A study on the design and TECHNICAL PAPER

material costs of tall wind Journal of the South African

Institution of Civil Engineering

turbine towers in South Africa Vol 57 No 4, December 2015, Pages 45–54, Paper 1231

A C Way, G P A G van Zijl ANDREW WAY, who is a student member of

SAICE, obtained his Bachelor’s (2012) degree and
his Master’s (2014) in Civil Engineering from
Stellenbosch University. This paper is a
The aim of this project was to study the structural design and material costing of various condensed version of his Master’s thesis, which
designs of tall wind turbine towers and the associated foundations in a South African context. focuses on wind power in South Africa and the
Design guidelines are proposed for the design of tubular steel, concrete and concrete-steel structural design of wind turbine support
structures.
hybrid towers and foundations for hub heights of 80, 100 and 120 m. The results indicate that
concrete and hybrid towers become viable alternatives to the conventional steel towers at hub Contact details:
heights equal to and above 100 m. Institute of Structural Engineering
Stellenbosch University
Three heights – 80 m, 100 m and 120 m – of each type of tower (steel, concrete and hybrid)
Private Bag X1
and their foundations were designed according to the relevant design standards. The designs Matieland 7602
were verified using the Abaqus CAE finite element software (SIMULIA 2010). The material costs South Africa
of the designs were calculated for a South African environment, according to the increases in T: +27 (0)72 679 0586
material cost with increasing hub height. E: [email protected]

In this paper, the required foundation sizes for the concrete and hybrid towers were found to
be smaller than for the steel towers. The material costs of the concrete and hybrid towers were PROF GIDEON VAN ZIJL (Pr Eng MSAICE) is
professor in structural engineering at
shown to be lower than for the steel towers, especially at hub heights above 100 m. An increase
Stellenbosch University. He obtained his
in hub height caused an increase in energy generation of 3.52% and 6.28% for 80 m to 100 m, Bachelor’s (1986) and Master’s (1990) degrees in
and for 80 m to 120 m hub heights, respectively. It is postulated that the concrete and hybrid Civil Engineering from Stellenbosch University,
towers become viable alternatives to the conventional steel towers at hub heights above 100 m. and his PhD (2000) from Delft University in the
Netherlands. As Director of the Centre for
Development of Sustainable Infrastructure, his
research interests are structural and computational mechanics, including the
INTRODUCTION 120 114 development, characterisation, and constitutive and durability modelling of
advanced construction materials.
Background and literature review 100
90 Contact details:
The introduction of the Renewable Energy Department of Civil Engineering
80 74
Price (R/kWh)

Independent Power Producer Procurement Stellenbosch University

Programme (REIPPPP) in August 2011 Private Bag X1
60 Matieland 7602
has led to a fast-growing renewable energy
South Africa
industry in South Africa, particularly in the 40 T: +27 (0)21 808 4436
wind power sector. In three years, South E: [email protected]
Africa has procured more investment in 20
independent power generation than had
0
been achieved across the African continent Round 1 Round 2 Round 3
for the past 20 years (Eberhard et al 2014). Price of wind energy for
The REIPPPP saw fast-tracked competi- REIPPP rounds
tion develop in the South African wind
power sector. This resulted in a reduction Figure 1 R
 eduction in the cost of wind energy
of the price of wind energy across the three over REIPPPP rounds
bidding rounds, as can be seen in Figure 1
(Eberhard et al 2014). The rapid decrease in (IRENA 2012). In 2012, approximately 90%
the price of wind energy is a sign of a fast- of installed wind turbine towers were of
maturing industry, although many criticise the tubular steel type due to their cost-
the sudden drop in prices (due to over- effectiveness and ease of construction
competitiveness in the industry), arguing (World Steel Association 2012). For hub
that this has negatively affected the potential heights of up to 80 m, tubular steel towers
for increased local content and sustainability. have proven themselves to be the most cost-
The lowered price for wind energy has never­ effective solution. As the tower hub heights
theless made it one of the main renewable increase to 100 and 120 m, however, steel
energy sources in South Africa. towers start to lose their appeal (Harte &
The current trend in the global wind van Zijl 2007).
industry is to use taller wind turbine towers One of the main reasons for the increas-
to access the stronger and less turbulent ing hub height trend stems from the current
wind resources that occur at greater heights situation of wind power in Europe. Many of Keywords: wind turbine, structural design, material cost, concrete, steel

45
Way AC, Van Zijl GPAG. A study on the design and material costs of tall wind turbine towers in South Africa.
J. S. Afr. Inst. Civ. Eng. 2015;57(4), Art. #1231, 10 pages. http://dx.doi.org/10.17159/2309-8775/2015/v57n4a6
the most favourable wind sites have already
been exploited by currently operating wind –28
farms, leaving only low to medium wind
resource sites. In addition to this, recent –29
WM01

advances in low to medium wind resource

technology allow for the exploitation of sites –30
that were previously considered to have
unprofitable wind resources.

