High Steel Tubular Towers For Wind Turbines: (Histwin2)

High steel tubular towers
for wind turbines

(HISTWIN2)
Research and
Innovation EUR 27226 EN
EUROPEAN COMMISSION
Directorate-General for Research and Innovation
Directorate D — Key Enabling Technologies
Unit D.4 — Coal and Steel
E-mail: rtd-steel-coal@ec.europa.eu
RTD-PUBLICATIONS@ec.europa.eu
Contact: RFCS Publications
European Commission
B-1049 Brussels
European Commission
Research Fund for Coal and Steel

High steel tubular towers
for wind turbines
(HISTWIN2)
M. Veljkovic, C. Heistermann, O. Garzon, M. Limam, A.T. Tran, Marko Pavlovic

Luleå University of Technology
Universitetsområdet, Porsön, SE-97187 Luleå, Sweden
M. Feldmann, F. Möller, C. Richter

Rheinisch-Westfälische Technische Hochschule Aachen (RWTH)
Templergraben 55, DE-5062 Aachen, Germany
C. Baniotopoulos, S. Gerasimidis, I. Zygomalas

Aristotle University of Thessaloniki
University Campus, Administration Building, GR-54124 Thessaloniki, Greece
A. Matos Silva
Martifer Energia – Equipamentos para energia, SA
Zona Industiral de Oliveira de Frades, PT-3684-001 Oliveira de Frades, Portugal
L. Simões da Silva, C. Rebelo, P. Pinto, R. Matos, A. Moura, H. Gervásio

Faculdade de Ciências e Tecnologia da Universidade de Coimbra
Rua Silvio Lima, Universidade de Coimbra Polo II, PT-3030-790 Coimbra, Portugal
J. Siltanen
Rautaruukki Oyj
Suolakivenkatu1, FI-00810 Helsinki, Finland
Grant Agreement RFSR-CT-2010-00031

1 November 2010 to 31 October 2013
Final report
Directorate-General for Research and Innovation
2015 EUR 27226 EN

LEGAL NOTICE
Neither the European Commission nor any person acting on behalf of the Commission is
responsible for the use which might be made of the following information.
The views expressed in this publication are the sole responsibility of the authors and do not
necessarily reflect the views of the European Commission.
Europe Direct is a service to help you find answers

to your questions about the European Union
Freephone number (*):

00 800 6 7 8 9 10 11
(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.
More information on the European Union is available on the Internet (http://europa.eu).
Cataloguing data can be found at the end of this publication.
Luxembourg: Publications Office of the European Union, 2015
ISBN 978-92-79-48076-8
doi:10.2777/899646
© European Union, 2015

Reproduction is authorised provided the source is acknowledged.
Printed in Luxembourg
Printed on white chlorine-free paper

TABLE OF CONTENT
TABLE OF CONTENT.............................................................................................. 3
1 FINAL SUMMARY ............................................................................................ 5
2 SCIENTIFIC AND TECHNICAL DESCRIPTION OF THE RESULTS ............................ 13

2.1 Objectives of the project ..................................................................................... 13
2.2 Comparison of initially planned activities and work accomplished .............................. 13
2.3 Description of activities and discussion ................................................................. 14
2.3.1 Structural analysis of various tower types .......................................................... 14
2.3.1.1 Introduction ........................................................................................... 14
2.3.1.2 Tower loading and design methodology ...................................................... 15
2.3.1.3 Tower and foundation design and results.................................................... 17
2.3.1.4 Design results ........................................................................................ 18
2.3.2 LCA of various tower types .............................................................................. 21
2.3.2.1 Model for Life-Cycle Assessment ............................................................... 21
2.3.2.2 Analysis of environmental impacts for a life span of 20 years ........................ 22
2.3.2.3 Analysis of environmental impacts for a life span of 40 years ........................ 25
2.3.2.4 Comments and Conclusions ...................................................................... 28
2.3.3 Steel tubular towers: Alternatives for cross-sections ........................................... 28
2.3.3.1 Buckling resistance of plated structures vs. shell structures .......................... 29
2.3.3.2 Influence of longitudinal folds on the buckling stress .................................... 35
2.3.3.3 FEA of polygonal vs. circular tower segments .............................................. 38
2.3.3.4 Optimum spacing of bolts in single-lap connections of a hexagonal element

