsevier Ltd.
Bearman, P.W., Morel, T., 1983. Effect of free stream turbulence on the flow around bluff
bodies. Prog. Aerospace Sci. Pergamon 20 (2–3), 97–123. https://doi.org/10.1016/
The authors acknowledge the support of the European Commission’s 0376-0421(83)90002-7.
Framework Program “Horizon 2020”, through the Marie Skłodowska- Buresti, G., 2012. Elements of Fluid Dynamics. Imperial College Press, London. Available
Curie Innovative Training Network (ITN) “AEOLUS4FUTURE – Efficient at: https://books.google.com/books?id¼N6TNpwAACAAJ&amp;pgis¼1. (Accessed
27 January 2016).
harvesting of the wind energy” (H2020-MSCA-ITN-2014: Grant agree-
Carbo Molina, A., Bartoli, G., De Troyer, T., 2017. Wind tunnel testing of small vertical-
ment no. 643167) to the present research project. Also, the Cost Action Axis wind turbines in turbulent flows. In: Procedia Engineering, X International
TU1804 WinerCost – “Wind Energy to enhance the concept of Smart Conference on Structural Dynamics, EURODYN 2017. Elsevier, Rome,
cities” is gratefully acknowledged, for providing the possibility of a Short pp. 3176–3181. https://doi.org/10.1016/j.proeng.2017.09.518.
Cebeci, T., Mosinskis, G.J., Smith, A.M.O., 1972. Calculation of separation points in
Term Scientific Mission, used to conduct this experiment. A special thank incompressible turbulent flows. J. Aircraft 9 (9), 618–624. https://doi.org/10.2514/
goes to François Rigo for taking part in the construction of the setup and 3.59049.

16
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Conan, B., 2012. Wind resource accessment in complex terrain by wind tunnel modelling. Micallef, D., van Bussel, G., 2018. A review of urban wind energy research: aerodynamics
Orleans. Available at: http://www.theses.fr/2012ORLE2067. (Accessed 31 January and other challenges. Energies 11 (9), 2204. https://doi.org/10.3390/en11092204.
2018). Multidisciplinary Digital Publishing Institute.
Devenport, W., et al., 2010. Aeroacoustic Testing of Wind Turbine Airfoils: February 20, Milan, P., W€achter, M., Peinke, J., 2013. ‘Turbulent character of wind energy.’, Physical
2004 - February 19, 2008. Blacksburg, Virginia. Available at: http://www.osti.gov/ review letters. Am Phys Soc 110 (13), 138701. https://doi.org/10.1103/
bridge. (Accessed 5 April 2019). PhysRevLett.110.138701.
Devinant, P., Laverne, T., Hureau, J., 2002. Experimental study of wind-turbine airfoil Miley, S., 1982. A catalog of low Reynolds number airfoil data for wind turbine
aerodynamics in high turbulence. J. Wind Eng. Ind. Aerod. 90 (6), 689–707. https:// applications. Available at: http://wind.nrel.gov/public/library/3387.pdf. (Accessed
doi.org/10.1016/S0167-6105(02)00162-9. 8 June 2016).
Emeis, S., 2014. Current issues in wind energy meteorology. Meteorol. Appl. 21 (4), Mouzakis, F., Morfiadakis, E., Dellaportas, P., 1999. Fatigue loading parameter
803–819. https://doi.org/10.1002/met.1472. identification of a wind turbine operating in complex terrain. J. Wind Eng. Ind.
Evans, S.P., et al., 2017. The suitability of the IEC 61400-2 wind model for small wind Aerod. 82 (1), 69–88. https://doi.org/10.1016/S0167-6105(98)00211-6.
turbines operating in the built environment. Renew. Energy Environ. Sustain. 2, 31. Mücke, T., Kleinhans, D., Peinke, J., 2011. Atmospheric turbulence and its influence on
https://doi.org/10.1051/rees/2017022. D. Goodfield. the alternating loads on wind turbines. Wind Energy 14 (2), 301–316. https://
Frandsen, S.T., 2007. Turbulence and Turbulence-Generated Structural Loading in Wind doi.org/10.1002/We.422.
