Sie sind auf Seite 1von 13

WIND ENERGY

Wind Energ. 2010; 13:657669


Published online 20 January 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/we.386

RESEARCH ARTICLE

Low-cost mounting arrangements for


building-integrated wind turbines
David Udell1, David Infield2 and Simon Watson3
1 PDL Solutions (Europe), 1 Tanners Yard, Hexham, NE46 3NY, UK
2 University of Strathclyde, Glasgow, G1 1XW, UK
3 Loughborough University, Loughborough, Leicestershire, LE11 3TU, UK

ABSTRACT
Micro-generation is being widely promoted as a way for householders in the UK and elsewhere to take part in the Green
Revolution. Building-integrated wind turbines (BIWTs) provide a way to do this, enabling people to reduce their contribution to the problems of both climate change and decreasing fossil fuel availability. Although energy yields from
BIWTs for many householders have been shown to be low, there are still situations where such turbines can make a useful
contribution to electricity generation, e.g. in windier areas and for isolated detached buildings. The standards for the
installation of BIWTs are still being developed including those for the safe mounting of turbines on domestic buildings.
This paper investigates the current trend for mounting small wind turbines on the walls of domestic premises and compares
this with an approach which uses roof timbers. It identifies the main characteristics of building construction which affect
the integrity of such installations. European and British standards have been used to calculate wind and gravitational
loads. Finite element models are used to derive working stresses and, hence, some basic principles of good design. The
likely costs of wall and roof mounting are then compared. Installation and health and safety issues are also examined
briefly. Copyright 2010 John Wiley & Sons, Ltd.
KEYWORDS
building-integrated wind turbine; wind loads; finite element modelling
Correspondence
S. Watson, Centre for Renewable Energy Systems Technology, Electronic and Electrical Engineering Department, Loughborough
University, Ashby Road, Loughborough, LE11 3TU, UK.
E-mail: s.j.watson@lboro.ac.uk
Received 23 March 2009; Revised 26 November 2009; Accepted 29 November 2009

1. INTRODUCTION
Micro-generation is being promoted by politicians, media
and commercial organizations alike as a way for householders in the UK and elsewhere to take part in the Green
Revolution. Building-integrated wind turbines (BIWTs)
provide a way to do this, enabling people to reduce their
contribution to the problems of both climate change and
decreasing fossil fuel availability. Recent trials of BIWTs
have indicated that energy yields have been disappointing1
and simulations of expected yields in urban areas have
predicted low capacity factors.2 Nonetheless, there are
Copyright 2010 John Wiley & Sons, Ltd.

situations where such turbines have made a reasonable


contribution to on-site electricity generation, e.g. on isolated detached buildings in high wind speed areas and on
high-rise buildings. A number of commercial organizations in the UK already sell small wind turbines designed
for wall mounting. These installations are best suited to
gable end attachment, although this approach is not universally available for domestic housing. There are concerns that not enough has been done to assess the structural
integrity of such installations. This paper aims to address
some of these concerns and to examine alternative mounting arrangements. In particular, it looks at using the
657

Mounting of building-integrated wind turbines

timbers of pitched roofs. Horizontal axis machines rated


between 400 W and 1.5 kW are considered in the context
of UK wind conditions.

2. METHODOLOGY
The approach adopted here was firstly to understand the
loads which buildings and small wind turbines are normally subject to. A datum building geometry was defined
and has been used throughout the work; in addition, the
effects of changing some of the geometric parameters have
been evaluated. Then an understanding of material capabilities was developed in order to assess how close to the
limits building stresses are with and without a turbine
attached. And finally, finite element (FE) analysis has been
used to assess different mounting arrangements and to set
practical limits on wind turbine size and mounting
position.

2.1. Typical building loads


A selection of European and British standards were used
to calculate the main loads:
2.1.1. Wind loads
BS 6399-23 describes the methods for calculating wind
loads on buildings and BS EN 61400-24 defines the
wind speed that should be used for the design of small
wind turbines. BS EN 61400-2 states that an extreme wind
speed of 35 m s-1 should be used but combines this with a
gust factor of 1.4 to give a design wind speed of 49 m s-1.
By selecting particular parameters, it is also possible to
calculate a speed of 49 m s-1 using BS 6399-23 which is
desirable in that it gives consistency between the two standards. The value of 49 m s-1 will be referred to as design
speed 1. However, this wind speed, according to BS 63992,3 can only occur in exposed coastal regions of the UK
where there are no upstream obstructions and where, in
addition, the local topography causes speed-up effects.
This is not typical of an urban environment in the UK or
elsewhere in Europe, so, a second design speed has also
been considered. This has been taken as an urban location
in Preston, Lancashire, UK. The speed as calculated in BS
6399-23 is dependent on the height of the building. For the
geometry defined in Section 2.2 for the datum building,
the resulting wind speed is 26.4 m s-1. If, in addition, the
wind turbine hub is assumed to be 2 m above the ridge of
the roof, the wind speed increases to 28.6 m s-1. This is
taken to be design speed 2. Both of these wind speeds are
consistent with a 50-year return period.
Both these design wind speeds exceed those at which
most small wind turbines cut out, these being typically
around 15 m s-1. For design speed 1, this means that the
highest thrust experienced by the turbine is when it is in
its parked position. For the 1.5 kW and 400 W machines,