consisting of four folded steel sheets ......................................................................... 44
2.3.3.5 Bolt spacing and geometry of the overlapping parts on a simplified model ....... 45
2.3.3.6 Analysis of the “complete tower segment model” resistance under compression
and bending .......................................................................................................... 48
2.3.3.7 Segment Tests ....................................................................................... 51
2.3.4 Transversal friction connections ....................................................................... 58
2.3.4.1 Finger tests ........................................................................................... 58
2.3.4.2 Derivation of modified buckling curve ........................................................ 68
2.3.4.3 FE analysis of transversal friction connections for circular and polygonal wind
turbine towers ....................................................................................................... 70
2.3.5 Longitudinal bolted connections ....................................................................... 89
2.3.6 Design Recommendations for modularized steel tubular towers............................. 90
2.3.7 Feasibility study ............................................................................................. 92
2.3.7.1 Preparation Phase ................................................................................... 93
2.3.7.2 Feasibility Tests ...................................................................................... 94
2.3.7.3 Manufacturing Process ............................................................................. 96
2.3.7.4 Conclusions ......................................................................................... 109
3
2.3.8 Foundations ................................................................................................ 110
2.3.8.1 Introduction ......................................................................................... 110
2.3.8.2 Alternatives for tower foundations ........................................................... 111
2.3.8.3 Shallow foundations .............................................................................. 112
2.3.8.4 Foundations with steel micro-piles ........................................................... 114
2.3.8.5 Micropiles experimental tests.................................................................. 117
2.3.8.6 LCA analysis (shallow vs. hybrid foundations) ........................................... 131
2.3.9 Design recommendations for hybrid foundations (shallow reinforced with micropiles)
133
2.3.10 Accessories in steel towers ........................................................................ 135
2.4 Conclusions .................................................................................................... 136
2.5 Exploitation and impact of the research results .................................................... 136
2.5.1 Actual applictations ...................................................................................... 136
2.5.2 Technical and economic potential for the use of the results ................................ 136
2.5.3 Any possible patent filing .............................................................................. 137
2.5.4 Publications / conference presentations resulting from the project ....................... 137
2.5.5 Any other aspects concerning the dissemination of results ................................. 139
LIST OF FIGURES ............................................................................................. 140
LIST OF TABLES ............................................................................................... 144
LIST OF REFERENCES ....................................................................................... 147
4
1 Final summary
Alternative structural solutions for high towers –Design of case studies and LCA
The analysis of alternative competitive construction solutions for tubular wind towers with circular
cross section where the hub height is higher than 80m was one of the main objectives of the study.
Therefore, the comparison between structural solutions based on steel and hybrid concrete-steel
tubular towers has been performed and, in addition to this original scope the study, the design of
concrete towers was performed for the sake of comparison. The friction connection between tower
segments, initially developed in previous HISTWIN project, was included in the analysis of
comparison with the performance of traditional flange connections. Also, the implementation of a
new structural concept using modularized tube tower, where the tube is made of segments
connected by longitudinal bolted connection, was taken into account for both steel towers and steel
part of hybrid towers.
The results of this investigation are compiled in this report, which contains the global structural
analysis of 15 study cases comprising steel, concrete and hybrid steel-concrete wind towers with
heights of 80, 100 and 150 m supporting multi-megawatt turbines of 2, 3,6 and 5 MW power
respectively. The design of the towers and respective shallow foundations was made in accordance
with the structural Eurocodes assuming standard steel and concrete grades and some minor
simplifications, e.g. the linear variation of thickness and diameter of the steel tube along the height
and simplified, although IEC61400 compatible, wind loading. For steel towers and steel segments
of hybrid towers the two types of connections between segments were considered, welded flange
connection (WFC) and the bolted friction connection (FrC) (see Fig. 5.1). Additionaly, lifetime of 40
years was considered in case of using FrC in order to allow for the possibility of elongation of the
current 20 years lifetime. The compilation of the study cases is as follows:
Tower Height and Rated Power of Wind turbine

