Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

Huge 3-D lattice structure with small wind turbine for amenity space

Yoshitami Nonomura*, Hiroyuki Hirahara**, Md. Zair Hossain***,

Masaaki Kawahashi**, and Nobuyuki Kobayashi****

* Graduate School of Engineering, Tokyo Polytechnic University, 1583, Iiyama, Atsugi, Kanagawa, 243-0297
** Faculty of Engineering, Saitama University, 255, Shimo-Okubo, Sakura, Saitama, 338-8570, Japan
TEL +81-480-433245, hhira@mech.saitama-u.ac.jp
*** Graduate School of Science and Engineering, Saitama University
**** Faculty of Engineering, Tokyo Polytechnic University

Abstract
In order to develop a novel concept for the application of huge 3-D lattice structure with a small, friendly and
multi-purpose wind turbine, a prototype integration system has been designed and fabricated. The integrated system will
be used for not only power generation but also wind breaking or flow control. The integrated system has sited in several
location as a practical test. Successively, the preliminary examination was carried out experimentally and numerically for
the optimal design. The prediction of the wind flow is important to evaluate the windbreak deficit effect on the fluid (air).
Although the influence of a traditional windbreak on the reduction of wind speed is fairly well known, it is still yet been
done how these influence will be behave with arrays of wind turbines. In the present article, three-dimensional numerical
simulations of flow over two types of measurement features i.e. standalone and 5x5 array basic unit, are carried out. The
Navier-Stokes equations, which were modified to include drag terms as a sink term, have been solved using SIMPLE
method and standard two-equation model. The accuracy of these simulations is examined by comparing with experimental
data and explains the relationships between deficit effects and the structure of windbreak. Results indicate that the
horizontal extent of windbreak effects downwind found 3h (h=windbreak height) where velocity reduction coefficient was
found approximately 0.6. Reductions of 20% or more may extend to about 7h from the windbreak. The results are
stimulating and useful for design of wind control.

1.Introduction
The present work aims the extend of application of small wind turbines for amenity space or urban area. As
well known the development of large wind turbines, the application is extended to the wide area all over the world. The
wind profile and budget of wind power was inspected by the specialist for the source of renewable energy. On the other
hand, the small wind turbine is not prevailed enough due to
low power coeeficient and low capacity as energy source.
However, its application still involves a lot of feasibility for
multipurpose in buit environment engineering. In order to
develop a new concept for the application of huge 3-D lattice
structure with a small, friendly and multi-purpose wind
turbine, a prototype integration system has been designed and
fabricated. The development, especially analysis and design
aspects of an integrated system is a complex task involving a
wide range of scientific and engineering skills. These systems
will be used not only exploiting the energy from the wind and
enhance the farm energy production but also aspect of wind
breaking or flow control into different application packages.
Fig. 1 Integrated micro wind turbine system at Tokyo
Here, the integration concept means arrays of wind turbine
Polytechnic University 12x9 array.
and will be composed from hundreds to several thousands
windmill. Therefore, the challenge for the integrated system
researcher is to optimize the energy production and wind breaking effect.
In this context, a comprehensive methodology has been developed over the few years by the joint project with
Saitama University, Tokyo Polytechnic University, and Fujita Corporation [1-3]. In the experimental investigation, the
results show that a huge integrated system with high effcicency could be developed that it will provide a new application
concept and contrary to conventional belief. Figure 1 shows the example of the integrated system as 12x9 array at Tokyo
Polytechnic University. The system installed on the ground near a bank of the campus.
The flow behind an array of wind turbines is quite complicated. The interaction between the integrated system
and the airflow is complicated due to the turbulent characteristics of the wind and the complex behavior of natural
obstacles. The principal effects of the integrated system are the reductions of wind speed and turbulence intensity in the
downstream region due to the adsorption of momentum from the wind flow. To evaluate this effect an experimental study
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

