Applied Energy: Francesco Balduzzi, Alessandro Bianchini, Ennio Antonio Carnevale, Lorenzo Ferrari, Sandro Magnani

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Applied Energy 97 (2012) 921–929

Contents lists available at SciVerse ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Feasibility analysis of a Darrieus vertical-axis wind turbine installation


in the rooftop of a building
Francesco Balduzzi, Alessandro Bianchini, Ennio Antonio Carnevale, Lorenzo Ferrari ⇑, Sandro Magnani
‘‘Sergio Stecco’’ Department of Energy Engineering, University of Florence, Via di Santa Marta 3, 50139 Florence, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The renewed interest that is being paid by architects, project developers and local governments to small-
Available online 28 December 2011 size wind turbines is mainly connected to the attractive prospects of future applications in the urban
environment; the delocalized power production of these systems could indeed provide an effective
Keywords: answer to both the growing demand for renewable energy and the increased attention in buildings with
Darrieus a sustainable and low-energy design. In particular, Darrieus vertical-axis wind turbines (VAWTs) are
VAWT being considered as one of the most attractive solutions due to their low visual impact, the reduced
Urban context
acoustic emissions and their better response to a turbulent and skewed oncoming flow. The feasibility
Skewed flow
Microeolic turbine
of this scenario has, however, to be proved yet; in particular, doubts are still connected to the real pro-
ducibility in a complex terrain like the urban one and to the compatibility of microeolic machines with a
densely populated area. On these assumptions, the aim of this work is to critically evaluate the energetic
suitability of a Darrieus VAWT installation in the rooftop of a building in a reference European city. With
this goal in mind, a numerical CFD analysis was carried out to characterize the flow field in the rooftop
area of buildings with different shapes and geometrical proportions: the flow velocity modulus and direc-
tion were calculated for different oncoming wind profiles and the results were projected into a net avail-
able wind distribution in the rooftop of each building. As a second step, in order to provide a reliable
estimation of the real functioning of the turbine in the investigated environment, a specific numerical
model has been developed to account for the effects of a skewed flow on the power performance of
the Darrieus rotor. The results of these analyses were finally combined and synthesized in an energy-
oriented study to evaluate the feasibility of a rooftop installation.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction to this environment is not yet tested both in terms of real produc-
ibility and of structural compatibility with the buildings. As a re-
Increasing interest is being paid to understand where small-size sult, the wind profile in urban locations is quite different from
wind turbines can effectively be exploited to provide delocalized the classical log-law based profile [1], with the zero-velocity height
power in the built environment [1]. The main prospects of a similar shifted up to a peculiar value (displacement, d) which is a function
installation context can be readily understood; in particular, small of the average height of the surrounding buildings (Fig. 1).
rotors positioned at the top of a tall building could theoretically ex- Notwithstanding this aspect, some general principles to identify
ploit a higher zone of the wind profile with respect to that usually the main requirements of a suitable installation site are provided in
exploitable by means of the only turbine tower. Moreover, the en- technical literature (e.g. [1–4]). In detail, from a theoretical point of
ergy could be produced directly where it is needed, with a notable view, wind turbines in the urban environment require buildings
contribution to a sustainable design of new buildings in terms of that are reasonably higher than the average height of the surround-
energy consumption. ing constructions, in order to take advantage from the local flow
Due to the high roughness length of the terrain and the pres- deflection and acceleration [1], but only on condition that peculiar
ence of obstacles characterized by different shapes and permeabil- geometric proportions between the buildings are fulfilled [5].
ity along the flowpath, the wind conditions in urban locations are, Within this context, Darrieus vertical-axis wind turbines
however, very complex and the real adaptability of wind turbines (VAWTs) are increasingly appreciated and often considered as
the most promising solution in the built environment, due to their
very low noise levels and to their reduced sensitivity to a turbulent
⇑ Corresponding author. Tel.: +39 055 4796 570; fax: +39 055 4796 342.
oncoming wind [5–7]. In addition, recent studies [6,8] put in
E-mail addresses: balduzzi@vega.de.unifi.it (F. Balduzzi), bianchini@vega.de.u
evidence that some benefits in terms of power increase can be
nifi.it (A. Bianchini), ennio.carnevale@unifi.it (E.A. Carnevale), ferrari@vega.de.uni
fi.it (L. Ferrari), magnani@vega.de.unifi.it (S. Magnani). obtained from a Darrieus functioning under skewed flow, mainly

0306-2619/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2011.12.008
922 F. Balduzzi et al. / Applied Energy 97 (2012) 921–929