Latitude
–31
As the hub height increases, the towers WM09

need to be able to resist increased ultimate –32 WM03

WM02

loads, bending moments and increased WM10 10

9
fatigue loads and moments. This is as a –33 WM04
WM06
8
WM07 7
result of the greater wind loads acting on the 6
5
tower, nacelle and blade assembly. Also to –34 4
be taken into consideration is the stiffness WM08 Mean wind speed (ms–1) @ 100 m 3
WAsP modelled, 250 m resolution 2
requirement of the tower with regard to the –35
WM05

interaction between the natural frequency of 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

the tower and the rotational frequency of the Longitude
turbine (Nicholson 2011). In order to satisfy
these requirements, either the tower shell Figure 2 WASA wind resource map modelled in WAsP wind analysis software (WASA Project 2014)
thickness needs to be increased, or the tower
base diameter needs to be increased. The currently no commercial scale steel towers AIMS AND METHODOLOGY
main problem with the steel tower occurs that further split the sections into segments, This paper aims to:
when tower base diameters are required to due to the fatigue loads that would act on a. Acquire and analyse South African wind
be larger than the allowable road transporta- the connections between segments. In addi- data, ranging from 80–120 m above
tion height limit of 4.5 m. tion, even the smallest of manufacturing ground.
The development of turbines with imperfections in connection components b. Study the design of wind turbine support
nameplate capacity (i.e. maximum power cause high stress concentrations that lead to structures and foundations for steel, post-
generating capacity at optimal wind speed) fatigue failure. tensioned concrete and concrete-steel
in excess of 3 MW has also created the need hybrid type towers.
for access to stronger wind resources. The Precast concrete tower c. Determine whether an increase in tower
added weight of these larger turbines and The precast concrete tower is generally height is viable for South Africa or not, by
blades requires an increase in the structural manufactured off site in sections and fur- calculating the increase in the tower and
strength of the towers. It is generally due to ther into segments, which alleviates the foundation material costs as a function of
these reasons that different tower material transportation problems associated with the tower height.
and designs are currently being employed as large base-diameter sections prevalent in the d. Develop guidelines for the South African
an alternative to the conventional steel tower taller steel towers. The segments are then wind industry with regard to the material
for taller hub heights. transported to site where they are placed, costs and structural design of tall wind
grouted and post-tensioned. These towers turbine towers.
Tower types have distinct advantages, other than the ease Initially a literature study was conducted on
The three most common designs for tall of transportation, in that the thicker con- the global and South African wind industries
wind turbine towers are: crete sections are stiffer than a typical steel to gain a better understanding of the current
■■ the conventional tubular steel tower tower section. This allows for reduced lateral status of, and trends in, the wind industry.
■■ the precast, post-tensioned, segmented deflections, longer fatigue life and higher An investigation into the wind resource and
concrete tower tower natural frequencies. The disadvantages the available wind resource information in
■■ the concrete-steel hybrid tower. include the obvious lack of tensile strength of South Africa was performed to justify the
the concrete, creating the necessity for post- use of taller wind turbine towers. A total of
Tubular steel tower tensioning and the need for increased crane nine towers (concrete, steel and hybrid with
The conventional steel tower is manufac- hire time. heights of 80, 100 and 120 m) and their foun-
tured in 20–30 m sections that taper in dations were then designed according to the
diameter and shell thickness from top to Concrete-steel hybrid tower current international design methods.
bottom. This is the most common type The hybrid tower generally consists of a The respective tower designs were then
of tower in the world, and as such there lower post-tensioned concrete section, rang- used to determine the material costs associ-
are established manufacturers around the ing from 40–80 m, and an upper steel sec- ated with increased hub height. Thereafter
world who have optimised the tower for hub tion (Nordex 2007). The lower sections can the increase in revenue generated as a result
heights of up to 80 m. This tower type has be cast in-situ, but are generally manufac- of increases in hub height was determined.
the advantage of rapid construction, as there tured in segments like the precast concrete The last two processes were carried out to
are only three or four sections that need to tower and transported to and assembled on develop an indication of whether increasing
be lifted into place. As previously mentioned, site. This type of tower combines the benefits the hub height is a viable option in a South
the base diameter is limited due to transport of both the steel and concrete towers, and African context.
constraints, which causes a notable increase only has the disadvantages of the concrete Note that the cost analysis in this paper is
in tower shell thickness with increasing tower in the form of the post-tensioning limited to the material cost. Clearly total cost
hub height (Cotrell et al 2014). There are requirement. should be considered when alternatives are

46 Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015
 10 – Overturning
moment
Wind presure – Turbine
profile mass
9 e
– Thrust force
– Torsional
Mean annual wind speed (m/s)