658

D. Udell, D. Infield and S. Watson

with estimated swept areas of 6 and 1.9 m2, respectively,


the loads were estimated to be 1600 N and 600 N, respectively. These areas were estimated from a survey of small
wind turbine swept areas versus thrust conducted by the
New and Renewable Energy Centre (NaREC) in the UK.
However, the thrust calculated for a machine operating at
a wind speed of 15 m s-1 exceeds that for a parked machine
at design speed 2. For the large and small wind turbines,
the peak loads were thus estimated to be 800 N and 300 N,
respectively. These loads were used for the Preston location. Drag loads due to the support pole and attachment
structure were also estimated using BS EN 61400-24 and
included in the FE modelling. Note that the thrust on the
rotor, Trotor, was calculated using equation (1):
Trotor = ACTU 2

(1)

where A is the area of the rotor, U is the wind speed, is


the air density and CT is the thrust coefficient. Trotor was
calculated for two cases: (i) parked, where CT was assumed
to be 0.18 for a large machine and 0.21 for a small machine;
and (ii) operating, where CT was assumed to be 1.0 for
both a large and small machine.
The thrust on the pole, Tpole, was calculated assuming
that the pole was cylindrical with a drag coefficient,
CD, of 1.3, as specified in BS EN 61400-24, and using
equation (2):
Tpole = AfrontalCDU 2

(2)

where Afrontal is the frontal area of the pole given by the


product of the pole diameter and pole height.
Wind loads perpendicular to the gable end and perpendicular to the non-gable wall were calculated for design
speeds 1 and 2 (see Figure 4 for definitions of gable end
and non-gable wall). The pressures were applied to the
whole area of the walls even though, in an urban environment, this may not be realistic due to upstream obstructions. This approach ensures the calculations remain
conservative.
For a detached house, BS 6399-23 was used to calculate
the loads on the walls parallel to the wind direction. These
side loads on the building were shown to have a minimal
effect on the stresses in the walls facing the wind and so
were not considered further. Wind loads on the lee side of
the building are less than those on the windward side so
the former were not considered further.
For masonry structures designed using BS 5628-1,5 a
partial safety factor (PSF), for loads, of 1.4 should be
applied. BS 1995-1-1-6 gives the main calculations for
designing timber structures but does not contain a PSF
for loads. However, a value of 1.4 is given in BS 19907,
matching the figure for masonry structures.
The focus thus far has been on extreme wind speeds.
However, fatigue and gyroscopic loads also need to be
considered. Average urban wind speeds in the UK, at a
height of 10 m above ground, are estimated to be less than
5.5 ms-1 8 or even less than 4.0 ms-1 9. Because the thrust
Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

D. Udell, D. Infield and S. Watson

Mounting of building-integrated wind turbines

loads are proportional to the square of the wind speed, the


turbine load at 6 m s-1 is only 5% of what it is at 28.6 m s-1.
For a more conservative approach, BS EN 16400-24 recommends that cycling loads should be estimated on the
basis of the wind speed varying between 0.5 and 1.5 times
rated speed. Since rated speeds are typically up to 12 m s-1,
this is equivalent to varying the speed between 6 and
18 m s-1. Even at 18 m s-1, the thrust is less than 40% of
what it is at 28.6 m s-1. Provided the material fatigue capabilities exceed 40% of their ultimate capabilities, fatigue
should not be a major influence on the design (see Section
2.4). The gyroscopic yaw moment was shown, using BS
EN 61400-24, to cause reactions an order of magnitude
lower than those caused by the thrust loading.
2.1.2. Dead loads and imposed loads
BS 6399-110 describes the calculation of roof, floor and
ceiling loads. The main concern in buildings is the avoidance of tensile stresses because these are not well resisted
by mortar joints (see Section 2.3). Wind loads tend to
cause tensile bending stresses at the bottom of the wall
which may not be overcome by the compressive stresses
due to gravity. Wind turbines induce localized tensile
stresses around their support attachments.
For wall-mounted turbines, in order to be conservative,
floor, roof and ceiling loadings were calculated at their
minimum and snow loads were assumed to be zero.
However, for roof-mounted turbines, larger ceiling and
roof loads were considered, as this gives the more conservative case. In designing a timber roof structure, BS5268311 recommends combining the wind load calculated in

Section 2.1.1 with a ceiling load (for a loft used for storage
only) of 0.55 kN/m2 and half of the maximum imposed
load calculated in BS 6399-3,12 which accounts for snow
loads.
2.2. Datum building geometry
A small building with internal dimensions 5 m 5 m
5 m, excluding the roof, has been selected since it is the
largest practical size where a close couple roof (i.e. one
that does not require purlins) can be made using readily
available timbers,13 and hence, the timber stresses are
relatively high.
The foundations were modelled by extending the walls
to 0.5 m below ground level. To keep the model simple,
no windows or doors were included. This is not unreasonable as apertures are generally remote from the highly
stressed regions.
Each wall used a cavity construction with two leaves
100 mm thick, separated by a 50 mm gap. These dimensions were chosen because of their common use.13 The
leaves are assumed to be joined by wall ties spaced as
recommended in BS 5628-1,5 i.e. 2.5 ties per square metre.
Using guidance from Foster,13 the ceiling joists and
rafters were spaced 0.5 m apart and measured 100 mm
deep by 50 mm thick. The ceiling joists were joined by a
binder of section 75 mm 50 mm which was joined to a
ridge board of section 175 mm 32 mm by hangers of
section 50 mm 38 mm (see Figure 1). Note the noggings
(green), which attach the end rafters to the gable ends via
steel straps12 and the wind turbine support (cyan).