with a typical array was carried out in the Table 1 Practical Application
previous stage. Therefore, in order to get Installation Site Number of Unit Features
a clear overview of the wake flow behind Hokkaido Tomari Village 20 Coastal area
Nigata Prefecture 2 Inside of an open space, Snowstorm endurance test
the huge 3-d integrated system it is Fukushima Prefecture 4 Mountain area, Environmental application
necessary to study numerically due to the Tokyo 5 Educational application in Junior high school
Aomori Prefecture 100 Agricultural application, Windbreak test
limitation of wind tunnel capacity. So far Saitama University 32 Building roof, Urban area application
we know that, there is no such a reference Fujita Corporation Research Centre 48 Urban area application
Fujita Corporation Head Office 10 Urban area application
in the filed of wind turbines regarding this Tokyo Polytechnic University 108 Urban area application
aspect. However, Large-scale vegetative Musashi University 32 Building roof, Urban area application
Chiba Prefecture 5 Installation in a monument
windbreaks have been established Sanden Corporation 32 Mountain area application
regarding global environmental
sustainability [4].
To simulate the spatial pattern of wind speed reductions by integrated system, the model wind turbine needs to
be developed. The research attention is therefore given to development a model wind turbine together with modification
of Navier-Stokes equations and model for wind turbine. The first challenge is to create a model wind turbine, which gives
similar flow behavior to full-scale wind turbine. The second challenge, then, modification of Navier-Stokes equation for
wind turbine. Theoretically, the windbreak exerts a drag force on the flow field, causing a net loss of momentum in the
incompressible airflow and consequently wind breaking effect. Therefore, additional flow resistance should be added to
the Navier-Stokes equations as a sink term representing total drag of wind turbine. The third challenge, then, modification
of and transport equations. There is no reference regarding wind turbine in this area. The proposed mathematical
formulations are the first approach in wind turbine to consider the drag force term in the N-S equations. However, many
attempts have been carried out in the vegetation filed regarding above aspect. For example, modeling of atmospheric
flows over plant canopies with introducing drag related sink terms into the momentum as well as into the turbulent
transport equations was done by Wilson and Shaw [5]. Shimizu and Tsujimoto [6] and Lopez and Garcia [7] introduced a
multi-dimensional two-equation turbulence closure approach to simulate the steady uniform flows through vegetation of
uniform density. Neary carried out one-dimensional numerical simulation with introducing vegetative drag as a sink term
in the Navier-Stokes equations and source terms in the and to simulate fully developed open-channel flows [8].
Fischer-Antez et al. studied three-dimensional numerical simulation with considered only vegetative drag as a sink term in
the Navier-Stokes equations [5]. Cui and Neary reported large eddy simulation with introducing drag force as a sink term
to simulate the fully developed flow through vegetation [9]. Willson and Yee investigated flow behavior in an array of
porous fences with including the drag related term [10]. The inclusion of additional source/sink terms in the mean
momentum equations and the supporting and transport equations to represent the aggregate effect of the wind turbine on
the flow can be obtained if we consider first the addition of an instantaneous drag force term. The philosophical basis for
this idea resides in the expectation that it might lead to a better description of the turbulence effects of the drag force term
as well as to more accurate solution of the mean momentum equation.
To evaluate wind breaking effects of an integrated system CFD investigation was carried out in this study and
a relevant wind tunnel experiments as well. The present paper addressed the numerical strategy on the simulation of
integrated system wind breaking effect and some unique phenomenon associated with integrated system, in order to
develop and validate numerical prediction with the aim of create a huge integrated system with small wind turbines. Then,
the outline of the methodology was presented and the experimental techniques and a modeling of turbines for numerical
simulation was also presented. Finally the flow characteristics arround the integrated wind turbines were discussed in
detail.