Turbine Clusters. Mueller, T.J., et al., 1983. The influence of free-stream disturbances on low Reynolds
Freudenreich, K., et al., 2004. Reynolds number and roughness effects on thick airfoils for number airfoil experiments. Exp. Fluid 1, 3–14. https://doi.org/10.1007/
wind turbines. Wind Eng. 28 (5), 529–546. https://doi.org/10.1260/ BF00282261.
0309524043028109. SAGE PublicationsSage UK: London, England. Nakamura, Y., 1993. Bluff-body aerodynamics and turbulence. J. Wind Eng. Ind. Aerod.
Haan, F.L., Kareem, A., Szewczyk, A.A., 1998. The effects of turbulence on the pressure 49 (1–3), 65–78. https://doi.org/10.1016/0167-6105(93)90006-A.
distribution around a rectangular prism. J. Wind Eng. Ind. Aerod. 77 (78), 381–392. Nakamura, Y., Ohya, Y., 2006. The effects of turbulence on the mean flow past square
https://doi.org/10.1016/S0167-6105(98)00158-5. rods. J. Fluid Mech. 137 (1), 331. https://doi.org/10.1017/S0022112083002438.
Hand, M., Simms, D., Fingersh, L., 2001. Unsteady aerodynamics experiment phase VI: Cambridge University Press.
wind tunnel test configurations and available data campaigns. Available at: https:// Nakamura, Y., Ohya, Y., Ozono, S., 1988. The effects of turbulence on bluff-body mean
www.researchgate.net/profile/Scott_Larwood2/publication/255198729_Unsteady flow. J. Wind Eng. Ind. Aerod. 28 (1), 251–259. https://doi.org/10.1016/0167-
_Aerodynamics_Experiment_Phase_VI_Wind_Tunnel_Test_Configurations_and_Availa 6105(88)90121-3. Elsevier.
ble_Data_Campaigns/links/5457d2540cf26d5090ab52a1.pdf. (Accessed 30 May Nakamura, Y., Ohya, Y., Ozono, S., 2012. The effects of turbulence on bluff-body mean
2016). flow. Adv. Wind. Available at: https://books.google.co.uk/books?hl¼it&amp;lr¼&a
Hansen, M.H., 2007. Aeroelastic instability problems for wind turbines. Wind Energy 10 mp;id¼-Q6RfEiLKVAC&amp;oi¼fnd&amp;pg¼PA251&amp;ots¼bPGaZKAGK0&a
(6), 551–577. https://doi.org/10.1002/we.242. mp;sig¼PwxVCOlBloBI0EXkrWEpUrzQCeE. (Accessed 7 June 2016).
Hansen, M.O.L., et al., 2006. State of the art in wind turbine aerodynamics and Ohya, Y., 2004. Drag of circular cylinders in the atmospheric turbulence. Fluid Dynam.
aeroelasticity. Prog. Aero. Sci. 42 (4), 285–330. https://doi.org/10.1016/ Res. 34 (2), 135–144. https://doi.org/10.1016/j.fluiddyn.2003.10.002.
j.paerosci.2006.10.002. Pagnini, L.C., Burlando, M., Repetto, M.P., 2015. Experimental power curve of small-size
Hau, E., 2013. Wind Turbines: Fundamentals, Technologies, Application, Economics, wind turbines in turbulent urban environment. Appl. Energy 154, 112–121. https://
third ed. Springer-Verlag, Berlin Heidelberg. https://doi.org/10.1007/3-540-29284- doi.org/10.1016/j.apenergy.2015.04.117. Elsevier Ltd.
5. Pagnini, L., Piccardo, G., Repetto, M.P., 2018. Full scale behavior of a small size vertical
Hoffmann, J.A., 1991. Effects of freestream turbulence on the performance characteristics axis wind turbine. Renew. Energy 127, 41–55. https://doi.org/10.1016/
of an airfoil. AIAA J. 29 (9), 1353–1354. https://doi.org/10.2514/3.10745. j.renene.2018.04.032. Elsevier Ltd.