Nomenclature

Acronyms CS roughness constant


AR Blade Aspect Ratio D distance between UB and IB (m)
BEM Blade Element Momentum Dt turbine diameter (m)
DMSV Double Multiple Streamtubes with Variable interference Dtw tower diameter (m)
factors H IB height (m)
ESDU Engineering Science Data Units ^
H mean buildings height (m)
HAWT Horizontal Axis Wind Turbine KS sand-grain roughness (m)
MS Multiple Streamtubes Ht turbine height (m)
VAWT vertical axis wind turbine N number of blades
Nr number of rods
Greeks symbols Ns number of struts
c skew angle (°) V horizontal wind component (m/s)
e turbulent kinetic energy dissipation rate (m2/s3) c blade chord (m)
gel electrical efficiency d displacement (m)
# rotor azimuthal angle (°) h UB height (m)
j Von Karman constant k Turbulent kinetic energy (m2/s2)
r turbine solidity tr rods thickness (m)
u pitch angle of the turbine airfoils (°) ts struts thickness (m)
v sloping angle of the roof (°) u flow velocity (m/s)
u friction velocity (m/s)
Latin symbols yP height of the ground cells centroid (m)
Ah area occupied by the buildings (m2) z0 roughness length (m)
Aj turbine projected area (m2)
Cl turbulence model constant

Reynolds-Averaged Navier–Stokes (RANS) calculation for incom-


pressible flow without buoyancy effects.
From a fluid-dynamic point of view, the assumption of a 2D
behaviour of the flux that invests the turbine implies that the
streamlines are assumed to follow the same path in each plane per-
pendicular to the building façade (i.e. parallel to the rotational axis of
the VAWT). This reasonable hypothesis, often applied in numerical
studies of the built environment (e.g. [5, 14–16]) remarkably simpli-
fies the numerical model, providing a notable reduction of the com-
putational cost as well as enabling a finer in-plane discretization.
The pressure–velocity coupling was made with the SIMPLE
algorithm and the convergence to the final steady-state was as-
sessed with a maximum amount of 3  103 time steps, which were
Fig. 1. Wind profile in the internal boundary layer in a built environment.
always sufficient to reach the target final computed residual of
1  105 (see [14–16]).
connected to the possibility of exploiting an increased swept area
The wall boundary conditions at the bottom of the computa-
(projected perpendicularly to the mainstream direction).
tional domain were based on the standard wall functions [10] with
In this work, a wide-ranging analysis was performed in order to
the sand-grain based roughness modification [11]. The roughness
evaluate the feasibility of a Darrieus turbine installation in the
effect was taken into account by imposing suitable values of the
built environment. As a first step, CFD simulations were carried
sand-grain roughness (KS) and the roughness constant (CS, with CS
out to outline some general tendencies regarding the influence of
= 0–1) which can be determined from the roughness length (z0) by
the building geometry on the flow conditions in the rooftop area;
Eq. (1) (see [12]):
moving from the numerical results, the net available wind distri-
butions in the rooftop of two study cases were selected. As a sec- z0
ond step, a specific correction model to account for the effect of K s ¼ 9:793 ð1Þ
CS
the skewed flow was developed and applied to a numerical code
for the evaluation of the performance of H-Darrieus turbines. Within this modelling approach, the near-wall cell size is deter-
Finally, the results were combined in an energy-oriented analy- mined by the condition that its first nodal point must have a distance
sis to evaluate the effective energy production over a yearly time yP from the wall which satisfies Eq. (2):
horizon of a purposefully designed H-Darrieus turbine as a func-
tion of the installation site (in terms of building dimensions and yP > K s ð2Þ
proportions with respect to the surrounding buildings). In order to assess the best numerical configuration, the perfor-
mance of several turbulence models was investigated and vali-
2. Numerical analysis dated with experimental wind tunnel measurements of the
CEDVAL laboratory [13] by means of a test case based on a single
The CFD simulations were carried out with the OpenFOAM soft- building block (1:200 scale). In particular the standard k–e model,
ware package [9]. The modelling approach was a 2-D steady-state the RNG k–e model and the realizable k–e model were compared:
F. Balduzzi et al. / Applied Energy 97 (2012) 921–929 923

Fig. 4. Computational domain.