2.Experimental apparatus
2.1 Test facility and measurement
technic Tw in Y AG laser
In the present study, a Honey com b Traverse unit
hot wire, ultrasonic anemometer Pitot tube
and PIV were used in the velocity Test Section
measurements. All of the Flow

measurement was carried out with a Smoke

Goettingen type wind tunnel of the generator

Fujita Corp. The working cross CCD

camera
section is 3x1.8 m2. The test facility
is shown in Fig.2 schematically.
The free stream velocity was
measured at the center of the tunnel PC(data acquisition) PC(tim ing controller)
and its turbulence intensity was less 3000 12000
than 1% during the experiments.
Velocity profiles of the stand-alone Figure 2 Wind tunnel(PIV measurement)
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

original and the model wind turbine were measured using ultrasonic (KAIJO MODEL DA-600) and hot-wire anemometer
(KANOMAX, model 0251R-T5) respectively. In addition, in order to investigate the wake flow around a two-dimensional
typical array of model turbine, velocity measurement with hot-wire and particle image velocimetry (PIV) were carried out.
The blockage ratio should be considered due to its influence on the experimental results. The blockage ratio should be
small, preferably, less than 1/10.
Preparatory to studying the effect of the integrated system on the flow control, the flow field around a
standalone full-scale and a model windmill were investigated, respectively. The flow velocity for full scale was measured
with 3-dimensional ultrasonic anemometer.The sampling rate was 20 Hz and the velocity was measured for 30 s at a
measurement point. The ultrasonic anemometer was traversed around the turbine to detect the recirculation flows near the
vicinity of windmills, which is used in the full-scale turbine measurement only because 3-D measurement is possible with
it, although the inspection volume is larger than that of hot wire. Ultrasonic measurements were performed firstly in
streamwise and then the horizontal (x-y) plane. The wind turbine was mounted on a support. Alternatively, I-type hot wire
was used in scale model measurement.
The model rotor was mounted on a support. The diameter is 6mm.. The measurements were carried out with
I-type hot-wire probe. Sampling frequency was 500Hz. The velocity profile in approach flow was obtained at first, then
the wake was inspected in x-direction, vertical (x-z) plane, x-y plane respectively. For all of the measurements with
ultrasonic and hot-wire anemometer, the probe was mounted on a computer controlled three-dimensional traverse system.
2.2 Velocity measurement of 55 array model with PIV
In order to study the effect of integrated system on the flow control, a typical array of 55 turbines was
submerged in the boundary layer. The turbines were mounted on aluminium beams of 6 mm in diameter. Figure 6 shows a
bird’s eye view of integrated system installed in the wind tunnel. The dimensions of the rack were 500mm by 500mm.
The turbines were arranged in a square grid pattern, spaced 0.2d apart each other. The velocity measurement was carried
out with hot-wire and PIV. The measurement section in the hot-wire measurement is shown schematically in Fig. 7, 8 and
table 5 respectively. PIV measurement is also carried out to study the two-dimensional wake structure. The technique of
PIV is applied in the studies of wind turbine recently. Infield et al. [11] introduced the PIV technique to study the
aerodynamics in the wind tunnel test on a 0.9 m diameter wind turbine. Their investigation mainly was concerned in the
performance of the turbine and visualization of the flow filed immediate vicinity of the blade. Visualization of the wake
structure was also carried out by Whale et al. [12] using a 1/100th scale-model with two-dimensional water tunnel. In the
present study, the flow images were captured from the upstream region to 20d downstream in order to visualize the
detailed structure of the wake flow. The experimental arrangement is shown schematically in Fig. 9 and the measurement
sections are shown in Fig. 10. The integrated system was placed at the centre of wind-tunnel and aligned normal to the
upstream flow. The models used for the flow visualization were painted with an acrylic polymer black paint to minimize
laser reflection and increase the visibility of flow image. A twin Nd-YAG laser was used as an illumination light source.
The beam was introduced into the test section with a beam guide from the top wall of the wind tunnel. The wavelength of
the laser light was 532 nm and output power 30 mJ/pulse. The laser head and its related optics were mounted on a
computer controlled traverse unit for the scanning of the test section. The laser sheet was focused into an interrogation
section by adjusting the optical equipment. The thickness of the laser sheet at the measuring section was about 0.5mm.
The typical time delay between two illuminations was 200s and the frequency of image acquisition was at 15 Hz. An oil
smoke generator (Porta smoke generator) was used for seeding tracer. The nominal diameter of the smoke particle is less
than 1m. Uniform smoke streak was supplied in the main flow through a smoke regulator that consisted of honeycomb
and stainless fine mesh screen, which is located 5 m upstream of the test section. Preliminary, it was confirmed that the
smoke streak was almost laminar without a significant fluctuation. In order to ensure quality of the images, the smoke
concentration was maintained enough for the visualization by using this regulator. The flow image was recorded with a
CCD camera (Mega plus ES 1.0) from the direction perpendicular to the light sheet which resolution is 10181008 in space.
The CCD camera was placed 850mm apart from the illuminating sheet. A Nikon 38 mm lens with F number 2.1 was used.
128 image pairs were acquired for each experimental case. The dimension of the inspection area was 200200 mm for
every image acquisition. The calibration and flow images were analysed with a PIV software (IDT Provision). The size of
the correlation window was 20x20 pixels. 42x42 vectors were reconstructed from each pair of the images.