Wind Turbine Pole

Ridge Board

Hanger
Binder
Ceiling Joist
Rafter
Wall Plate,
(underside fixed to the inside skin of the non-gable wall)

Figure 1. Roof construction with cladding suppressed.

Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

659

Mounting of building-integrated wind turbines

This roof model represents the style of construction


used before trussed rafters became the norm. Most of the
analysis work was carried out using this kind of roof, but
trussed rafter construction is considered in Section 3.2.3.
In common with most domestic buildings,14 roof loads
are transmitted to the inner leaf of the wall cavity only.
This makes the outer leaf more susceptible to tensile
stresses so it is tied to the roof structure to minimize this.
A roof pitch angle of 30o was chosen as this is typical of
UK construction.

2.3. FE meshing detail


The full house FE mesh used 20 node brick elements of
side length 100 mm. This was sufficient to assess the bulk
effects of wind and gravity. Loads and displacements from
this model were applied to a more finely meshed submodel which was used to assess the local stresses due to
the thrust and weight of the wind turbine around the attachment points. Bolt fixing loads were applied at the midheight of the bricks causing localized peak stresses there.
However, the region of interest is the mortar line where
the greatest material weakness exists. This location was
two elements away from the point load so that stresses
should be accurately predicted there. In assessing the stress
around a particular attachment point, an average value
along the length of the brick (215 mm) was calculated,
before comparing against the known material limits (see
Figure 2).

2.4. Material properties


Many kinds of brick, mortar and timber are used in the
construction of buildings. This is further complicated by

D. Udell, D. Infield and S. Watson

their often highly variable densities, stiffnesses and


strengths. In this work, typical densities and stiffnesses and
minimum reasonable strengths were assumed.
2.4.1. Wall materials
Wall and roof material densities were obtained from
either BS 64815 or Illston and Domone.16 The compressive
stiffness of walling materials is well documented, but it is
the flexural stiffness which is the main concern here. In
Illston and Domone,16 the modulus of elasticity for such
materials is estimated to be 1 to 2 GPa so this range of
values was investigated. The flexural strength of masonry
made with clay bricks with medium water absorption and
type (iii) mortar is given in BS 5628-15 as 0.4 MPa.
However, BS 5628-15 recommends applying a PSF for
material strength of 3, giving a design limit of 0.13 MPa.
Masonry fatigue stress limits are not given in BS 5628-1.5
However, concrete seems to perform well in fatigue, at
least up to 106 cycles. A survey in Lee and Barr17 shows a
number of data sets where the 106 strength for plain concrete is around 65% of its ultimate strength, while Singh
et al.18 records a 2 106 strength of 58% of the same.
Hence, for masonry materials, following the argument of
Section 2.1.1, fatigue loading from a wind turbine is
expected to be significantly less severe than that caused by
the extreme once in 50 years wind.
2.4.2. Roof and floor materials
The roof timbers were considered to be made from UKgrown Douglas Fir as this is the weakest of the wood types
given in BS EN 338.19 The strength grade is C14 and its
density is 350 kg m-3. The characteristic bending strength
for short-term loading of a 100 mm deep beam, parallel to
the grain, is 16.7 MPa. A PSF of 1.3 was applied to bring
the limit to 12.8 MPa. The equivalent strength across the

Figure 2. Fine mesh showing brick and mortar outline.

660

Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

D. Udell, D. Infield and S. Watson

Mounting of building-integrated wind turbines

grain is only around 0.65 MPa, so any significant loading


in this direction should be avoided.
Timber fatigue stress limits are not given in BS EN
338.19 However, timber also performs well in fatigue, with
a 107 strength of 60% of the ultimate strength recorded in
Illston and Domone.16 This is largely species independent.
Hence, for timber too, fatigue loading from a wind turbine
is a less severe load case than that due to the 50-year gust.
The ground floor is normally supported by sleeper walls
built up from the foundations and does not interface with
the walls. The first floor was modelled with 225 50 mm
timber joists at 0.5 m centres.

the support poles were made of hollow steel tubing of


diameters as follows: 100 and 62.5 mm for the non-gablemounted large and small turbines, respectively, and 80 and
50 mm for the gable-mounted large and small turbines,
respectively.
2.5.1. Wall mounting
Wall mounting is the current state of the art but is most
suited to gable end attachment as the support pole length
can be minimized. Non-gable wall attachment was also
evaluated. There are currently three main types of attachment. These are depicted in Figure 3 and described below.

2.5. Wind turbine installations


To begin with, the stresses in buildings without wind turbines were investigated. Vertical tensile stresses were
found to be the main issue because of mortars weakness
under this regime. Next, loads deriving from typical wind
turbine installations were assessed. As already mentioned,
the initial height chosen for the hub was 2 m above the
ridge of the roof. This is slightly arbitrary as there are
many opinions on the subject. Renewable Devices20 advise
a minimum roof-to-blade height of 0.5 m for their 1.5 kW
rated machine, while Martin8 recommends 10 m (see
Section 4).
For a hub height of 2 m above the ridge, the attachment
point reactions due to the support pole drag loads represented around 20% of those caused by the 1.5 kW turbine
itself. It was also estimated that the weight of the support
pole and brackets would be around 1.5 to 2 times the
weight of the turbine. Obviously, much depends on the
chosen materials, but for this research, it was assumed that