Fig. 2. Vertical velocity profiles along the first test model (125  125  125 mm)
[13]: comparison between experimental data and simulations with the Standard k–
e turbulence model. elements are chosen in order to preserve the wind profile which
is imposed at the inlet. In this study, two urban schemes were ana-
lysed, having different proportions between the investigated build-
the standard k–e model showed a better fit with the experimental ings. The case studies were selected from a previous work [5], in
data and was therefore selected as the base scheme for the whole which a generic urban environment was modelled (Fig. 4): by
set of simulations. hypothesizing a wind turbine installation on a specific building
In detail, the numerical approach was validated by means of a (installation building, IB), the oncoming flow on the turbine was cal-
comparison between 3D simulations with the numerical code culated as a function of the IB height, the height of its upwind build-
and the experimental data in terms of velocity profiles obtained ing (UB) and the distance between IB and UB.
by [13] on two different buildings’ shapes. The first model (1:200 As suggested by technical literature [14–17], in order to avoid
scale) had a cubic form (H = L = 125 mm and a frontal width of any influence on the final solution, the dimensions of the compu-
125 mm), whereas the second model (1:200 scale) was character- tational domain must be referred to the tallest building height
ized by a height (H) of 125 mm, a length in the wind direction (L) of (H). In this work, however, since two different values of the instal-
100 mm and a frontal width of 150 mm). The comparisons (see lation building height were investigated, all the simulations were
Figs. 2 and 3), carried out in terms of vertical velocity profiles, carried out with fixed dimensions of the boundaries; in further de-
showed constant agreement with the experiments. tail, with reference to H = Hmax, the computational domain was ob-
The main issue in applying the scale factor to the final model tained extending its boundaries from the IB centre by 15Hmax
was, however, to satisfy the yP constraint both at the ground and vertically, 5Hmax upstream and 15Hmax downstream.
on the building walls (e.g. [5,12]); this constraint would lead in fact The mesh discretization was obtained by a hexahedral struc-
for z0 values typical of the urban context (z0  1.0 m) to the crea- tured grid with high resolution: the expansion ratio of adjacent
tion of excessively extended cells at the ground (yP  30 m) with cells was kept below 1.15 (a value lower than 1.3 is suggested in
respect to the near-building size (z0 = 3.0  104 m, yP  0.01 m). [14–16]). In order to ensure the grid-independency of the results,
To overcome this criticality, an explicit modelling of the roughness a sensitivity analysis on the cells number was undertaken: five
elements was applied: these elements were considered as squared refinement levels of the mesh were tested and compared (see
blocks having the function to virtually reproduce the real rough- Fig. 5, which reports the calculated mean velocity in a 10 m height
ness effects on the flow. By doing so, the ground roughness of an zone in the corner of IB).
open-landscape environment can be modelled using a smaller va- As a result, a number of cells of about 8  104 was selected as
lue of z0 (3.0  102 m). The roughness elements height ( H) b must
the best compromise (the velocity error between the chosen
be calculated from the average building height of the investigated refinement level and the maximum one is lower than 0.1%).
urban area, whereas the number, width and spacing between the As prescribed in [18], the inlet boundary conditions were im-
posed by the assumption of a constant shear stress with the height

Fig. 3. Vertical velocity profiles along the second model (125  100  150 mm)
[13]: comparison between experimental data and simulations with the Standard k–
e turbulence model. Fig. 5. Sensitivity analysis on the mesh discretization.
924 F. Balduzzi et al. / Applied Energy 97 (2012) 921–929

Table 1
Numerical model settings.

Algorithm Simple
Interpolation schemes Always second order
Turbulence model Standard k–e
Boundary conditions Inlet uðzÞ ¼ uj ln ðzdÞ

u 3 2
u ffiffiffiffiffi
z0 ; eðzÞ ¼ jz ; jðzÞ ¼
p
Cl
Top Symmetry plane
Outlet Zero gradient
Roughness length Ground 3.0  102 m
Wall 3.0  104 m

(the hypothesis has been also validated by [4, 8–10]). By the assign-
ment of the friction velocity (u), the resulting inlet velocity profile
was therefore a logarithmic boundary layer; k and e were imposed
by a constant and a hyperbolic function, respectively. The final set-
tings of the numerical model are summarized in (Table 1).

3. Case study

The roughness parameters and the wind characteristics


exploited in the simulations were extrapolated from a real city data b h = 1.5 H;
Fig. 6. Velocity field and streamlines for: (a) H = 2 H; b D = 0.5 H;
b (b) H = 4 H;
b
b D = 0.5 H.
h = 2.75 H; b
set; in particular, some literature data for London city [19,20] were
chosen (Table 2). The displacement value was calculated, following
the indications of the ESDU [21], by Eq. (3):
Table 3
b  4:3z0 ð1  Ah Þ ¼ 13 m
d¼H ð3Þ Case studies.

b were purpose- IB height UB height Distance between Inclination


Three roughness elements of constant height H (H) (h) IB and UB (D) angle (v)
fully placed before UB and IB, in order to reproduce over UB the
Case 1a b
2H b
1:50 H b
0:5 H 0°
correct urban inlet profile in terms of velocity distribution and dis-
Case 1b 8°
placement. Two geometric configurations (Fig. 6) were conse- Case 1c 18°
quently selected from a previous work [5], in order to achieve at Case 2a b
4H b
2:75 H b
0:5 H 0°
the rooftop of the investigated buildings a lower and a higher avail- Case 2b 8°
Case 2c 18°
able wind distribution with respect to the undisturbed flow,
respectively. In detail, the two case studies have a double and qua-
druple IB height than the city average height Hb (following the basic
requirements proposed by [1]) and the same distance between IB
b
and UB (D = 0.5 H).
In addition, each geometric configuration was simulated with
three roof typologies (Table 3): in particular, a flat roof (e.g. tall
modern buildings, skyscrapers, etc.) was compared with a sloping
roof, with two inclination angles of 8° (roof slope in central Italy)
and 18° (average slope of the Italian roofs) respectively, in order
to evaluate the flow variations in term of velocity magnitude and
skew angle approaching the turbine (e.g. Fig. 7).