3. Velocity distribution
An example of wake profile around a
4
full-scale turbine is shown in Fig.3. The data were
acquired by the ultra sonic anemometer. This figure
Y/D [ -]

shows the mean velocity profile in the wake. x/d is 0

the distance normalized by the rotor diameter. The -2

result shows the velocity is deficit in the region that -4

0<x/d<5. 0 5 10 15 20 25
Figure 4(a) and (b) show the X/D [ -]

dependency of velocity deficit and turbulent Fig.3 Velocity distribution around a full-scale turbine (contour
intensity on rotor torque when free stream velocity map)
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

was 5m/s. Theses curves are obtained in the

different electric load of the generator. When 0.9
we changed the electric load, the torque exerted
on the turbine shaft is changed. The figure 0.8
shows that the velocity deficit was largest when 0.7
the electric load was 30 . Under this condition,
0.6
the maximum torque was generated between
20 and 30 . Then, the velocity was defected 0.5

u/ur
most at 30 . Following this reason, the 0.4 0ohm
turbulent intensity shows the largest value 5ohm
0.3
when R=30 30ohm
0.2 infinity
4.Numerical model 0.1

4.1 Turbine model in CFD 0

We carried out a numerical 0 1 2 3 4 5 6 7 8
simulation to study the flow around the x/d
integrated system. In order to analyze the flow (a) Velocity deficit in mean flow
field involving the near and far field around the 0.800
wind turbine, a modeling of turbine is required
to express the drag and rotor work. So, we 0.700
introduce a simplified geometry for the micro 0.600 0ohm
wind turbine. Figure 5 shows the schematic
0.500 5ohm
idea of turbine model in CFD. In the numerical 30ohm
 ur

simulation for the wide area including the 0.400 infinity

integrated turbine system, it is not adequate to
solve the flow field with direct simulation in 0.300

which the rotating blade was introduced. 0.200

Therefore, we introduce a body of generator
0.100
and a cylindrical interaction zone in CFD
instead of the rotating blade. The body was 0.000
inevitable object as a source of strong velocity 0 1 2 3 4 5 6 7 8
deficit such as circulation zone just behind the x/d
turbine. The interaction zone represents a
strongly influenced flow region by turbine (b) Turbulent intensity profile
blade. In this region, a drag force exerted on the Fig.4 Influence of turbine torque on the wake flow (full scale turbine)
fluid that passes through this region. The
formulation will be described in the next section.