Huang, R.F., Lee, H.W., 1999. Effects of freestream turbulence on wing-surface flow and Rezaeiha, A., Montazeri, H., Blocken, B., 2018. Towards optimal aerodynamic design of
aerodynamic performance. J. Aircraft 36 (6), 965–972. https://doi.org/10.2514/ vertical axis wind turbines: impact of solidity and number of blades. Energy 165,
2.2537. AIAA. 1129–1148. https://doi.org/10.1016/J.ENERGY.2018.09.192. Pergamon.
IEC 61400-2:2013, 2013. Wind turbines - Part 2: small wind turbines. Available at: Sareen, A., Sapre, C.A., Selig, M.S., 2012. Effects of leading-edge protection tape on wind
https://webstore.iec.ch/publication/5433. (Accessed 18 November 2019). turbine blade performance. Wind Eng. 36 (5), 525–534. Available at: https://m-selig
Ishugah, T.F., et al., 2014. Advances in wind energy resource exploitation in urban .ae.illinois.edu/pubs/SareenSapreSelig-2012-WindEngineering-WPT.pdf. (Accessed
environment: a review. Renew. Sustain. Energy Rev. 37, 613–626. https://doi.org/ 5 April 2019).
10.1016/J.RSER.2014.05.053. Pergamon. 
Sarkic Glumac, A., Hemida, H., H€ offer, R., 2018. Wind energy potential above a high-rise
Kc, A., Whale, J., Urmee, T., 2019. Urban wind conditions and small wind turbines in the building influenced by neighboring buildings: an experimental investigation. J. Wind
built environment: a review. Renew. Energy 131, 268–283. https://doi.org/10.1016/ Eng. Ind. Aerod. 175, 32–42. https://doi.org/10.1016/J.JWEIA.2018.01.022.
J.RENENE.2018.07.050. Pergamon. Elsevier.
Kelly, M., et al., 2014. Probabilistic meteorological characterization for turbine loads. Seddighi, M., Soltani, M., 2007. The influence of free stream turbulence intensity on the
J. Phys. Conf. 524 (1), 012076 https://doi.org/10.1088/1742-6596/524/1/012076. unsteady behavior of a wind turbine blade section. In: 45th AIAA Aerospace Sciences
IOP Publishing. Meeting and Exhibit. American Institute of Aeronautics and Astronautics, Reston,
Kistler, A.L., Vrebalovich, T., 2006. Grid turbulence at large Reynolds numbers. J. Fluid Virigina. https://doi.org/10.2514/6.2007-630.
Mech.. Cambridge University Press 26 (01), 37. https://doi.org/10.1017/ Sicot, C., Devinant, P., et al., 2006. Experimental study of the effect of turbulence on
S0022112066001071. horizontal axis wind turbine aerodynamics. Wind Energy 9 (4), 361–370. https://
Kosasih, B., Saleh Hudin, H., 2016. Influence of inflow turbulence intensity on the doi.org/10.1002/we.184. John Wiley & Sons, Ltd.
performance of bare and diffuser-augmented micro wind turbine model. Renew. Sicot, C., Aubrun, S., et al., 2006. Unsteady characteristics of the static stall of an airfoil
Energy 87, 154–167. https://doi.org/10.1016/j.renene.2015.10.013. subjected to freestream turbulence level up to 16%. Exp. Fluid 41 (4), 641–648.
van Kuik, G.A.M., et al., 2016. ‘Long-term research challenges in wind energy – a research https://doi.org/10.1007/s00348-006-0187-9.
agenda by the European Academy of Wind Energy’. Wind Energy Sci 1 (1), 1–39. Sicot, C., et al., 2008. Rotational and turbulence effects on a wind turbine blade.
https://doi.org/10.5194/wes-1-1-2016. Investigation of the stall mechanisms. J. Wind Eng. Ind. Aerod. 96 (8–9), 1320–1331.