4. Performance correction model for skewed flow

Recent wind tunnel tests ([6,8]) carried out on a low-solidity


H-Darrieus wind turbine, have shown that the rotor performance
is strongly affected by incidental misalignments of the flow with
respect to its main flowpath, which is generally perpendicular to
the rotor axis. Surprisingly, VAWTs are characterized by an opposite
behaviour in skewed flow than that of the conventional horizontal-
axis wind turbines (HAWTs), which are negatively affected by
skewed flow conditions [22]: an increase of the power output of

Table 2
Fig. 7. Examples of the velocity field and streamlines for the same geometric
London data.
configuration with a flat (a) or a sloping roof (b), respectively.
b
Mean building height ð HÞ 13.6 m
Percentage of the total area occupied by buildings 55% the rotor was in fact noticed for relatively small skew angles. Mov-
Roughness length (z0) 0.29 m ing from this evidence, two simplified models have been developed
Friction velocity (u) 0.55 m/s
by Mertens et al. [6] and Simão Ferreira et al. [8]. The first model [6]
F. Balduzzi et al. / Applied Energy 97 (2012) 921–929 925

consists in a modification of the classical BEM approach, by assum-


ing a reduction of the effective flow velocity as a function of the co-
sine of the skew angle and a modification of the swept area of the
turbine by projecting it perpendicularly to the oncoming flown
direction. On the other hand, the second approach [8] is limited
to an a posteriori correction of the attended power coefficients on
the basis of the flow momentum variation due to the oncoming
skewed flow.
On these bases, a new aerodynamic model to account for the
effects of the skewed flow in an H-Darrieus turbine was specifically
developed and embedded in the VARDAR code of the ‘‘Sergio Ste-
cco’’ Department of Energy of the University of Florence (see
[23,24]). The VARDAR code makes use of the Double Multiple
Streamtubes with Variable interference factor (DMSV, [25]) for the
performance estimation of H-Darrieus rotors but the possibility
to use a Multiple Streamtubes approach [26] is also provided and
Fig. 8. Schematic view of the turbine model divided in 5 sectors.
adopted whenever very low solidity two-bladed rotors are ana-
lysed [6]. Several sub-models to account for the main secondary
(e.g. streamtubes expansion, dynamic stall) or parasitic effects inder changes with #; as a result, the torque output of a blade at a
(e.g. parasitic torque of the struts and shadowing effect of the cen- certain azimuthal angle is determined by a different combination
tral tower) are also embedded [24]. of blade portions working in accordance with a DMSV or a MS
In detail, the theoretical approach that was followed for the model.
proposed skewed-flow model was to improve the original scheme The calculation scheme of the proposed model can be summa-
proposed in [6] by providing a more refined description of the rized as follows:
flow-blade interaction in the rotor as a function of both the azi-
muthal position and the span position of the blade itself.  The aerodynamic performance of sectors II, III and IV are calcu-
With regard to the contribution of the modified velocity, mov- lated with a Double Multiple Streamtubes approach with the
ing from the results of Hoerner [27,28] and Jones and Cohen [29], VARDAR code;
the lift and drag forces produced by an airfoil in skewed flow con-  The performance of sectors I and V are defined by a modified
ditions are only due to the wind component perpendicular to the Multiple Streamtubes approach which considers only one inter-
blade, whereas no effect is connected to the parallel component. action between the flow and the blades;
The same assumption was included in the code, where a reduced  The torque value of the machine at each azimuthal angle (#) is
velocity, connected to the cosine of the skew angle, was considered calculated with an averaged mean on the basis of the height
(see [6]). ratio between the sectors and the final performance of the tur-
Focusing on the contribution of the modified area, one can read- bine is finally computed.
ily understand that the turbine swept area is no longer constant,
but experiences a variation as a function of the skew angle [7,8]: The validity of the new model has been verified by means of the
in particular, the most correct cross-section which has to be taken only experimental data available in the technical literature [8], re-
into account to model the turbine functioning is the projection of ferred to a low-solidity two-bladed H-Darrieus turbine. The com-
the frontal area (i.e. Aj = Ht  Dt) on a plane normal to the oncoming parison between simulations and wind tunnel measurements has
wind direction. As a consequence of this choice, the projected area shown a notable agreement (Fig. 9), with a remarkable improve-
increases for relatively small skew angles, due to the contribution ment in the trend description if compared to the previous theoret-
of portions of the upper and lower planes of the virtual cylinder ical model [8]. On this basis, although a new experimental
swept by the rotor. A limit value to the increment of the projected campaign has been already planned by the authors to achieve a
area, mainly dependent on the turbine geometry, can however be more extensive validation of the proposed approach, the model
found: beyond this value, part of the flow does not find any blade
element on its path (the turbine is seen like a hollow cylinder),
thus reducing the effective frontal area.
Moving from these assumptions, in order to evaluate the aero-
dynamic performance of the rotor, the swept volume of the blades
was divided into five sectors with slicing planes parallel to the
oncoming flow direction, as shown in Fig. 8.
Focusing on the aerodynamic implications of this model, the
oncoming flow which passes through sectors I and V experiences
a single interaction with the blades, whereas the flow passing
through sectors II, III and IV, is subjected to a double interaction
with the blades (i.e. in the upwind and in the downwind halves
of the rotor). As a result of this schematization, the proposed model
studies the aerodynamic interactions in sectors II, III and IV with
the DMSV approach; sectors I and V are conversely analysed with
a modified Multiple Streamtubes (MS) approach, in which the
power extraction deriving from only one blade-flow interaction
is considered in the calculation of the thrust force of the rotor.
In addition, from a perusal of Fig. 8, one can notice that the
height of each sector (with the only exception of sector III, which
is characterized by a constant height) in the external surface of cyl- Fig. 9. Comparison between simulated and experimental data.
926 F. Balduzzi et al. / Applied Energy 97 (2012) 921–929

has been here applied to the calculation of the performance of an


H-Darrieus turbine in the built environment.