4 off C,
Brackets

2 off H,
Brackets

Poles supported using two or three tripods. The feet


of a tripod are fixed to the wall, while its head supports the pole. One tripod supports the lowest point
on the pole, while a second is located as high up the
wall as possible to minimize the length of unsupported pole. Other tripods can be added in between.
The height of the tripod (in the horizontal plane) is
set such that the support pole avoids the roof overhang. The effects of varying the distances between
the pole and the wall and the distance between the
tripods were considered. Two tripod arrangements
can be seen in Figures 3 and 4 showing a typical
installation.
Poles supported using three or four C-shaped brackets attached directly to the wall such that the pole
breaks through the roof overhang. In Figure 3, these
brackets are represented by short cross beams.
Poles supported at two positions using flat H-shaped
structures which are close to the wall. Once again,
the support pole has to break through the roof
overhang.

Dual Tripod,

Triple Tripod

0.9m
1.2m
1.3m

Figure 3. Support arrangements for wall mounting.

Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

661

Mounting of building-integrated wind turbines

D. Udell, D. Infield and S. Watson

Gable End Wall

Non-Gable Wall

Figure 4. Stress distribution due to wind and gravity.

An important aspect in the mounting of wind turbines


on buildings is the minimization of structure-borne vibration. The configurations detailed above are those most
commonly used and vibration isolation devices such as
rubber strips are often used to minimize the transmission
of vibrations to a building. The majority of domestic
installations have not experienced major problems with
structure-borne vibrations using the above types of mounting arrangements, but there was noted to be a vibration
problem for 29% of the installations which formed part of
the UK Energy Savings Trusts domestic scale wind trial.21
There has been very little work to date in this area and a
proper study of structure-borne vibration is outside the
scope of this paper. This paper does not deal with the possible transmission of vibrations to a domestic building, but
it is noted that measures must be taken to minimize this
potential problem. In this paper, the emphasis is on static
loading in order to make the analysis tractable.
2.5.2. Roof mounting
Mounting from the ridge board and an additional horizontal member supported by two rafters are discussed in
detail. Figure 1 shows the mounting close to the gable
wall, but the same mounting was also considered for a
mid-roof position because gable end access may be
restricted by a chimney or television aerial.

3. RESULTS
3.1. Wall mounting
Unless otherwise stated, stress refers to the vertical
tensile stress. The results for the wall mounting arrange-

662

ments are summarized in Table I, at the end of Section


3.1.2.
For design speed 2 with the wind load applied to the
non-gable wall, the peak stress in the lower wall was
0.165 MPa after gravity effects were accounted for. For
the wind load applied to the gable wall, the equivalent
stress was 0.235 MPa, as shown in Figure 4. These exceed
the limit given in Section 2.4.1 suggesting that the PSF
for material strength given in BS 5628-15 is rather
pessimistic and perhaps not strictly adhered to in practice.
Therefore, for the purposes of this report, non-gable
and gable limits of 0.165 and 0.235 MPa were set, respectively, for design speed 2. Provided the stresses in the
wall due to the wind turbine pole support feet do
not exceed these values, then the wall will not experience
stresses higher than it would if there were no wind
turbine.
Note that the stress due to gravity at the top of the wall
is virtually zero, while the stress in the lower wall due to
the turbine supports is virtually zero. Therefore, the
stresses induced by the turbine are effectively independent
of what is going on at the bottom of the wall. This means
that provided the local stresses due to the pole supports are
less than those in the bottom of the wall (under the same
wind conditions), the former must be acceptable. The
limits of 0.165 and 0.235 MPa can be said to be relative
because they depend on what the peak stress in the wall is
without the wind turbine present.
For design speed 1, tensile stresses due to the wind
increase in proportion to the wind speed squared. The
tensile stresses due to the wind are offset by the effect of
gravity which is clearly greater at the bottom of the wall
than it is at the top. Relative to design speed 2, the effect
of gravity is reduced which means that wall local stresses
Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

D. Udell, D. Infield and S. Watson

Mounting of building-integrated wind turbines

Table I. Wall mounting assessments for design speed 2.


Turbine power
(kW)
1.5
1.5
0.4
1.5
1.5
0.4
0.4
1.5
0.4
1.5
0.4

Height above
ridge (m)

Description

Support
spacing (m)

Wall

Stress for Perpdr


wind (MPa)

Stress for parallel


wind (MPa)

Acceptable?

2
2
2
2
2
2
5
2
2
2
2

Tripods 2
Tripods 3
Tripods 2
Tripods 2
Tripods 2
Tripods 2
Tripods 2
C Brakt 4
C Brakt 4
H Brakt 2
H Brakt 2

0.75
0.6
0.75
0.75
1.2
1.2
1.2
0.9
0.9
1.3
1.3

Non-gable
Non-gable
Non-gable
Gable
Gable
Gable
Gable
Gable
Gable
Gable
Gable

0.58
0.52
0.16
0.34
0.21
0.11
0.37
0.84
0.16
0.82
0.23

0.68
0.57
0.31
0.36
0.32
0.14
0.46
0.85
0.36
0.59
0.23

No
No
No
No
No
Yes
No
No
No
No
Yes

Figure 5. Sub-model stresses for dual tripod mount excluding bulk effects of wind and gravity.