Kurian, T., Fransson, J.H.M., 2009. Grid-generated turbulence revisited. Fluid Dynam. https://doi.org/10.1016/j.jweia.2008.01.013.
Res. 41 (2), 021403 https://doi.org/10.1088/0169-5983/41/2/021403. IOP Sørensen, J.N., 2011. Aerodynamic aspects of wind energy conversion. Annu. Rev. Fluid
Publishing. Mech. 43, 427–448. https://doi.org/10.1146/annurev-fluid-122109-160801.
Li, Q., et al., 2016. Effect of turbulent inflows on airfoil performance for a Horizontal Axis St Martin, C.M., et al., 2016. Wind Turbine Power Production and Annual Energy
Wind Turbine at low Reynolds numbers (Part II: Dynamic pressure measurement). Production Depend on Atmospheric Stability and Turbulence. https://doi.org/
Energy 112, 574–587. https://doi.org/10.1016/j.energy.2016.06.126. 10.5194/wes-2016-21.
Lindeboom, R.C.J., 2010. Determination of unsteady loads on a DU96W180 airfoil with Stathopoulos, T., et al., 2018. Urban wind energy: some views on potential and
actuated flap using Particle Image Velocimetry. Available at: https://repository.tud challenges. J. Wind Eng. Ind. Aerod.. Elsevier 179, 146–157. https://doi.org/
elft.nl/islandora/object/uuid:f3bb4ee0-0bd6-402a-858f-45160d5ef1dc?collectio 10.1016/J.JWEIA.2018.05.018.
n¼education. (Accessed 5 April 2019). Stratford, B.S., 1959. The prediction of separation of the turbulent boundary layer.
Lyon, C.A., Selig, M.S., Broeren, A.P., 1997. Boundary layer trips on airfoils at low J. Fluid Mech.. Cambridge University Press 5 (01), 1. https://doi.org/10.1017/
Reynolds numbers. In: 35th Aerospace Sciences Meeting and Exhibit. American S0022112059000015.
Institute of Aeronautics and Astronautics Inc, AIAA. https://doi.org/10.2514/ Sunderland, K., et al., 2013. Small wind turbines in turbulent (urban) environments: a
6.1997-511. consideration of normal and Weibull distributions for power prediction. J. Wind Eng.
Maldonado, V., et al., 2015. The role of free stream turbulence with large integral scale on Ind. Aerod. 121, 70–81. https://doi.org/10.1016/j.jweia.2013.08.001. Elsevier.
the aerodynamic performance of an experimental low Reynolds number S809 wind Suryadi, A., Herr, M., 2015. Wall pressure spectra on a DU96-W-180 profile from low to
turbine blade. J. Wind Eng. Ind. Aerod. 142, 246–257. https://doi.org/10.1016/ pre-stall angles of attack. In: 21st AIAA/CEAS Aeroacoustics Conference. American
j.jweia.2015.03.010. Elsevier. Institute of Aeronautics and Astronautics, Reston, Virginia. https://doi.org/10.2514/
Mary, I., Sagaut, P., 2002. Large eddy simulation of flow around an airfoil near stall. AIAA 6.2015-2688.
J. 40 (6), 1139–1145. https://doi.org/10.2514/2.1763. Swalwell, K., Sheridan, J., 2001. The effect of turbulence intensity on stall of the NACA
Matyushenko, A.A., Kotov, E.V., Garbaruk, A.V., 2017. Calculations of flow around 0021 aerofoil. In: 14th Australasian. Available at: http://people.eng.unimelb.edu.
airfoils using two-dimensional RANS: an analysis of the reduction in accuracy. St. au/imarusic/proceedings/14/FM010235.PDF. (Accessed 7 October 2016).
Petersburg Polytech. Univ. J: Phys. Mathemat 3 (1), 15–21. https://doi.org/10.1016/ Swalwell, K., Sheridan, J., Melbourne, W., 2004. The effect of turbulence intensity on
J.SPJPM.2017.03.004. No longer published by Elsevier. performance of a NACA 4421 airfoil section. In: 42nd AIAA Aerospace. Available at:
http://arc.aiaa.org/doi/pdf/10.2514/6.2004-665. (Accessed 8 June 2016).