5. Feasibility analysis

The CFD results and the developed model to account for the per-
formance variation due to the skewed flow were combined in order
to perform a feasibility analysis of an H-Darrieus installation in the
rooftop of a building in the built environment.
As a first step, a study turbine was designed, whose main fea-
tures are reported in Table 4: the H-Darrieus scheme was adopted
(Fig. 10) and structural and dimensional constraints, which could
be compatible with a rooftop installation, were taken into account
in designing the machine (see [23]); in particular, due to the high
rotational speeds of these rotors, thick supporting struts and aux-
iliary tie-rods have been included in the model in order to resist
the centrifugal loads acting on the blades.
The CFD simulations for the selected buildings geometries (see
Table 3) were hence analysed. A 2 m height zone over the corner of
IB was taken into account as the reference section for the turbine
functioning [5]. Within this zone, the mean velocity variation DU
Fig. 10. Scheme of the designed H-Darrieus turbine.
(%), with respect to the undisturbed wind profile at the same
height, and the skew angle of the wind flow were calculated on
the hypothesis that the turbine slightly affects the oncoming flow
field. In the averaging process, a maximum mean square error of 3%
on the velocity modulus and of 7% on the skew angle were noticed.
In the present analysis, turbulence effects were neglected and they
may require further evaluations in the next future.
The main results of the analysis are summarized in Fig. 11.
Upon examination of Fig. 11, it is worth noticing that the geometric
proportions between the UB and the IB have a remarkable influ-
ence on the velocity variation at the rooftop of IB; Case 1 configu-
rations (with or without the sloped roof) show indeed a constant
decrease of the flow velocity with respect to the undisturbed wind
profile, whereas Case 2 configurations constantly offer a positive
velocity variation, which is however maximized by the application

Table 4
Main features of the study turbine.

Parameter Symbol Unit Value


Turbine main features
Turbine diameter Dt (m) 2.0
Turbine height Ht (m) 2.0
Swept area A (m2) 4.0
Blades’ number N 3
Blades’ type Straight
Blades’ chord c (m) 0.3
Aspect ratio of the blades AR 6.7
Solidity r 0.45
Aerodynamic design of the blades
Airfoil NACA0018
Thickness t (% of c) 18%
Maximum thickness location (% of c) 25%
Camber line type Straight
Pitch angle u (°) 0
Resistant elements
Struts’ number Ns 6
Fig. 11. Mean velocity variations (a) and skew angles (b) in the analysis section at
Struts’ thickness ts (m) 0.030
the rooftop of the investigated buildings.
Tie-rods’ number Nr 3
Tie-rods’ thickness tr (m) 0.003
Diameter of the central tower Dtw (m) 0.060
of the 8° sloped roof. Moreover, in Case 1 configurations a more
Power and efficiency parameters
Electrical efficiency gel 0.8 pronounced dependence from the inclination angle of the roof
Nominal wind speed unom (m/s) 16.0 was found both in terms of velocity variation and of skew angle;
Rated power @ unom P (W) 1730 in Case 2 configurations, conversely, a lower dependence from
Cut-in speed ucut-in (m/s) 3.0 the roof slope can be appreciated and almost the same skew angle
Cut-out speed ucut-out (m/s) 20.0
was found with sloping roofs with either v = 8° or v = 18°.
F. Balduzzi et al. / Applied Energy 97 (2012) 921–929 927

Moving from the results of Fig. 11, the attended performance of


the Darrieus turbine was calculated using the developed model
embedded in the VARDAR code.
The turbine characteristic curves were calculated for all the
wind velocities in the range between 0 and 20 m/s (selected cut-
out velocity) and for the skew angles highlighted by the analysis
of Fig. 11b. Then, the new curves were matched with the inverter
load curve, which was determined as the envelop of the best oper-
ating points of the turbine in aligned flow [30]: as a result, the ex-
pected power that is extracted by the rotor in each flow condition
was calculated. It is worth pointing out, however, that, the original
load curve is not optimized anymore in case of an oncoming
skewed flow, due to the shape modification of the characteristic
curve at each wind velocity. In order to overcome the potential
power losses connected to this problem, a real-time adaptation
of the load request could be provided as a function of the skew an-
gle: this solution seems, however, to be very complex and was not
taken into account here. As an example, the power curve for the
aligned-flow condition is compared to that at c = 33° (Case 2a) in
Fig. 12. Upon examination of Fig. 12, a power increase in skewed
conditions up to 6% can be appreciated; in addition, one can also
notice that the performance improvement due to the effects of
the skewed flow can also lead to a reduction of the minimum
cut-in speed, thus extending the operating range of the rotor and
increasing the energy harvesting for the low-wind conditions.
Finally, in order to perform a yearly calculation of the available
power production, some assumptions were introduced:

 A prevailing wind direction was considered, in which all the


oncoming wind during the year was supposed to blow. This
assumption is often verified in urban applications, in which
the installation building is often chosen to ensure a façade per-
pendicular to the prevailing wind.
 The wind in that direction was assumed to have a Rayleigh
velocity distribution [5], with the scale factor to match the
attended mean velocity of 6 m/s at Href.