due to the pole support feet increase by a smaller amount


than stresses in the lower wall. Therefore, because relative
limits are being used for the wall mounting studies, design
speed 2 gives similar but slightly more conservative limits
than design speed 1, so, only results for the former are
detailed.
At this point, it is worth noting the importance of wall
ties and floor bracing. If the first floor were removed, the
stress at the bottom of the non-gable wall would increase
from 0.165 to 0.20 MPa, and if there were no wall ties, it
would be 0.42 MPa. These results are for equal leaf stiffnesses of 2 GPa, but if the outer leaf stiffness was half that
of the inner leaf, the stress at the bottom of the wall would
decrease by around 30%. These results emphasize the
importance of obtaining a detailed understanding of the
building construction in assessing its suitability for a wind
turbine installation.
Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

3.1.1. Non-gable wall attachment


The first option considered was the dual tripod with the
supports 750 mm apart and with the support pole 450 mm
from the wall. For the large turbine with the wind acting
perpendicular to the outside surface of the wall, the average
mortar line stress across the worst brick was 0.58 MPa
which is well in excess of the limit and cracking is predicted at four of the six attachment points. With the wind
applied parallel to the wall, the worst brick stress was
0.68 MPa with cracking at three other attachment points
also likely (see Figure 5).
When the distance of the support pole to the wall was
decreased to 250 mm, the peak stress increased by about
5%. When three tripods were used instead of two, spaced
at 600 mm, the worst brick stresses for perpendicular and
parallel winds were 0.52 and 0.57 MPa, respectively, not
much lower than with twin tripods.

663

Mounting of building-integrated wind turbines

D. Udell, D. Infield and S. Watson

For the small (400 Watt) turbine installation, with two


supports, the worst brick stresses reduced to 0.16 and
0.31 MPa for the perpendicular and parallel winds, respectively. For the latter case, four attachment points failed the
criterion.
Only the tripod mounting systems have been considered
here because, as the next section shows, the other systems
cause higher stresses. This means that none of the mounting systems provides an acceptable solution for attachment
to the non-gable wall.

pendicular and parallel winds, respectively. For the small


turbine installation, both these stresses reduced to
0.23 MPa, this being marginally acceptable from a structural point of view.
For the large turbine supported by brackets 1.3 m apart
and with the turbine hub height 2.6 m above the top
bracket, the worst brick stresses were 0.82 and 0.59 MPa
for the perpendicular and parallel winds, respectively. For
the small turbine installation, both these stresses reduced
to 0.23 MPa, this being marginally acceptable from a
structural point of view.

3.1.2. Gable wall attachment


3.1.2.1. Tripod type attachment. For the large (1.5 kW)
turbine supported by dual tripods 750 mm apart and with
the support pole 450 mm from the wall, the worst brick
stresses were 0.34 and 0.36 MPa for the perpendicular and
parallel winds, respectively.
When the distance between the tripods was increased to
1200 mm, the worst brick stresses were 0.21 and 0.32 MPa
for the perpendicular and parallel winds, respectively, with
cracking predicted at two points. For the small turbine
installation, the worst brick stresses reduced to 0.11 and
0.14 MPa for the perpendicular and parallel winds,
respectively.
Using the widely spaced supports and a hub height of
5 m above the ridge, loads from the small machine gave
worst brick stresses of 0.37 and 0.46 MPa for the perpendicular and parallel winds respectively. It is estimated that
the small turbine installation would be acceptable at hub
heights up to 3.2 m above the ridge.

3.2. Roof timber attachment

3.1.2.2. Close to wall C bracket-type attachment. For


the large turbine supported by four brackets 0.9 m apart
and with the turbine hub height 2.6 m above the top
bracket, the worst brick stresses were 0.84 and 0.85 MPa
for the perpendicular and parallel winds, respectively. For
the small turbine, these stresses reduced to 0.16 and
0.36 MPa, respectively, with two attachment points predicted to crack in this latter case.

3.2.1. Assessment at design speed 2


With the wind approaching the non-gable end, the
arrangement in Figure 1, with the large turbine, gave a
peak bending stress of 5.8 MPa in the ridge board (see
Figure 6). This was the average stress across the ridge
board and includes both an allowance for the bolt hole and
the PSF for loads of 1.4. The final stress was well below
the 12.8 MPa target.
For a turbine mounted at the roof mid-span, the peak
stress for the non-gable wind increased by around 10% due
to the lack of support from the gable end wall.
Repeating the assessment with the small turbine
mounted 5 m above the ridge line gave a peak stress of

3.1.2.3. Close to wall H-type attachment. For the


large turbine supported by brackets 1.3 m apart and with
the turbine hub height 2.6 m above the top bracket, the
worst brick stresses were 0.82 and 0.59 MPa for the per-

For a roof without a turbine, the stresses due to wind and


gravity were found to be well below the limits given in
Section 2.4.2. Absolute stress limits could therefore be
used here rather than relative stress limits which had to be
used for the brick wall mounting (see Section 3.1). The
results for the roof timber mounting arrangements are summarized in Table II, at the end of Section 3.2.3.
Mounting from the ridge board and ceiling joists might
be desirable in that it minimizes the amount of support
required, and thus the potential cost. However, any vibration would be transmitted directly to the ceiling which
would be unacceptable to the occupants, particularly at
night. All options, therefore, have the bottom of the pole
supported by an additional horizontal beam (see Figure 1)
itself supported by rafters.

Table II. Roof mounting assessments for design speeds 1 and 2.