17
G. Vita et al. Journal of Wind Engineering & Industrial Aerodynamics 204 (2020) 104235

Tabrizi, A.B., et al., 2015. Extent to which international wind turbine design standard, Toja-Silva, F., et al., 2018. A review of computational fluid dynamics (CFD) simulations of
IEC61400-2 is valid for a rooftop wind installation. J. Wind Eng. Ind. Aerod. 139, the wind flow around buildings for urban wind energy exploitation. J. Wind Eng. Ind.
50–61. https://doi.org/10.1016/j.jweia.2015.01.006. Elsevier. Aerod. 180, 66–87. https://doi.org/10.1016/J.JWEIA.2018.07.010. Elsevier.
Tang, H., et al., 2019. ‘Wake effect of a horizontal Axis wind turbine on the performance Traub, L.W., 2011. Experimental investigation of the effect of trip strips at low Reynolds
of a downstream turbine’, energies. MDPI AG 12 (12), 2395. https://doi.org/ number. J. Aircraft 48 (5), 1776–1784. https://doi.org/10.2514/1.C031375.
10.3390/en12122395. Vermeer, L.J., Sørensen, J.N., Crespo, A., 2003. Wind turbine wake aerodynamics. Prog.
Tangler, J., 2002. The nebulous art of using wind-tunnel airfoil data for predicting rotor Aero. Sci. 39 (6–7), 467–510. https://doi.org/10.1016/S0376-0421(03)00078-2.
performance. ASME 2002 Wind Energy. Available at: http://proceedings.as Vita, G., et al., 2018. Generating atmospheric turbulence using passive grids in an
medigitalcollection.asme.org/proceeding.aspx?articleid¼1572110. (Accessed 30 expansion test section of a wind tunnel. J. Wind Eng. Ind. Aerod. 178, 91–104.
May 2016). https://doi.org/10.1016/j.jweia.2018.02.007. Elsevier.
Timmer, W.A., 2010. Aerodynamic characteristics of wind turbine blade airfoils at high Wagner, R., 2010. Accounting for the Speed Shear in Wind Turbine Power Performance
angles-of-attack. In: TORQUE 2010: the Science of Making Torque from Wind, Measurement. Risø National Laboratory.
pp. 71–78. https://doi.org/10.1533/9780857097286.2.210. Walker, S.L., 2011. ‘Building mounted wind turbines and their suitability for the urban
Timmer, W.A., van Rooij, R.P.J.O.M., 2003. ‘Summary of the Delft university wind scale—a review of methods of estimating urban wind resource’. Energy Build. 43 (8),
turbine dedicated airfoils’. In: ASME 2003 Wind Energy Symposium. ASME, 1852–1862. https://doi.org/10.1016/j.enbuild.2011.03.032.
pp. 11–21. https://doi.org/10.1115/WIND2003-352. Wang, S., et al., 2014. Turbulent intensity and Reynolds number effects on an airfoil at
Toft, H.S., et al., 2016. Uncertainty in wind climate parameters and their influence on low Reynolds numbers. Phys. Fluids 26 (11), 115107. https://doi.org/10.1063/
wind turbine fatigue loads, 90. Renewable Energy, pp. 352–361. https://doi.org/ 1.4901969. AIP Publishing.
10.1016/j.renene.2016.01.010. Zdravkovich, M.M., 1997. Flow around circular cylinders, flow around circular cylinders.
Toja-Silva, F., et al., 2016. An empirical-heuristic optimization of the building-roof Available at: http://www.scopus.com/inward/record.url?eid¼2-s2.0-00307
geometry for urban wind energy exploitation on high-rise buildings. Appl. Energy 44960&amp;partnerID¼tZOtx3y1.
164, 769–794. https://doi.org/10.1016/j.apenergy.2015.11.095. Elsevier Ltd.

18

You might also like