4.2 Formulation
For deriving k-epsilon turbulence model for flow in a wind turbine, we start from the general flow equations
that include the time acceleration term, the convective inertia term, the pressure gradient term, viscous diffusion term and
the total drag term. To model the additional flow resistance, we introduce a source/sink term representing total drag to the
momentum equation. The governing equations for the flow of an incompressible fluid at high Reynolds number (Re)
based on the Reynolds-averaged Navier–Stokes approach are:
Ui
0 (1)
xi
2 1
Ui 1 P Ui 1 2
U iU j ui u j Uj uj U j uj Ui ui (2)
t xj xi xj xj xj 2
Where the Reynolds averaging of a quantity is denoted by drawing a bar over the quantity. Here, ( ui ) and ( U i ) represent
the time averaged and fluctuating quantities respectively. In addition and are the coefficients. The fourth term in the
right hand side of Eq. (2), the instantaneous drag force term, can be reduced as follows. By assuming that U jU j K u j u j ,
and neglecting the higher order terms , the following relation is obtained.
1 U ju j
U jU j 2U ju j U jU j 2 1
(3)
U jU j 2
Substituting of Eq. (3), along with the time averaging rules, into the fourth term, we have the following form,
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

Blade modeling cylinder 1

2
1
2
U ju j
Uj uj U j uj Ui ui U jU j 1
Ui ui
2
U jU j
1
2
Uj
U jU j Ui 1 u j ui
2
U jU j
Flow
(4)
Here, we only use the first order approximation of the
instantaneous drag force term. Therefore, only linear
terms are kept in equation (4). The higher order terms
are dropped. Finally we can get the following
M icro w ind turbine
equations.
Generator Unit 2
Ui 1 P Ui 1 1
2
Fig.5 Modeling of micro wind turbine in CFD U jU i u j ui U jU j Ui
t xj xi xj x j xj 2
(5)
4.3 Deriving the k and epsilon equations
Basic concept for modeling We have considered only the addition of a
a huge integrated system
drag force term to the mean momentum equation.
However, the flow filed around wind turbine cannot be
properly simulated without the inclusion of an
appropriate source/sink term on the k equations.
In view of this, the effect of form and viscous drag in
the wind turbine will also require the inclusion of
additional sink term on the k and equations.
An example of integrated unit The extra sink term in the transport equations for
Blade modeled by square slab k represents the loss from kinetic energy to
smaller scales and physically corresponds to the loss
of energy from the turbulent eddies in the wind turbine
as these eddies do work against the form and viscous
drag of the wind turbine.
Fig.6 Basic modeling concept of integrated turbine system The transport equation for the turbulent
kinetic energy can be obtained from the transport
equation for the reynolds stress tensor uiu j .The Reynolds stress tensor equation with drag force is as follows:

u j ui ui u j Ui ui u j uk Uj 1 ui uj 1
Uk u j uk ui uk p p ui jk uj ik
t xk xk xk xk xj xi xk

1
2
U kU i uk u j U k u k u iu j
U kU k ui u j 1 1
(6)
2 2
ui u j ui u j 1 U kU k U kU k
2
xk xk xk xk 2 1 U kU j ukui U k uk u j ui
2
U kU k u j ui 1 1
2 2
U kU k U kU k
After manipulation the above Eq. (6), the turbulent kinetic energy equation without drag force can be written as:
k k T k
Uj P (7)
t xj xj k x j

By following the derivation method of basic k equation, Let, j= i and k=j and introduce to the equation drag force
term of Eq. (6), the expression yields,

1 1
2 1 U jU i u j ui 1 U j u j ui ui
2 U jU j ui ui 2 1
2 1
(8)
2 2 U jU j 2 2 U jU j 2

In the present research, we only retained the first term of Eq. (10) and multiplying by 1/2 yield
1
2
k U jU j (9)
Finally, The k equation with drag force is as follows
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

k k k 1
2
Uj P k U jU j (10)
t xj xj k xj
The equation without drag force is as follows:
2
T
Uj C 1 Pk C 2 (11)
t xj k k xj xj
The drag term before and after time averaging is defined as