Roof description

Traditional
Traditional
Traditional
Traditional
Traditional
Traditional
Modern

664

Turbine power
(kW)

Height above
ridge (m)

Wind direction

Approx peak
stress (MPa)

Design speed

Acceptable?

1.5
0.4
1.5
0.4
1.5
0.4
1.5

2
5
2
5
2
5
2

Non-gable
Non-gable
Gable
Gable
Non-gable
Non-gable
Gable

5.8
7.9
4.4
6.1
12
17
34

1
1
1
1
2
2
1

Yes
Yes
Yes
Yes
Yes
No
No

Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

D. Udell, D. Infield and S. Watson

Mounting of building-integrated wind turbines

Figure 6. Uncorrected ridge board bending stresses for design speed 2.

7.9 MPa. Note that 30% of this stress was due to the aerodynamic drag on the support pole.
With the wind approaching the gable end, the peak
stress occurred in the rafters at the attachment points (see
Figure 7). When the bolt hole and the PSF of 1.4 were
accounted for, a realistic peak stress of 4.4 MPa was calculated for the large machine mounted 2 m above the ridge
line.
Repeating the assessment with the small turbine
mounted 5 m above the ridge gave a peak stress of
6.1 MPa. This is approximate only as much depends on
the diameter of the support pole.
The stresses have so far been assessed against the limit
parallel to the grain. This is reasonable at the ridge board,
but the rafters are at such an angle that cross grain stresses
become more important and these would increase as the
pitch angle of the roof increased. It is, therefore, recommended that some rig testing is carried out to establish
accurate limits. Alternatively, a significant reduction in
rafter stresses can be obtained by adding an additional
angled brace from the bottom of the support pole to a
position on the ridge board between the 3rd and 4th rafters.
This ensures that the non-gable wind is the limiting case.
This additional brace also markedly decreases the deflection of the wind turbine and thereby keeps the axis of blade
rotation closer to the horizontal giving better yawing
behaviour.
For the same heights above the ridge line, the roof is
capable of supporting turbines around twice the sizes of
the ones considered above. It is estimated that the large
machine could be mounted up to 4.2 m above the ridge
Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

line with acceptable stresses. For the pole, this would


necessitate the use of either a stronger material or a larger
diameter: the former increasing cost and the latter drag.
FE models were created with both fully fixed and
pinned joints at the rafter-to-ridge board and rafter-to-wall
plate connections. The models with pinned connections
gave stresses about 19% higher. As the real joints are
considered to be somewhere between the two, average
stresses are given here.
3.2.2. Assessment at design speed 1
Design speed 1 gives stresses just over twice those for
design speed 2. This means that the large machine at 2 m
and the small machine at 5 m produce peak stresses of 12
and 17 MPa, respectively, for wind approaching the nongable end. So, because relative stresses were used for the
brick wall limits, the roof capability could be said to be
only a little better than for the wall. However, this only
applies to extremely exposed and windy locations where
dwellings may well suffer damage on a once in 50-year
basis. If so, this would invalidate the use of relative stress
limits. In any case, design speed 1 is not realistic for urban
locations, so, the superiority of roof mounting is clear.
3.2.3. Modern trussed rafter roofs
Unfortunately, because modern roofs do not usually
have a ridge board, the arrangement in Figure 1 is not
possible. It is not even possible to add supports between
trusses, at ridge level, because the timber close to the ridge
level is hidden by the nail plates used to join the various

665

Mounting of building-integrated wind turbines

D. Udell, D. Infield and S. Watson

Figure 7. Unmodified end rafter bending stresses for design speed 2.

parts of the truss (Figure 8). The trusses are attached to


each other using braces at ceiling height and below the
ridge. The braces are typically fairly thin at 100 mm
25 mm but butt up against the brick walls where available,
for extra support.
None of these bracing elements provides the necessary
vertical face, and, if they were used (in conjunction with
wedges to create a vertical face), significant cross grain
stresses could be induced. The trussed rafter alone would,
therefore, need to provide all of the support. Unfortunately, trussed rafters tend to have smaller sections. For a
roof of the size considered here, sections 72 mm 35 mm
are typical. The bending and shear stresses, therefore, tend
to be greater than in the traditional roof, and the effect of
the bolt holes is more marked.
However, this is partly offset by the strength of the
timber. The wood used for the truss is grade TR26 which
is similar to grade C27 and much stronger than C14, in
particular, the bending stress limit along the grain is
24.7 MPa rather than 12.8 MPa.
For the large machine mounted at 2 m from the ridge,
wind approaching the gable end and design speed 2, the
arrangement in Figure 8 gives a peak stress over 40 MPa
when all effects are accounted for (18.5 MPa for turbine
loading alone, excluding the effect of the bolt hole).
Including an additional angled brace drops this peak stress
to 34 MPa and moves the peak stress position to the apex
of the truss. This suggests that for a turbine mounted 2 m

666

above the roof line, the size limit is around 2/3 of the large
machine, i.e. a swept area of 4 m2.