These assumptions are in fact needed to get some general ten-


dencies, but it is readily noticeable that they impose a heavy sim-
plification of the real urban environment, whose precise
description can be ensured only by 3D simulations of the peculiar
analysed case.
On these hypotheses, the net available wind distributions in the
investigated zone in the rooftop of IB were finally calculated for the
considered buildings geometries (Fig. 13).
Upon examination of Fig. 13, one may notice that different
installation choices can lead to a remarkable modification of the

Fig. 13. Available wind distributions for plan roof cases (a) and sloped roof cases
[v = 8° – (b) and v = 18° – (c)].

wind statistical distributions, which are shifted in terms of both


the average velocity and the characteristic frequencies; conversely,
the distributions are slightly affected by the roof slope when a
small inclination angle is chosen, whereas an high inclination angle
can substantially modify the wind conditions in the rooftop, espe-
cially in the case of a short building. Under these assumptions, the
variation of the yearly energy production of the turbine with re-
spect to a hypothetic installation in the undisturbed wind at
Href = 36 m (Case 0 with null skew angle) are presented in
Fig. 14; in detail, the energy production of each configuration
was calculated either with or without accounting for the skew an-
gle effects (thus with the modified cut-in speed and power curve).
Fig. 12. Turbine power curve modification in skewed flow. From a perusal of Fig. 14, it is worth noticing that:
928 F. Balduzzi et al. / Applied Energy 97 (2012) 921–929

investigated installations schemes (Table 5) with respect to the


hypothetic one in Case 0.
It is readily noticeable how the relatively low velocity and skew
angle variations discussed in Fig. 11 have a remarkable impact on
the potential yearly production of the machine, mainly due to the
performance benefits induced at low wind velocities (which are,
however, the most frequent conditions in a long-time horizon).
Furthermore, on the basis of the consideration that the conve-
nience of a turbine installation is proportional to the capacity factor
[31], Table 5 clearly indicates that a microeolic installation in the
built environment could indeed represent a profitable solution,
but only on condition that a suitable installation site is chosen; in
detail, buildings reasonably higher that the surrounding construc-
tions can take advantage from the flow acceleration over their faç-
ade induced by the interaction with the upwind building, with a
notable increase of the potential energy available for the turbine.
Fig. 14. Energy yield variation with respect to Case 0 for the investigated study Moreover, the application of Darrieus VAWTs to this environment
cases. is deemed to represent a promising solution thanks to the perfor-
mance improvement connected to the effects of a skewed flow.
On the other hand, a wrong choice of the installation building, i.e.
 The right choice of an urban installation site, i.e. a geometric a condition in which a detriment of the rooftop flow conditions is
condition which enables the turbine to exploit the acceleration induced by the surrounding buildings, could dramatically compro-
induced by the upwind building (Case 2 geometries), can indeed mise the energy production and the convenience of the investment,
provide a remarkable increase of the turbine production, as well.
whereas a constant detriment of the available power can be suf- Finally, in order to get an idea of the electricity saving potentially
fered when a wrong installation site is chosen (Case 1 obtainable with the energy yield of the turbine, statistical indica-
geometries). tors for the energy consumptions of residential and office buildings
 The sloping angle of the roof has a relevant impact on the were considered; in detail, a daily amount of 25 Wh/m3 was as-
attended performance of a Darrieus turbine installed in the sumed as the reference electricity consumption of this buildings’
rooftop area. In the case of tall buildings, a sloping roof is typology [32,33]. On this basis, the energy harvesting of the single
deemed to ensure an improvement of the flow conditions in turbine investigated in this work is assumed to provide a total en-
the investigated area; in addition, small inclination angles ergy saving per year for the considered buildings in the range be-
(approximately 8–10°) seems to guarantee the best combina- tween 1.4% in Case 1a without skew effects and 2.5% in Case 2c
tion between velocity increases and skew angles. On the other with skew effects. Although the global amount appears small for
hand, in the case of an unfavourable buildings geometry, small a single rotor, one can consider that multiple turbines are generally
inclination angles are supposed to improve the flow conditions placed in the rooftop: e.g. four turbines could provide up to 10% of
in the rooftop, whereas steep sloping roofs could even worsen the global electricity demand in Case 2c, which could be sufficient,
the oncoming flow. for example, to satisfy the common energy consumptions of the
 The opportunity of exploiting a skewed flow with Darrieus wind building (e.g. lighting of the shared spaces, etc.).
turbines is deemed to represent a very important opportunity,
due to the sensible increase provided in the yearly power yield 6. Conclusions
of these machines. This effect could be further increased with a
more sophisticated logic of control, although a similar solution A wide-ranging analysis was carried out to evaluate the ener-
would increase the complexity of the electrical systems and the getic suitability of a Darrieus VAWT installation in the rooftop of
control apparatus. a building in a reference European city. With this goal in mind,
the first step of the analysis consisted on a numerical CFD analysis
Finally, although proper economic and cost analyses cannot be to characterize the flow field in the rooftop area of two buildings
carried out in this context, due to the fact that the contributions with different proportions with respect to both the average sur-
of all the others wind directions are here neglected in the energy rounding buildings height and their upwind building; in addition,
yield evaluation, some interesting remarks can be done by criti- the application of either a plan or a sloping roof was considered.
cally analysing the attended variation of the capacity factor (rela- The flow velocity modulus and direction (skew angle) were calcu-
tive to the main wind direction considered) of the turbine in the lated for different oncoming wind profiles and compared to their