moment
8
– Eccentricity
moment

z
5

4
0 20 40 60 80 100 120
Height above ground (m)
Humansdorp Napier Beaufort West Vredendal
Sutherland Vredenburg Calvinia Alexander Bay Figure 4 I llustration of loads acting on wind
turbine structure
Figure 3 Increase in mean annual wind speed vs height (2011)
the wind, assuming a yaw-misalignment of
compared for particular wind farms. For this have anemometers at altitudes of 10, 20, 15 degrees. The design wind speed, Ve50(z), is
emerging technology in the South African 40, 60 and 62 m in order to get an accurate distributed along the tower according to IEC
context it is left to the design team to ­representation of the wind profile at the 6400-1:2005, where Vref is the ten-minute
consider particular scenarios of overall cost, given sites. The data was condensed and mean wind speed and z denotes height:
including, amongst others, the construction logarithmic extrapolation techniques were
z 0 .11
method (for instance slip or climbing form- employed by the authors to extend the data Ve50(z) = 1.4 ∙ Vref (1)
work versus hybrid construction requiring to an altitude of 120 m. The data for eight zhub
large crane-hoisting capacity); logistics of of the ten masts (two contained large gaps
on-site precast versus precast in existing in data) were used to calculate the increase Note that in Equation 1 Vref is the wind
technology hubs where specialised labour in wind speed as a function of hub height, speed at the hub height zhub. The EWM is
and materials are available, and subsequent as can be seen in Figure 3. The average a case of an ultimate limit state (ULS), and
transportation to site; and the transportation increases in mean wind speed values from the design reflects this state. Load factors as
of steel tower segments from producers in 62 m hub height to 80, 100 and 120 m are prescribed in IEC 6400-1:2005 were used,
industrial zones (Coega in the Eastern Cape, 4.1%, 7.2% and 9.8% respectively. in conjunction with factors from EN 1997-
and Atlantis in the Western Cape) to site, 1:2004 (Eurocode 7) (EN 1997 2004) that
etc. Here light shed on possible dispropor- are not contained in IEC 6400-1:2005. The
tional raise in material cost with increase in DESIGN pressure distribution around the circumfer-
tower height may be of significant value to The design philosophy of the International ence of the tower is done in accordance
designers, although it is acknowledged that Electrotechnical Commission IEC 6400- with section 8.10 in SANS 10160-3:2009,
such additional cost may be offset by savings 1:2005 Wind turbines–part 1: Design with particular use of Figure 29. In addition
in other cost items. requirements (IEC 2005) was followed. Each to the wind loading on the tower, there
of the nine tower-and-foundation combina- are also wind loads that act on the blades
tions was subjected to the Extreme Wind and nacelle. The loads are summarised in
WIND RESOURCE ANALYSIS Speed Model (EWM), as set out in IEC Figure 4.
The wind resource analysis was based on 6400-1:2005.
data obtained from the Wind Atlas for South Loads on foundations
Africa (WASA 2014) project. This project Wind loads The loads from the tower transfer to the
aimed to set up a numerical wind atlas In the EWM, the wind turbine is subjected square foundation directly through an
database for South Africa, a sample of which to an extreme three-second wind gust (one- anchor cage. The loads that act on the foun-
can be seen in Figure 2, which uses colour in-fifty year return period) of 52.5 m/s at dation are thus the sum of the loads that act
coding to differentiate between areas of high hub height, as for IEC class IIIA. This value on the wind turbine. Two cases are assumed.
and low mean wind speeds. was compared to the equivalent in SANS The first is for the case of wind acting in
The project erected ten wind masts 10160-3:2009 (SANS 2009) and was found to the x-direction, as shown in Figure 4. The
around South Africa (Western, Northern be more conservative than the wind speeds second case is for wind acting at 45 degrees
and Eastern Cape) in order to verify the in the presence of even the worst terrain to the x-direction, as seen in plan, which
database. This data was made freely available category (Category A). In this circumstance, creates the necessity for the design of the
to the public and currently has three full the wind turbine is in a non-operational, foundations to consider these two different
years’ worth of data (2011–2013). The masts parked state, with the blades feathered out of wind orientation cases.

Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015 47
 typical wind turbine site along the coast of
South Africa (c = 60 kN/m2, φ = 30°, defined
below).
The foundations were designed with
particular attention to the resistance against
overturn, resistance against sliding of the
base, soil and foundation stiffness, tensile
reinforcing and resistance against punching
1P 3P shear. The calculation of the soil’s bearing
capacity, q f, is carried out according to Craig
(2004), where ii refers to inclination factors,
si refers to foundation shape factors, Ni refers
to bearing capacity factors and cd, γ, De and
b denote the soil cohesion intercept, the bulk
density of the soil, the embedded depth of
the foundation and the effective breadth of
the foundation respectively:
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Frequency (Hz) 1
q f = cdNcscic + γDeNqsqiq + ∙ γbNγsγ iγ(3)
2
Figure 5 Tower natural frequency exclusion zones (red)
The sliding resistance of the foundation is
not usually a governing factor, but should
ds
be checked nevertheless, according to DNV/
Fc Fw Ft Risø (2002), for drained and undrained soil
conditions respectively, where φ denotes
the design soil angle of shear resistance, Aeff
Fxy denotes the effective area of the foundation,
Vd is the vertical design load and Hd the
eq horizontal design sliding force.
gd
Hd < Aeff ∙ cd + Vd ∙ tanφ(4)

σsoil Hd < Aeff ∙ cd(5)

2b One of the most important criteria of a foun-

dation is to prevent overturn of the structure
Figure 6 Illustration of foundation flexural reinforcing and applied loads by an acceptable factor of safety. Put simply,
the sum of the stabilising moments, ∑ MR,
Tower natural frequency reproduced from Manwell et al (2010) in must outweigh the sum of the overturning
A crucial aspect of the design of any tower is Equation 2, and is accurate to within 15%, moments, ∑ Mo, by a factor, Fs:
the need to separate the natural frequency of but will be verified using the Abaqus FEM
the tower from the blade-passing frequency software. The natural frequency of the tower Fs >
∑MR (6)
of the turbine. The natural frequency of the is affected by the tower material Young’s ∑Mo
tower must be completely separated from the Modulus, E, the second moment of inertia
frequency range at which one blade passes of the tower, I, the tower length, L, and the The stiffness of the foundation is incorporat-
the tower, across the operational frequency mass of the rotor and tower. ed into the FEM model using linear springs,
range of the turbine (denoted by 1P), plus calculated according to the equations from
15% above and below. Similarly, the three 1 3EI section 8.4 of DNV/Risø (2002) for vertical,
fn = ∙ L3 (2)
blade passing frequency (3P) at rated power 2π 0.23 ∙ mtower + mrotor horizontal, rocking and torsional stiffness.
generation must be avoided by 15% on either The formulae are based on work done by
side (ASCE/AWEA 2011). This is to ensure For the turbine used in this study, a Vestas Gazetas (1983) and Elsabee (1973).
that there are no problems associated with V112 3MW, the feasible allowable tower The flexural reinforcing in the founda-
resonance, due to the interaction between natural frequency zone, is shown as being tion is present in both the top and bottom
the turbine and the tower. between 0.245 Hz and 0.544 Hz in Figure 5. edges to resist the tensile stresses caused
Resonance between the tower and the by the overturning moment from the wind
blade passing frequencies causes highly Foundation design loading, as shown in Figure 6. The design
increased deflections of, and vibrations in, The foundations were all designed to be of the reinforced concrete foundation is
the tower. Resonant effects lead to increased square shallow-gravity foundations. Water performed according to EN 1992-1-1:2004
and premature fatigue damage or, in the depths, varying from well below the founda- (EN 1992 2004).
worst case, catastrophic failure of the tower. tion to ground level, were used in the design The foundations were designed to negate
An approximate equation for the calcula- of the foundations. The soil conditions the need for punching shear reinforcing. The
tion of a tower’s natural frequency, f n, is for this project were chosen to represent a requirements were checked using section