1 1
2
1
2
U jU i u j U j u j ui
U jU j Ui U jU j ui 1 1
(12)
2 U jU j 2
U jU j 2

1 1
2
Uj
U jU j Ui 1
u j ui (13)
2 U jU j 2

Subtracting equation (12) from equation (13). Then, differentiation of this equation with respect to xk and multiplication
ui
by 2 and averaging of that equation yields
xk
1 ui ui U jU i ui Uj ui Uj ui
U jU j 2
ui 1
uj 1
u j ui 1
u ju i (14)
xk xk U jU j 2 xk xk U jU j 2 xk xk U jU j 2 xk xk

In the case of a dissipation equation, a dimensionally consistent analogy to the above k equation. Therefore, we only
retained the first term of the equation (14) e.g.
1
2
U jU j (15)
Finally, the modeled equation for
2 1
T 2
Uj C 1 Pk C 2 U jU j (16)
t xj k k xj xj
4.4 Model overview
The sketch below depicts the
geometry of the model wind turbine. The
1/10-scale turbine was used for the
computation in order to compare with
experimental results. The wind turbine was
simplifies as a cylinder. The body of the
wind turbine modeled as original wind
turbine except the blade area. The length of
the cylinder, l , is the larger than the
original wind turbine and the diameter of
the cylinder, d1 , taken as the identical of
the original wind turbine, e.g. 0.5m. The
simulation was carried out into the wind
tunnel with dimensions L 5 and 7m,
w 1m , and h 1m . The arrangement of
the wind tunnel and model turbine unit is
sketched in Fig.7. Fig.7 Layout of computational model. Two types of layout is shown in this
The computation was executed figure.
with Star-CD. The computational grids is
unstructured tetrahedral. For the present simulation model, we discretize the solution domain for a total of 560000cells.
The grid extends 2.5 meters upstream of the wind turbine origin and it extends to a 17.5 meters downstream. The
momentum equations and k transport equation are modified for drag forces.
The algorithm of the Star-CD, based on the Semi-Implicit Method for pressure Linked Equations, discretizes
the computational domain by a finite volume method in a cell-centered arrangement on unstructured meshes used to
achieve the pressure-velocity coupling for stability. In the SIMPLE algorithm, the continuity equation is converted into a
discrete Poisson equation for pressure. The differential equations are linear zed and solved implicitly in sequence: starting
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

with the pressure equation (predictor stage), 1.2

followed by the momentum equations and the 1.0

RNG
pressure correction equation (corrector stage). The open Short
equations for the scalars (turbulent quantities) are 0.8

solved after the updating of both pressure and 0.6 50 Velocity deficit in wake
velocity components. Within this loop, the linear 0.4 30
zed equations for each variable, as they arise, are 5
treated using a linear system solver (i.e., multigrid, 0.2

Preconditioned Conjugate Gradient, PCG, etc). 0.0

k low Reynolds
Several turbulence models in Star-Cd were used to k high Reynolds
-0.2
study the wind turbine in the wind tunnel. In all the -5.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
x
of the simulations SIMPLE has been used with D
third-order accurate order scheme QUICK for the Fig. 8 Comparison of wake distribution between experimental
mean flow and first order upwind scheme for the result and CFD(standalone)
turbulence equations.
4.5 Inlet boundary condition
In our simulation, the wind turbine is
exposed to a uniform flow of air at U = 5m/s. The
coordinate system has its x axis in the direction
of the airflow, passing the center of the wind
turbine. The inflow boundary conditions were used.
The inlet turbulence intensity and mixing length
were specified as 10 percent of the inlet velocity
magnitude and 0.001m respectively. At the outlet,
flow without split was imposed. No-slip condition
was considered at solid boundaries of the wind
turbine body. Consequently, all three mean velocity
components were set to zero.
Fig.9 Velocity distribution measured by PIV for 5x5 array
A thick turbulent boundary layer was
model.
generated as the approaching flow. In the present
study, inner layer of the approaching flow 10.0