Table 5
Attended variation of the capacity factors as a function of the installation site.

Test case Case 1a Case 1b Case 1c


UB = 1.50Ĥ – IB = 2.00Ĥ
Roof slope 0° 8° 18°
Theoretical approach No skew Skew No skew Skew No skew Skew
Capacity factor variation (%) 37.2% 34.2% 35.5% 32.4% 47.4% 44.5%
Case 2a Case 2b Case 2c
UB = 2.75Ĥ – IB = 4.00Ĥ
Roof slope 0° 8° 18°
Theoretical approach No skew Skew No skew Skew No skew Skew
Capacity factor variation (%) +71.3% +74.4% +84.7% +95.9% +81.3% +93.0%
F. Balduzzi et al. / Applied Energy 97 (2012) 921–929 929

level in the undisturbed wind. Under the assumption of Rayleigh [10] Launder BE, Spalding DB. The numerical computation of turbulent flows.
Computer Methods Appl Mech Eng 1974;3(2):269–89.
distribution of the blowing wind, the results were projected into
[11] Cebeci T, Bradshaw P. Momentum transfer in boundary layers. New York
net available wind distributions in the rooftop of each building. (USA): Hemisphere Publishing; 1977.
As a second step, a numerical model was developed to account [12] Blocken B, Stathopoulos T, Carmeliet J. CFD simulation of the atmospheric
for the effects of the skew angle of the flow on the power perfor- boundary layer: wall function problems. Atmos Environ 2007;41(2):238–52.
[13] Leitl B, Shatzmann M. CEDVAL, compilation of experimental data for validation
mance of the H-Darrieus turbine: the modified turbine power of microscale dispersion model. Hamburg (Germany): Meteorological
curves were hence evaluated with the developed model with the Institute, Hamburg University; 1998.
flow conditions previously calculated for the study configurations. [14] Franke J, Hirsch C, Jensen AG, Krüs HW, Schatzmann M, Westbury PS.
Recommendations on the use of CFD in wind engineering. In: Proceedings of
Finally, the results of the CFD simulations and the new turbine the international conference on Urban wind engineering and building
model were combined in a comparative feasibility analysis of a aerodynamics; 2004 May 5–7; von Karman Institute, Sint-Genesius-Rode
medium-size H-Darrieus turbine in the built environment. (Belgium); 2004.
[15] Franke J, Hellsten A, Schlünzen H, Carissimo B. Best practice guideline for the
The analysis showed that notable increments (up to 70%) of the CFD simulation of flows in the urban environment. Brussels: COST Office 2007.
attended capacity factor in the rooftop area of an installation build- [16] Tominaga T, Mochida A, Yoshie R, Kataoka H, Nozu T, Yoshikawa M, et al. AIJ
ing in the urban environment can be achieved whenever a building guidelines for practical applications of CFD to pedestrian wind environment
around buildings. J Wind Eng Ind Aerodynam 2008;96(10–11):1749–61.
reasonably higher than the average of the surrounding construc- [17] Wang X, McNamara KF. Evaluation of CFD simulation using RANS turbulence
tions is selected and suitable geometric proportions of the building models for building effects on pollutant dispersion. Environ Fluid Mech
itself with respect to its upwind building are fulfilled; otherwise, a 2006;6(2):181–202.
[18] Richards PJ, Hoxey RP. Appropriate boundary conditions for computational
constant detriment of the energy potential was noticed.
wind engineering models using the k–e turbulence model. J Wind Eng Ind
In addition, a positive influence on the velocity increment in the Aerodynam 1993;46–47:145–53.
rooftop area of the sloping angle of the roof was appreciated; this [19] Ratti C, Di Sabatino S, Caton F, Britter R, Brown M. Analysis of 3-D urban
effect was maximized by the application of a sloping roof with an databases with respect to pollution dispersion for a number of European and
American cities. Water, Air Soil Pollut: Focus 2002;2(5–6):459–69.
inclination angle of 8°, which was deemed to guarantee the more [20] Martin CL, Longley ID, Dorsey JR, Thomas JR, Gallagher MW, Nemitz E.
effective guidance to the flow which overcomes the building. Ultrafine particle fluxes above four major European cities. Atmos Environ
Focusing on the application of a Darrieus turbine, it is also 2009;43:4714–21.