48 Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015
6.4.3 of EN 1992-1-1:2004. The shear force, Table 1 Pre-stressing losses information
vED, at the first control perimeter, u1, with
Tower κ x α δanchor δrelax
an effective depth, deff, must be less than the µ
allowable shear stress for sections without Unit (rad/m) (m) (rad) (%) (%)
shear reinforcing, vRd,c:
Concrete 80 m 0.1 0.0016 80 0.028 3 2.5

β ∙ V ED, reduced Hybrid 80 m 0.1 0.0016 40 0.036 3 2.5

vED = < vRd,c
uI ∙ deff Concrete 100 m 0.1 0.0016 100 0.023 3 2.5
= CRd,c ∙ k(100 ∙ ρI ∙ fck)⅓(7)
Hybrid 100 m 0.1 0.0016 60 0.024 3 2.5

where: Concrete 120 m 0.1 0.0016 120 0.019 3 2.5

e
β = 1 + 0.6π ∙ Hybrid 120 m 0.1 0.0016 80 0.018 3 2.5
Dtower + 4deff
e; Dtower = eccentricity of applied load;
tower diameter Table 2 Effect of mesh size on analysis accuracy
V ED, reduced = reduced vertical force acting
Element Elements over Natural
on control perimeter size (mm) shell thickness
Analysis time (s)
frequency (Hz)
Error (%)
0.18
CRd,c =
γ m,c 2 000 1 2 0.88 4.97

γ m,c = partial material factor for 1 000 1 2 0.91 1.73

concrete
200 600 1 3 0.918 0.86
k = 1 +
deff 500 1 3 0.92 0.65
ρI = average reinforcing ratio
250 1 13 0.924 0.22
fck = characteristic cylinder
strength of concrete 125 2 117 0.925 0.11

80 4 20 935 0.926 0.00

Steel tower design
The design of the steel tower is generally
governed by fatigue (not covered here), cylinder strength, fcd, cylinder , of the 50 MPa Concrete-steel hybrid tower design
buckling resistance or stiffness requirements concrete (taking tension as negative and com- As previously mentioned, the hybrid tower
(either to limit deflections or to satisfy pression as positive) as shown in Equation 9. consists of a lower precast concrete section
natural frequency requirements). The buck- The symbols A, y, Nps and Mps denote the and a tubular steel tower upper section. The
ling requirements, as laid out in DNV/Risø cross-sectional area, distance to the extreme individual concrete and steel parts are designed
(2002), are satisfied through the inequality as fibre of section, axial force and moment with the same criteria as mentioned above for
shown in Equation 8. The buckling require- caused by the post-tensioning, respectively. the steel and concrete sections respectively.
ments consider the axial force, Nd, bending
moment, Md, tower shell thickness, t, tower Nd Nps Md ∙ y Mps ∙ y
fctmd < + ± ±
radius, R, and the Euler elastic buckling load, A A I I FEM ANALYSES
Nel. A combination of the applied axial force < fcd, cylinder (9) All nine of the designs were modelled using
and bending moment must be less than the the Abaqus CAE FEM software. The tower
critical compressive stress, σcr : The losses in pre-stressing force associ- and foundation combinations were modelled
ated with post-tensioning (wobble friction, using 3D solid elements. Initially there was
Nd Nel Md curvature friction, strand relaxation, elastic concern as to whether the 3D elements would
+ ∙ ≤ σcr (8)
2πRt (Nel – Nd) πR2t shortening, as well as anchoring) were con- result in sufficient accuracy, due to the small
sidered using Equation 10, adapted from EN number of elements required across the
The stiffness of the tower is acceptable when 1992-1-1:2004 in conjunction with the values thickness of the tower shell. Four elements
the natural frequency of the tower is within shown in Table 1. over the tower shell thickness resulted in the
the acceptable limits, as shown in Figure 5. number of elements in the model reaching
Wind turbine manufacturers sometimes δtotal = (e–(µα + κx) + δanchor + δrelax)(10) into the millions. The mesh size was thus
limit the maximum lateral deflection of the varied from four elements to one element
tower, but such a limit is not included in where: over the shell thickness and the accuracy was
this paper. δtotal = fraction of pre-stressing force lost compared for a natural frequency analysis.
to friction effects The results are shown in Table 2. Note that
Concrete tower design µ = curvature friction coefficient a simplification is possible by using shell ele-
The design of the concrete tower is dominated α = angle change along tendon length ments. In an accompanying paper (Van Zyl
by the tensile resistance (or the lack thereof) (rad) & van Zijl 2015) (see pages 38–44 of this edi-
of the concrete. The tower therefore employs κ = wobble friction coefficient (rad/m) tion), where physical inelasticity in the form
post-tensioning in order to limit the tensile x = tendon length (m) of cracking was incorporated, this was in fact
stresses in the tower. The tower is designed so δanchor = fraction of prestressing force lost done for feasibility.
that the stresses that develop in the concrete due to anchorage It was concluded that mesh sizes up to
sections are greater than the design mean ten- δrelax = fraction of prestressing force lost three times the tower shell thickness yield
sile strength, fctmd, but lower than the design to tendon relaxation effects results with errors of less than 1.5%, but only

Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015 49
Table 3 Tower design dimensions

Steel tower dimensions Concrete tower dimensions

Parameter Unit 80 m 100 m 120 m Parameter Unit 80 m 100 m 120 m

Top outer diameter (m) 3 3 3 Top outer diameter (m) 3 3 3

Bottom outer diameter (m) 4.5 4.5 4.5* Bottom outer diameter (m) 7.5 7.5 7.5

Top shell thickness (mm) 15 15 15 Top shell thickness (mm) 200 250 250

Top shell thickness (mm) 34 55 75 Top shell thickness (mm) 275 325 350

Hybrid towers – steel section dimensions Hybrid towers – concrete section dimensions

Parameter Unit 80 m 100 m 120 m Parameter Unit 80 m 100 m 120 m

Top outer diameter (m) 3 3 3 Top outer diameter (m) 4.6 4.6 4.6

Bottom outer diameter (m) 4.3 4.3 4.3 Bottom outer diameter (m) 7.5 7.5 7.5

Top shell thickness (mm) 25 25 25 Top shell thickness (mm) 200 200 300

Top shell thickness (mm) 40 40 40 Top shell thickness (mm) 200 200 300

Concrete section height (m) 40 60 80

Table 4 Square foundation design dimensions

Concrete Foundation steel

Tower Tower Breadth Depth
volume reinforcing
height type (m) (m)
(m3) (ton)

Steel 21 2.2 606 53.12

80 Concrete 20 1.5 459 38.27

Hybrid 20.5 1.75 598 43.98

Steel 21.75 2.5 707 62.85

100 Concrete 20.25 2.05 622 47.04

Hybrid 21.75 2.35 721 62.89

Steel 23.25 2.75 921 82.46

120 Concrete 22.75 2.45 774 72.81

Hybrid 22.75 2.45 832 71.65

factor relative to the buckling load to describe ■■ Own weight of tower and turbine
how safe or under-designed the model is. A ■■ Torsional moment on the tower
Figure 7 I llustration of loads for the FEM static factor of 1 indicates that the model is exactly ■■ Tower top weight eccentricity
and buckling analysis on the brink of failure due to buckling. Values ■■ Vertical post-tensioning loads on the
less than 1 indicate failure, and values over 1 tower, where applicable.
when the stresses and strains across the shell indicate safety against buckling failure.
thickness are not of importance. The final analysis was a two-step static Effect of underlying soil
The models consisted of 8-node linear load analysis with the ultimate limit state The effect of the underlying soil is taken into
hexagonal “brick” elements for the tower and loads applied to the wind turbine structure. account through the use of linear springs.
non-inclined portions of the foundations, with This was carried out with the intent of verify- The spring values are based on the previously-
4-node linear tetrahedral elements being used ing the hand-calculated stresses and to check mentioned work by Gazetas (1983) and
for the inclined portion of the foundations. tower-top deflections. The second part of Elsabee (1973) found in DNV/Risø (2002). The
the analysis then used the deflections of the springs were distributed around the underside
Analyses performed towers to determine the additional moments of the foundation in the Abaqus model, simu-
Each of the nine tower and foundation combi- exerted on the tower due to the permanent lating the vertical, horziontal, rocking and
nations was subjected to three analyses. First, loads acting at an eccentricity from the static torsional stiffness of the underlying soil.
a modal frequency analysis was performed to centre of gravity (known as the P − ∆ effect).
accurately determine the natural frequency
of the model and determine if it satisfied the Loads RESULTS
natural frequency range requirements. The loads for the static and buckling analy-
The second analysis involved the calcula- ses included, as illustrated by Figure 7, are: Tower and foundation dimensions
tion of the buckling stress of each tower. In ■■ Wind loads acting directly on the tower The final tower and foundation dimensions
this analysis, the ultimate loads are applied to ■■ Wind loads acting on the nacelle, blades that satisfy the requirements as described
the structure and the analysis determines a and nose-cone above are shown in Tables 3 and 4.

50 Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015
 80 m steel S.F.D. 80 m steel B.M.D.
10 30

Bending moment (MNm)

5 20
Shear force (MN)

0 10

–5 0

–10 –10

–15 –20
0 5 10 15 20 0 5 10 15 20
Foundation length (m) Foundation length (m)

Figure 8 Shear force and bending moment diagrams for the 80 m steel tower foundation