distribution will be approximated by Power law, 9.0

CFD
which was determined through the wind tunnel 8.0 PIV
measurement in 1/10-scale model.
z/d

n 7.0
Hit-wire
U Z (17) 6.0
U ref
5.0
In general, the boundary layer profile will be
expressed with empirical exponent for different 4.0

surface roughness. In the present condition, n = 3.0

0.15 was obtained in the estimation between the 2.0

measurements and the predictions of the power-law
1.0
model within a good accuracy.
0.0
0.0 0.2 0.4 0.6 0.8 U /U 1.0 1.2
5.Discussion x ref

5.1Wake flow behind a turbine Fig.10 Comparison of wake distribution between experimental
First of all, we show the comparison of result and CFD (5x5 array), wake profile behind 5x5 array at
the wake profile between the experimental results x/d=18, x/h=3.
and CFD. Figure 8 shows the comparison.
Numerical results shows a good agreement in the
deficit of velocity, however the recovery does not. For various turbulent models, the tendency shows the similar results.
5.2 Typical integrated system
We assumed two typical array of integrated system, upright 5x5 array and inclined 5x6 array as shown in Fig.7.
For 5x5 array model, experimental velocity measurement was carried out by PIV in the wind tunnel and its velocity
distribution was compared with CFD results. Figure 9 shows the experimental results. Velocity vectors in 6 planes were
presented. The velocity data was measured by PIV analysis. These PIV data was compared with hot-wire data. The results
shows a good agreement. In the vectors diagram, decelerating region behind the array and accelerating region over the
array was clearly measured. Furthermore, the drifting stream between the turbines are observed. These streaming forms a
complicated wake interacting each others.
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

Figure 10 shows the wake profile

behind 5x5 array at x/d=18, x/h=3. As shown in
this figure, two experimental results show a good
agreement. CFD profile deviates from the
experimental results, especially near the ground
wall. In our CFD model, drag force only
considered in the equations. The drag force is
considered as a simplified control volume such as
cylindrical zone. The results indicate a necessity of
improvement of mesh resolution and turbulent
(a) Velocity distribution in wake of 5x5 array upright model.
model. A flow analysis for different array
patter as shown in Fig. 7 is demonstrated in Fig.11.
In these figures, two type of array model is shown.
Upright 5x5 array is a standard constellation ((a)
and (b), and an alternative inclined model was
designed to consider the velocity gradient of
boundary layer. The array was immersed in the
boundary layer. The computation was carried out
1/10-scale model. The thickness of the boundary
layer is about 300 mm, so the height of the array is
comparable with the boundary layer thickness.
As shown in Fig.11(a), in the wake, a
strongly wind deficit area was formed by the array.
The length of the wake was approximately 6h and
its shape is chevron. Fig.11(b) shows the velocity (b) Variation of stream-wise wind profile of 5x5 array upright
variation in streamwise in the center cross section. model.
The boundary layer upstream was blocked by the
array, so the approach flow was decelerated ahead
of the array. The wake is very conformable, then a
reattachment of main flow was not observed in the
computational result. The experimental results in
Fig.9 also does not show the reattachment. On the
other hand, we carried out wind breaking
experiment with the normal fence. Although the
result is not shown in the present paper, the
reattachment flow was clearly observed in the
experiment. Of course the reattachment depends
on the blocking ratio such as porosity, it was found
that the integrated system with micro wind turbine
has a synergistic effect. (c) Velocity distribution in wake of 5x6 array inclined model
Figure 11(c) and (d) are the result for
inclined 5x6 model. The array was arrange as a
backswept constellation. The arrangement should
be effective for the energy budget because we can
utilize the displaced flow from the boundary layer.
These figure shows that a strong wake region does
not formed as contrasted with 5x5 upright array.
The In Fig.11 (d), the strong deceleration of
approaching flow does not observed. Therefore, it
was found that the configuration of the integrated
system is very important flow control and wind
power budget.