[21] Engineering Science Data Unit. Strong winds in the atmospheric boundary
worth noticing that the skew angles attended in the rooftop of a layer, Part 1: mean-hourly wind speeds. London: ESDU 82026 with
building in a urban environment (15–35°), seem to ensure a further Amendment A and B; 1984.
increase (up to 12%) of the attended energy harvesting thanks to [22] Burton T, Sharpe D, Jenkins N, Bossanyi E. Wind energy handbook. Oxford
(UK): J. Wiley & Sons Ltd.; 2001.
the improved behaviour of these machines in skewed-flow condi- [23] Bianchini A, Ferrari L, Magnani S. Analysis of the influence of blade design on
tions; on the basis of a specific model that was developed to ac- the performance of an H-Darrieus wind turbine. In: Proceedings of the ASME-
count for this effect in the performance estimations of Darrieus ATI-UIT 2010 Conference on Thermal and Environmental Issues in Energy
Systems; 2010 May 16–19; Sorrento (Italy).
VAWTs, this contribution is deemed to be maximized whenever a [24] Bianchini A, Ferrari L, Magnani S. Start-up behavior of a three-bladed h-
skew angle of approximately 25° is achieved. Darrieus VAWT: experimental and numerical analysis. Paper No GT2011-
45882. In: Proceedings of the ASME Turbo Expo 2011; 2011 June 6–10;
Vancouver (Canada); 2011.
References
[25] Paraschivoiu I. Double-multiple streamtube model for darrieus wind turbines.
NASA CP-2186. In: Proceedings of the 2nd DOE/NASA wind turbines dynamics
[1] Mertens S. Wind energy in the built environment. Brentwood (UK): Multi- workshop, 1981 February 24–26, Cleveland, OH (USA); 1981. p. 19–25.
Science; 2006. [26] Paraschivoiu I. Wind turbine design with emphasis on Darrieus
[2] Dayan E. Wind energy in buildings: power generation from wind in the urban concept. Montreal (Canada): Polytechnic International Press; 2002.
environment – where it is needed most. Refocus 2006;7(2):33–8. [27] Hoerner SF, Borst HV. Fluid-dynamic lift. New York (USA): Hoerner Fluid-
[3] Banks D, Cochran B, Denoon R, Wood G. Harvesting wind power from tall Dynamics; 1985.
buildings. In: Proceedings of the CTBUH 8th world congress; 2008 March 3–5; [28] Hoerner SF. Hoerner fluid-dynamics. New York (USA): Hoerner Fluid-
Dubai (UAE), Chicago: Council on Tall Buildings and Urban Habitat; 2008. p. Dynamics; 1965.
320–7. [29] Jones RT, Cohen D. Aerodynamics of wings at high speeds, aerodynamic
[4] Beller C. Urban Wind Energy – State of the Art 2009. Roskilde (Denmark): Risø components of aircraft at high speeds, high speed aerodynamics and jet
Laboratory - DTU; 2009 Report No.: Risø-R-1668(EN). propulsion. In: Donovan AF, Lawrence HR, editors. vol. VII. Princeton, NJ (USA):
[5] Balduzzi F, Bianchini A, Carnevale EA, Chesi A, Ferrari L. Influence of the Princeton University Press; 1957.
building geometry on microeolic installations in the urban context. In: [30] Ferrari L, Bianchini A. Critical aspects in the design of a small-size Darrieus
Proceedings of world renewable energy congress XI; September 25–30, wind turbine. In: Proceedings of world renewable energy congress XI; 2010
2010; Abu Dhabi (UAE); 2010. September 25–30; Abu Dhabi (UAE); 2010.
[6] Mertens S, van Kuik G, van Bussel G. Performance of an H-Darrieus in the [31] Syngellakis K. Urban wind turbines: development of the UK market. In:
skewed flow on a roof. J Solar Energy Eng 2003;125:433–40. Proceedings of the European wind energy conference; 2006 February 27–
[7] Mertens S. The energy yield of roof mounted wind turbines. Wind Eng March 2; Athens (Greece); 2006.
2003;27(6):507–17. [32] US Energy Information Administration (EIA). 2005 Residential energy
[8] Simão Ferreira CJ, van Bussel G, van Kuik G, An analytical method to predict the consumption survey: energy consumption and expenditures tables; 2006.
variation in performance of an H-Darrieus in skewed flow and its experimental [33] Macchi E, Campanari S, Silva P. La microcogenerazione a gas naturale. Milano
validation. In: Proceedings of the european wind energy conference 2006; (Italy): Polipress; 2006.
2006 February 27–March 2, Athens (Greece); 2006.
[9] OpenCFD Ltd. website: <http://www.openfoam.com>.

You might also like