steel and hybrid tower foundations were all

governed by the need for foundation weight
to counter the overturning moment from the
wind loading. The 80 m concrete tower was
governed by the same case, but as the weight
increased, the 100 m tower foundation was
governed by simultaneous overturn and
bearing capacity failure.
As the weight increased even more, the
120 m concrete tower foundation design
was governed by the bearing capacity of the
Figure 9 Natural frequency modes of the concrete 80 m tower soil. The fictional site had quite favourable
soil conditions, and so the bearing capacity
The asterisk in Table 3, under the 120 m weight to counter the tower overturning was not prevalent. In weaker soils (lower cd
steel tower, indicates that this tower does moment, to provide extra foundation height and φ values), bearing capacity is likely to be
not satisfy the natural frequency stiffness negating the need for punching shear more prevalent as the governing foundation
requirements. A frequency separation of only reinforcing, to increase the spread area of design parameter. Thus in the presence of
12% could be achieved between the 1P and the foundation to prevent bearing capacity weak soils, the steel and hybrid towers will
tower natural frequency, despite keeping the failure, or to provide more weight at the base be more appropriate, as they are lighter than
tower diameter at 4.5 for the first 45 metres in order to raise the natural frequency of the the concrete towers.
of the tower. As can be seen, the concrete system.
sections are notably thicker than their steel An example of the shear force and FEM Results
counterparts. This is due to concrete’s rela- bending moment diagram for the 80 m
tively limited capacity for tension and com- steel tower foundation is shown in Figure 8. Natural frequency analyses
pression strength, as well as a lower material The maxima and minima from the graphs A natural frequency analysis was car-
stiffness in comparison to steel. will be even greater for IEC wind classes II ried out on all nine of the models, all of
As can be seen in Table 4, the founda- and I. which, except the 120 m steel tower, satis-
tion volume is proportional to the tower The foundation designs were all gov- fied the natural frequency requirements
height. An increase in tower height requires erned by the case where the water table is (0.245 Hz < f n < 0.544 Hz). The hand
an increase in foundation concrete volume at ground level, as a considerable buoyancy calculations compared well with the FEM
for one of the following reasons: to provide force acts upwards on the foundation. The results, the greatest error in natural fre-
quency hand calculation being 12.6%. The
Table 5 Results of FEM analyses first four natural frequency modes of the
80 m concrete tower are shown in Figure 9.
Tower Tower Natural frequency Tower deflection
height  type  (Hz)
Buckle value 
(m)
The first (lowest) natural frequency of each
tower can be seen in Table 5. As can be seen
Steel 0.285 1.65 0.92 in Table 5, the concrete and hybrid towers
80 Concrete 0.432 10.1 0.37 do not have problems associated with low
natural frequencies (lack of tower stiffness).
Hybrid 0.407 2.25 0.52
Conversely though, they can sometimes
Steel 0.251 2.84 1.35 have natural frequencies that are too high
(overly stiff tower), depending on the choice
100 Concrete 0.333 4.65 0.60
of turbine.
Hybrid 0.338 2.25 0.80
Buckling analyses
Steel 0.238 3.48 1.48
Similarly, a buckling analysis was carried
120 Concrete 0.261 2.38 0.99 out on all nine models, the results of which
compare well with the buckling stress hand
Hybrid 0.297 2.27 1.06
calculations and confirm that the buckling

Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015 51
 Concrete 80 tension side Concrete 80 compression side
20 35

30
15
25
Stress (MPa)

Stress (MPa)
10
20

15
5

10
0
5

–5 0
0 20 40 60 80 0 20 40 60 80
Tower height (m) Tower height (m)
Abaqus Hand calculation Tensile limit Abaqus Hand calculation Compressive limit

Figure 10 Comparison of stresses from hand calculations and FEM results

Table 6 Tower pre-stressing force requirements Tower stresses

The tower stresses were obtained from the
Total PS force in tower PS force for costing
Height Tower Losses
(MN) (MN)
static analysis carried out on all the designs.
As predicted, the hand calculations are con-
Concrete 17.8% 23.9 33.4 servative in comparison to the FEM results,
80
Hybrid 12.0% 33.2 43.4 as illustrated in the graphs in Figure 10. This
figure illustrates the tension and compression
Concrete 20.5% 39.3 56.8 side (windward and leeward side) of the 80 m
100
Hybrid 14.9% 49.5 66.8
concrete tower. Tensile stresses are shown as
negative, and compression as positive values.
Concrete 23.1% 60.3 90.2 One can see that the hand-calculated ten-
120
sion stresses vary from the FEM values. This
Hybrid 17.7% 64.6 90.2
is likely due to the fact that the hand calcula-
tions assume that the tower is completely
Table 7 Component/material cost summary fixed at the base, whereas the FEM model
considers the non-fixity of the base (in the
Material/Component Unit Value
form of the springs on the underside of the
Foundation concrete (R/m3) 1 400 foundation, allowing small deflections due to
the elasticity of the soil).
Tower concrete (R/m3) 2 007
Most importantly, the extreme values
Reinforcing steel (R/ton) 15 251 (tension in concrete and compression in steel)
in the FEM outputs are less critical than the
Tower plate steel (R/ton) 18 912
hand calculation, which indicates safe design.
Pre-stressing: anchors and couplers (R/tower) 33 440
Pre-stressing
Pre-stressing: tendons and support clips (R/MN/m) 120
The loss in pre-stressing force for the concrete
towers is considerably higher than that of the
capacity is not a governing factor in the of the turbine and are thus of little concern, hybrid towers. This is due to the extra length
design of the towers studied in this project. whereas excessive ULS deflections are more of pre-stressing required in the concrete
The results of the buckling analyses can be likely to damage the turbine. towers. Even though the concrete portion of
seen in Table 5. It should be noted that the Similar studies prescribe a rule-of-thumb the hybrid tower is shorter than the concrete
hand calculations were acceptably conserva- maximum tower top deflection of 1.25% tower, the hybrid tower requires more overall
tive for all nine designs. of the tower height in order to protect the pre-stressing force than the concrete towers,
turbine against deflection-induced dam- as can be seen in Table 6. This is mainly due
Tower deflection age (Nicholson 2011). One can see that the to the concrete sections in the hybrid towers
The deflection of the towers at the ULS is deflections of the steel towers are consider- being thinner than the concrete towers.
also shown in Table 5. Deflections are not ably higher than that of the stiffer concrete
calculated at SLS, as wind speeds at SLS only and hybrid towers. Tower top deflections are Material cost comparison
reach 25 m/s before the turbine automatical- sometimes limited by turbine suppliers and The costs used in the material cost compari-
ly shuts down and its blades are feathered out it is thus a possibility that the steel towers son were obtained from South African man-
of the wind to reduce excess drag at higher may have to be further stiffened in order ufacturers and suppliers, and are exclusive of
ULS wind speeds (52.5 m/s). Furthermore, to comply with this requirement, although labour, due to the material cost comparison
due to the nature of the structure, SLS all of them are lower than 1.25% of the nature of this project. The costs used are
deflections do not hinder the functionality tower heights. an average of the 2014 prices obtained from