4. DISCUSSION
Computational fluid dynamics modelling from Renewable
Devices, cited in Renewable Devices,20 predicts the wind
speed over a 4 m pitched roof building where the pitch
angle is around 15o and the wind approaches perpendicular
to the building. This shows a maximum speed-up effect of
approximately 2 m s-1 but also high wind shear for 2 m
above the ridge height. However, it should be noted from
Encraft 20091 that this speed-up effect is reduced if the
building is within an urban area as opposed to an isolated
location. High wind shear generates high cyclic stresses
which reduce component life and this is in addition to
turbulence effects. As the pitch of the roof increases, wind
shear effects and turbulence are likely to increase, but
more work is required to quantify these effects. All that
can be said here is that to maximize energy capture and
turbine life, turbines should be mounted well above the
ridge. Manufacturers recommend that wind turbines be
mounted greater than 1 m above the roof line and the
height of 2 m used in this paper is commonly used. The
energy that can be captured will increase as the turbine is
mounted further above the ridgeline, but there are often
Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

D. Udell, D. Infield and S. Watson

Mounting of building-integrated wind turbines

Figure 8. Modern trussed rafter construction: peak stress position due to turbine load only.

planning restrictions to limit the mounting height to minimize visual impact.


For brickwork, the PSFs of 3 for material strength and
1.4 for load give a total safety factor of 4.2, which is rather
conservative. In aerospace applications, where the loads
and material properties are well understood, equivalent
factors might be 1.2 and 1.1 giving a total of just 1.3. It
would, therefore, be prudent, given the large potential
energy resource, for the wind and construction industries
to undertake further testing so that more accurate limits
can be established.

4.1. Comparison of mounting


arrangements
Mounting the turbine on the non-gable wall did not
produce a workable solution. The peak stress was almost
twice that for the equivalent gable end mounting. Gable
wall mounting only produced solutions for the small 400
Watt turbine. The best arrangement was the wide-spaced
dual tripod gable mounting.
It would be possible to reduce the mortar line stresses
by mounting the turbine closer to the corner of the building
rather than at the mid-span of the wall, but then, windows
and drain pipes would have to be accounted for. This
would also require a longer pole to maintain sufficient
height above the ridge line for the purposes of maintaining
energy yield.
It would be worth giving some further consideration to
the design wind speeds used. For a house, a 50-year return
Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

period seems entirely reasonable as excessive stresses


could destabilize an entire wall. However, for a wind
turbine, only mortar line cracking around a few bricks may
result, requiring some re-pointing after a heavy storm. All
the arrangements considered here had multiple attachment
points which should prevent the mounting arrangement
from failing completely. It may then be considered acceptable to use a shorter return period of, say, 20 years. BS
6399-23 then shows that the basic wind speed for Preston
would be reduced by 17% in this case, which would reduce
the loads by about one-third. However, it would be difficult to build this into a standard because of the multitude
of possible mounting arrangements which would have to
be considered.
It is worth noting that the wall integrity test carried out
by Windsave (http://www.windsave.comDecember
200722), using a point load of 4 kN perpendicular to the
wall, is a good test for assessing the capability of a wall
as it introduces an average vertical tensile stress of
0.61 MPa across a brick. This exceeds the stresses for
gable end tripod-type installations (see Section 3.1) for
design speed 2.
Unless cleverer ways of wall mounting are developed,
it seems that there is more potential in roof mounting. It
may be slightly more expensive in the first place (see
Section 5), but it offers both the potential of capturing
more energy and shorter payback times. For design speed
2, the capability of the conventional roof was estimated to
exceed that of the wall by a factor of around 4.
The capability of the trussed rafter mounting was much
less than that of the conventional roof and only a little

667

Mounting of building-integrated wind turbines

better than the best of the wall mountings. However,


because the trussed rafter is a single self-contained unit, it
would be possible to design a special truss capable of
withstanding the turbine loads and to brace this to the
standard trusses in order to spread the load. As the UK
government intends to build 3 million new homes
over the next 20 years, one could envisage a relatively
cheap mass-produced trussed rafter designed to mount
BIWTs.

D. Udell, D. Infield and S. Watson

6. CONCLUSIONS
From the research presented, a number of conclusions can
be drawn:

4.2. Health and safety and


installation issues
Wall-mounted turbines are reasonably well insulated from
the internal space of a house. Their only connection to the
inside wall is via wall ties and any vibration in these may
be damped, to some extent, by cavity wall insulation.
Roof-mounted machines have a more direct connection
with the inner wall via the rafters and wall plates. However,
between the turbine and the inner skin of the wall, there
are a large number of timber-to-timber and timberto-masonry joints which will provide damping. Wood also
has more internal damping capability than most other construction materials because of its multicellular construction and this limits the transmission of vibration energy.
Hornbostel23 quotes noise reduction coefficients for brick
as 0.02 to 0.05 and wood as 0.11 to 0.28. Some kind of
anti-vibration mounting is still likely to be required, but
this could be of a similar specification for roof-mounted
and wall-mounted machines alike.
A disadvantage of the roof-mounted machine is that
installation is likely to be more difficult, certainly for retrofits. Internal and external access to the roof will be
required and some form of scaffolding to do the necessary
roof modifications. Clearly, the wall-mounted machines do
not break through the external skin of the house and it may
be possible to install, at least the small machine, without
the use of scaffolding.