6.Conclusion
In order to develop a novel concept for
the application of huge 3-D lattice structure with a (d) Variation of stream-wise wind profile of 5x6 array inclined
small, friendly and multi-purpose wind turbine, a model
prototype integration system has been designed Fig.11 Flow analysis for different array pattern.
and fabricated. The experimental and numerical
analysis was carried out for full-scale turbine and
 Proceeding of European Wind Energy Conference 2006, Athens, P0392, 2006

1/10 scale model. The velocity distribution was measured with hot-wire, ultrasonic anemometer, and PIV. The similarity
was discussed and successively a numerical computation was conducted for the integrated turbine system. In CFD, a
turbine model for a wind turbine and combined unit was proposed in k-e model.
The velocity deficit and turbulent intensity in the experimental result for full scale turbine shows a strong
dependency on the turbine torque. The deficit was minimum and turbulent intensity maximum under the rated condition
(optimum driving condition).
CFD model gave a fair agreement with experimental result. The macro structure agree with experimental
result. It should be discussed that the not only drag but also the variation of torque for precise analysis.
Upright array model of integrated system shows a good wind breaking effect. The reattachment of main flow
was not observed in contrast to the experimental result with wind braking fence. On the other hand, the backswept model
shows a very narrow wake area. The arrangement has a advantage for wind power budget. Consequently, changing the
arrangement of integrated micro wind turbine, the wind control will suitably designed for the different purpose in built
environment.

Acknowledgement
The present project was supported by Grant-in Aid for the Development of Innovative Technology, No.12304
by Ministry of Education, Culture, Sports, Science and Technology (MEXT) in Japan. The experiment was carried out by
Mr. Masaru Fujinami, Mr. Tetsuya Sasazawa, and Mr. Tetsuji Matsushita. We appreciate for their great help.

References
1. Hirahara H, Hossain MZ, Kawahashi M, Nonomura Y. Testing basic performance of a very small wind turbine
designed for muilti-purpose, J. Renewable, Energy, Vol. 30, Issue 8 (2004) pp. 1279-1297.
2. Hossain MZ、Hirahara H, Nonomura Y, Kawahashi M. The wake structure in a 2D grid installation of the horizontal
axis micro wind turbines Renewable Energy (in print).
3. Hossain MZ、Hirahara H, Nonomura Y, Kawahashi M. Flow visualization around an array of small wind turbine
installed on the building. Transaction of Visualization Society of Japan (in print).
4. Willson JD, Yee E. Calculation of winds disturbed by an array of fences. Agricultural and Forest Meteorology 2003;
115: 31-50.
5. Willson NR, Shaw RH. A higher order closure model for canopy flow. Journal of Applied Metrology
1997;16:1197-1205.
6. Shimizu Y, Tsujimoto T. Numerical analysis of turbulent open channel flow over a vegetation layer using a k
turbulence model. Journal of Hydroscience and Hydraulic 1994:11(2): 57- 67.
7. Lopez F, Garcia M. Open channel flow through simulated vegetation: Turbulence modelling and sediment transport.
USGS wetlands research program technical report WRP-CP-10.
8. Neary VS. Numerical simulation of fully developed flow with vegetative resistance. Journal of Engineering
Mechanics 2003; 129(5): 558-63.
9. Fischer-Antize T, Stoesser, Bates P, Olsen NRB. 3D numerical modelling of open-channel flow with submerged
vegetation. Journal of Hydraulic Research 2001; 39(3): 303-10.
10. Cui J, Neray VS. Large eddy simulation (LES) of fully developed flow through vegetation. Proceedings of the fifth
international conference on hydro informatics, Cardiff 2002.
11. Infield DG, Grant I, Smith G, Wang X. Development of particle image velocimetry for rotor flow measurements.
Proceedings British wind energy conference, Stirling, Scotland 1987