52 Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015
various suppliers/manufacturers in South Table 8 Tower material use and cost
Africa and can be seen in Table 7. The mate-
Tower material use and cost
rial cost of the tower and foundations are
shown in Figure 11 and Table 8 and include Steel in Concrete
Pre-
Tower
Pre-
Tower
stressing stressing
the tower material used, the foundation con- Tower tower in tower length cost
force cost
crete and reinforcing and the pre-stressing (ton) (m3) (m) (R)
(MN) (R)
material costs.
80 183.6 – –     3 471 608
Material costs are not the only costs
associated with the production and erection Steel 100 330.6 – –     6 252 465
of the tower and foundation. Other costs
120 685.7 – –     12 968 527
associated with transport, and labour and
lifting costs, amongst others, also play a 80 – 306 33.4 80 354 157 968 705
role in determining the cost of a finished
Concrete 100 – 457 56.8 100 714 963 1 632 051
product. As can be seen from Figure 11, the
concrete and hybrid towers are less material- 120 – 608 90.2 120 1 332 343 2 552 493
cost-intensive than the steel towers, par-
80 72.2 147 43.4 80 450 372 1 661 436
ticularly for the 100 and 120 m towers. The
trend seems to show that the hybrid towers Hybrid 100 72.2 221 66.8 100 835 232 1 808 977
will become more cost effective than the
120 72.2 434 90.2 120 1 332 343 2 236 468
concrete towers at hub heights greater than
110–115 m. The steel towers are shown to be
disproportionately material-cost-intensive at 16
hub heights greater than 100 m.
14

Increase in revenue generation 12

The data obtained from the WASA project 10

Cost (Rm)

for the Napier mast was used to calculate the 8

increase in revenue generated as a function
6
of increases in hub height. This specific site
has a middle-of-the-range increase in wind 4
speed as function of hub height, and serves 2
as a good example of how increasing the hub
0
height can be beneficial to both investor and 70 80 90 100 110 120 130
power utility. In Table 9 the annual average Tower height (m)
wind speed for the three years of data (2011 Steel Concrete Hybrid
to 2013) are given for the 80 m hub height,
as well as the logarithmically extrapolated Figure 11 Material cost comparison
values for the 100 m and 120 m hub heights.
A single Vestas V112 3MW turbine was used 3 500
in the calculation of the increase in energy
generated, with the 80 m hub height as refer- 3 000
Power generated (kW)

ence. For the total annual energy generated,

2 500
the following expression is used:
2 000
N 1
Ea = ηelecηunavail ∑ Pi(vi )(11) 1 500
i=1 6
1 000
where vi is the wind velocity for the particular
site obtained from WASA and extrapolated 500

to the relevant hub height, Pi(vi) is the power 0

generated by the particular turbine at the 0 5 10 15 20 25
particular velocity vi, ηelec is a coefficient Wind speed (m/s)
correcting for losses in generation and feeding
into the grid, and ηunavail is a coefficient for Figure 12 Power curve for the Vestas V112 3 MW turbine
losses due to unavailability of the turbine. See
Figure 12 for the particular power curve used unit price R0.74/kWh (see Figure 1) of REIPPP CONCLUSIONS
in this paper. Note further that N is the total Round 3. For an increase in hub height The designs of the wind turbine foundations
number of wind readings per year, i.e. six per from 80 to 100 m and from 80 to 120 m, an were highly dependent on the choice of under-
hour for 24 hours and 365 days per year, and increase in revenue of 3.52% and 6.28% was lying soil parameters. Given the favourable
typical values of ηelec = 0.97 and ηunavail = 0.95 found. The results can be seen in Table 9. soil conditions used in this project, the design
have been used here (Feng & Tavner 2010). It should be noted that the 20 year average of the foundation for the steel and hybrid
Subsequently the yearly revenue is calculated shown in Table 9 is an extrapolation of the towers was governed by the weight of the
by multiplying the annual energy Ea with the three years of data that was analysed. foundation needed to stabilise the structure

Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015 53
Table 9 Revenue generation summary factors and lessons. Washington DC: Public-Private
Infrastructure Advisory Facility (PPIAF).
Hub height
Elsabee, F 1973. Static stiffness coefficients for circular
Year Parameter Unit
80 100 120 foundations embedded in an elastic medium. MSc in
Civil Engineering, Boston, MA, US: Massachusetts
Annual average wind speed (m/s) 8.71 8.97 9.19
Institute of Technology.
Wind speed increase (%) 0.00 2.91 5.42 EN (European Standard) 1992 (2004). EN 1992-1-1:
2011
Energy generated (MWh) 13 136 13 593 13 967 Eurocode 2: Design of Concrete Structures. Part 1-1:
General Rules and Rules for Buildings. Brussels,
Revenue generated (Rm) 9.72 10.10 10.34
Belgium: European Committee for Standardization
Annual average wind speed v (m/s) 8.27 8.53 8.74 (CEN).
EN (European Standard) 1997 (2004). EN 1997-1:
Wind speed increase % 0.00 3.18 5.77
2012 Eurocode 7: Geotechnical Design. Part 1: General
Energy generated MWh 13 199 13 666 14 025 Rules. Brussels, Belgium: European Committee for
Revenue generated (Rm) 9.76 10.11 10.38 Standardisation (CEN).
Feng, Y & Tavner, P 2010. Introduction to wind turbines:
Annual average wind speed v (m/s) 8.91 9.20 9.43
Reliability and availability. Durham, UK: Durham
Wind speed increase % 0.00 3.21 5.84 University.
2013 
Harte, R & Van Zijl, G P A G 2007. Structural stability
Energy generated MWh 13 930 14 422 14 801
of concrete wind turbines and solar chimney towers
Revenue generated (Rm) 10.31 10.67 10.95 exposed to dynamic wind action. Journal of Wind
Averaged 20-year revenue (Rm) 198.64 205.62 211.11 Engineering and Industrial Aerodynamics (JWEIA),
95 (9–11): 1079–1096.
Revenue increase % 0.00 3.52 6.28
Gazetas, G 1983. Analysis of machine foundation
vibrations: State of the art. Soil Dynamics and
against overturning. This was for the case requirements associated with the foundation Earthquake Engineering, 2(1): 32.
of the water table being at ground level. The of concrete and hybrid wind turbine towers IEC (International Electrotechnical Commission) 2005.
foundation design of the concrete towers was are lower than those of the steel towers for IEC 6400-1:2005. Wind turbines. Part 1: Design
governed by a combination of bearing capac- the given design assumptions. Consequently, requirements. Geneva: IEC.
ity and foundation weight required to stabilise and additionally, the material cost of the IRENA (International Renewable Energy Agency) 2012.
against overturning, also for the case of water studied steel towers and foundations in a Renewable energy technologies. Cost analysis series
table at ground-surface level. South African context are higher than their – Wind power. Abu Dhabi, United Arab Emirates:
The amount of reinforcing steel in the foun- concrete and hybrid counterparts, particu- IRENA.
dation was governed by the minimum reinforc- larly for hub heights in excess of 100 m. Manwell, J, McGowan, J & Rogers, A 2010. Wind
ing requirement in EN 1992-1-1, although this The increased revenue, due to increases in Energy Explained: Theory, Design and Application.
may not be the case when designing for IEC hub height from 80 m to 100 m and 120 m for Chichester, UK: Wiley.
wind classes II or I. None of the foundations are a Vestas V112-3MW turbine, was shown to be Nicholson, J C 2011. Design of wind turbine tower and
subject to punching shear failure, even without in the vicinity of 3.52% and 6.28% respectively, foundation systems: Optimization approach. MSc
punching shear reinforcement, although the with average capacity factor increases of the thesis, Iowa, IA, UK: University of Iowa.
concrete towers were just under the limit same magnitudes. It remains to be verified Nordex 2007. Added yields in non-coastal regions.
for requiring punching shear reinforcement. whether these additional revenues exceed the Hamburg: Nordex Wind Power SE. Available at:
Foundations for concrete towers taller than added total costs to realise the higher towers. http://www.nordex-online.com/en/products-
120 m will not be able to provide sufficient services/hybrid-towers.html [accessed on 18
shear resistance without shear reinforcing, September 2014].
while the steel and hybrid tower foundations REFERENCES SANS (South African National Standard) 2009. SANS
may still have adequate shear resistance. ASCE/AWEA 2011. Recommended practice for 10160-3:2009. Basis of Structural Design and Actions
The steel tower design was governed by compliance of large land-based wind turbine support for Buildings and Industrial Structures. Part 3: Wind
the natural frequency stiffness requirements structures. Washington DC: American Society of Actions. Pretoria: SABS Standards Division.
of the tower, primarily, to ensure that the Civil Engineers and the American Wind Energy SIMULIA 2010. Abaqus CAE Finite element analysis
natural frequency lies within the acceptable Association. software. Version 6.10-2. Vélizy-Villacoublay, France:
limits as determined by the choice of tur- Cotrell, J, Stehly, T, Johnson, J, Roberts, J O, Dassault Systems.
bine, but also to limit deflections. The design Parker, Z, Scott, G & Heimiller, D 2014. Analysis of Van Zyl, W S & Van Zijl, G P A G 2015. Dynamic
of both the concrete and hybrid towers was transportation and logistics challeges affecting the behaviour of normally reinforced concrete wind
dictated by the lack of tensile resistance of deployment of larger wind turbine towers: Summary turbine support structures. Journal of the South
the concrete. results. Denver, CO, US: National Renewable Energy African Institution of Civil Engineering, 57(4):
The tensile stresses in the sections Laboratory. 38–44.
(reduced by post-tensioning) thus governed Craig, R F 2004. Craig’s Soil Mechanics, 7th ed. New WASA (Wind Atlas for South Africa) 2014. Mean wind
the design of the concrete sections, although York: Spon Press. speed atlas for parts of South Africa, Digital image.
buckling may become an important design DNV/Risø 2002. Guidelines for Design of Wind Pretoria: SA Department of Energy (DoE) [accessed
factor when designing towers taller than Turbines. 2nd ed. Copenhagen, Denmark: Det 22 September 2014].
120 m. Norske Veritas (DNV) and Risø National Laboratory. World Steel Association 2012. Steel solutions in the
According to the results from this Eberhard, A, Kolker, J & Leigland, J 2014. South Africa’s green economy: Wind turbines, Brussels, Belgium:
project, it can be seen that the material Renewable Energy IPP Procurement Program: Success World Steel Association.

54 Journal of the South African Institution of Civil Engineering  •  Volume 57  Number 4  December 2015