Both the 400 W and the 1.5k W wall-mounted turbines, mounted 2 m above the ridge line, cause mortar
stresses in excess of those suggested by BS5628,5 but
so do wind alone loads calculated by BS6399.3
Of the various wall mountings considered, the dual
widely spaced tripod arrangement was the best but
gave an acceptable solution for the small (1.9 m2
swept area) turbine only. It could support the small
machine to a maximum height of 3.2 m above the
ridge line.
A traditional close couple roof was shown to be four
times as capable of absorbing the loads as the brick
wall. It could support the large (6 m2 swept area)
turbine at 4.2 m above the ridge.
A typical trussed rafter roof was capable of supporting a machine with a 4 m2 swept area 2 m above the
ridge. However, it would be possible to design a
special truss capable of supporting larger machines
and fix this to standard trusses in the usual way.
More work is needed to understand the wind strength
above domestic housing, but mounting from roof
timbers is both viable and, at a small extra initial cost,
likely to provide solutions with more structural integrity and shorter payback times than those using brick
walls.

ACKNOWLEDGEMENTS
The author would like to express his gratitude to Mr A.
Wilson of NaREC for his helpful advice. Thanks also to
Mr J. Gardiner of J G Design Services for his development
of the FE code throughout the project.

REFERENCES
5. WALL AND ROOF MOUNTING
RELATIVE COSTS
Most of the support structure for a roof-mounted machine
is internal so, compared to a wall mounting, appearance
and corrosion are less important issues. For the supports,
much use could be made of conventional scaffolding.
Besides being relatively cheap and readily available, the
mechanical integrity of scaffolding tubes and couplings is
covered by BS1139-1.124 and BS1139-2.2.25 Overall costs
for roof and wall mounting are expected to be similar.
The main difference will be in the roof modifications
which involve the cost of replacing a ridge tile with a ridge
vent.26 This is likely to add only 510% to the total cost
of installation of a roof-mounted turbine compared to a
wall-mounted device.

668

1. Encraft 2009. Warwick wind trials final report.


[Online]. Available: http://www.warwickwindtrials.
org.uk/resources/Interim+Report+August+2008.pdf.
(Accessed 2 December 2009).
2. Heath MA, Walshe JD, Watson SJ. Estimating the
potential yield of small building-mounted wind turbines. Wind Energy 2007; 10: 271287.
3. BS 6399-2. Loading for buildingsPart 2: code of
practice for wind loads. 1997.
4. BS EN 61400-2. Wind turbine generator systems
Part 2: safety of small wind turbines. 1996.
5. BS 5628-1. Code of practice for the use of masonry
Part 1: structural use of unreinforced masonry. 2005.
6. BS 1995-1-1. Eurocode 5Design of timber structures. 2004.
Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

D. Udell, D. Infield and S. Watson

7. BS 1990. EurocodeBasis of structural design. 2002.


8. Martin N. Can we harvest useful wind energy from the
roofs of our buildings? Building for a Future; Winter
2005: 4652.
9. Dutton AG, Halliday JA, Blanch MJ. The Feasibility of Building-Mounted/Integrated Wind Turbines
(BUWTs): Achieving their potential for carbon emission reductions. Energy Research Unit, CCLRC, May
2005, final report on a project part-funded by the
Carbon Trust (2002-07-028-1-6), 118 pp.
10. BS 6399-1. Loading for buildingsPart 1: code of
practice for dead and imposed loads. 1996.
11. BS 5268-3. Structural use of timberPart 3: code of
practice for trussed rafter roofs. 2006.
12. BS 6399-3. Loading for buildingsPart 3: code of
practice for imposed roof loads. 1988.
13. Foster JS. Mitchells Structure and Fabric Part 1 (6th
edn). Longman: Harlow, Essex, UK, 2000; 78153.
14. Emmitt S, Gorse C. Barrys Introduction to Construction of Buildings. Blackwell: Oxford, UK, 2005;
6187.
15. BS 624. Schedule of weights of building materials.
1964.
16. Illston JM, Domone PLJ. Construction Materials:
Their Nature and Behaviour. Spon Press: London, UK,
2001; 281510.
17. Lee MK, Barr BIG. An overview of the fatigue behaviour of plain and fibre reinforced concrete. Cement and
Concrete Composites 2004; 26: 299305.

Wind Energ. 2010; 13:657669 2010 John Wiley & Sons, Ltd.
DOI: 10.1002/we

Mounting of building-integrated wind turbines

18. Singh SP, Mohammadi Y, Madan SK. Flexural fatigue


strength of steel fibrous concrete containing mixed
steel fibres. Journal of Zhejiang UniversityScience
A, 2006; 7: 13291335.
19. BS EN 338. Structural timberStrength classes. 2003.
20. Renewable Devices. Swift rooftop wind energy
system. Technical and Planning InformationRev F1,
2005.
21. Energy Saving Trust 2009. Location, location, location. Domestic small-scale wind field trial report.
[Online]. Available: http://www.energysavingtrust.
org.uk/cym/Global-Data/Publications/Locationlocation-location-The-Energy-Saving-Trust-s-fieldtrial-report-on-domestic-wind-turbines. (Accessed 8
April 2009).
22. Wind Save 2007. Wall strength testing. [Online].
Available: http://www.windsave.com. (Accessed December 2007).
23. Hornbostel C. Construction Materials: Types, Uses
and Applications. Wiley-IEEE: New York, USA,
1991, 910.
24. BS1139-1.1. Metal scaffoldingPart 1: tubes
Section 1.1 Specification for steel tube. 1990.
25. BS1139-2.2. Metal scaffoldingPart 2: couplers
Specification for couplers, fittings and accessories for
use in tubular scaffolding. 1991.
26. BCIS. The Property Makeover Price Guide. RICS:
London, UK, 2007, 72.

669

Das könnte Ihnen auch gefallen