80m/2MW 100m/3.6MW 150m/5MW
Lifetime (years) 20 40 20 40 20 40
Steel towers and steel WFC x x x
segments in hybrid
towers FrC x X x x x x
Concrete towers and
concrete segments of CT x X x x x x
hybrid towers
Main conclusions found were as follow. For towers up to 80 meters, the use of steel tubular
sections and flange connections are the most suitable. This conclusion supports the widespread use
of this type of connection in current tower heights. However, friction connections have shown to be
more suitable for higher towers, leading to lower material consumption and allowing for
modularization of large diameters of the tube. When seismic risk is considered in the analysis the
use of concrete towers is penalized, particularly for heights above 100 meters, and the foundations
of concrete and hybrid towers increase largely in order to withstand seismic forces.
The design results of these 15 study cases ere sufficient for a detailed Life Cycle Analysis (LCA) of
the different solutions. Main assumption for the LCA consists on the existence of two different
scenarios concerning the designed lifetime length of the towers. The first scenario considers the
lifetime of 20 years and two different construction methods for the in-situ assembling connection of
the steel segments. The first assembling concept is based on current technology using flange
connections and the second concept using friction connections.
The life cycle analysis was carried out according to international standards ISO 14040 [4] and ISO
14044 [5] and to the CML methodology [6].
Assuming equal importance for all environmental categories and the current lifetime of 20 years, it
was concluded that for heights up to 100 m hybrid towers with friction connections in the steel
tubular segments are the most efficient solution. For towers of higher height (150 m) the concrete
tower becomes more efficient (see Fig 5-3). The second scenario considers an increased total
5
lifetime up to 40 years, assuming the reuse of the tower after 20 years of operation. In this case,
the use of friction connections in steel towers enhances the possibility of dismantling and reusing
the tower potentiating much better performance in relation to the environmental category of global
warming (see Fig 5-5).
Alternative structural solutions for high towers – Polygonal cross sections and friction
connections
Another main objective of the work dealing with alternative structural solutions for tubular steel
wind tower has been the investigation related to the efficiency of using polygonal cross sections
together with friction connections, also including the possible use of high strength steel (HSS) and
the presence of openings.
Preliminary numerical investigations were performed in order to compare the buckling resistance of
plated structures vs. shell structures. The parametric studies based on ‘perfect’ tubular structures
with varying number of ‘edges’have shown that in general the use of polygonal cross sections
allows for a material reduction compared to circular cross sections due to better buckling
resistance. However, this preliminary conclusion is not robust and depends on the inclusion of
connection in the analysis. Therefoe, the numerical studies were extended with the consideration in
the FEM of imperfections in the shell, geometry of the connection and the possible use of high
strength steel. Behaviour at failure of the friction connections in circular and polygonal wind tower
structure was assessed through the bending resistance influenced by buckling in the vicinity of the
connection and fatigue performance of the bolts. Quasi-static analysis was made with
ABAQUS/Explicit solver coupled with damage material models. Dimensions of the tower segment,
as well as the design and fatigue loads are taken from a real design of a tower with classical ring
flange connection, provided by Vestas. Diameter of the shell of D = 3374 mm with thickness of t =
24 mm (steel grade S460) and high-strength bolts M48 (grade 10.9) were used. Buckling of the
shell segment in the vicinity of the friction connection is analysed in a parametric study considering
various design imperfection shapes and sizes according to EN1993-1-6 [20]. In order to provide
validation of the FE modelling and analysis techniques used for the real tower with friction
connections, experimental results from previous HISTWIN project [1] were used where friction and
ring flange connections with bolts M20 and shell diameter of 1000 mm (8 mm thickness) were
tested.
Bending resistance of the friction connection is analysed in a parametric study, considering:

different shapes of tower shells (circular and polygonal), assembling tolerances (gap between the
shells), length of fingers and few different steel grades (S355, s460 and S650MC) for the shell. The
assembling tolerances, mostly relevant in the case of friction connection, define the necessary gap
between the outer and inner shell. This leads to pre-bending of the fingers of the inner shell to
achieve contact between shells and developing friction forces. Two alternative fabrication
techniques for providing assembling tolerances were evaluated.
The study included also sensitivity analysis of both connections to imperfections of the shell in the
vicinity of the connection, comparison analysis of fatigue performance of the bolts in the ring flange
and in friction connection and analysis of the influence of buckling of the fingers on the bending
resistance of the connection.
Main conclusions of this advanced FEA are summarized as follows:
- If the lengths of the segments of polygonal tower shape are chosen such as to satisfy the
slenderness limit of the class 4 cross section the full effectiveness is achieved and no local
buckling occurs.
- Slots at the corners of polygonal tower shape are needed to provide assembling tolerances.
Those corner slots reduce the bending resistance, approximately 7% in the case analysed
here.
- Approximately 12% higher bending resistance is achieved with circular tower shape when
compared to the polygonal tower shape of the same outer diameter if no imperfections are
considered. However, the design imperfections in the case of polygonal tower are lower
and therefore the same bending resistance as in circular tower is achieved.
6
- If the design imperfections are considered the use of friction connection leads to 5% higher
bending resistance than the comparable ring flange connection. Additionally, the
eccentricity of the friction connection does not disturb the buckling resistance of the shell in
the vicinity of the connection if the design imperfections are included.
- In the friction connection there is no need to use slotted holes that are significantly longer
than the group of bolts in the connection. There are no stress concentrations in fingers,
since the initial tolerance gap between the shells is closed by the deformations of the
shells. On the other hand, if longer sloted holes are used bending resistance is reduced. For
the case of the longest fingers analysed in this study (total length 650 mm) the ultimate
bending resistance is reduced by 9-12%, due to bending of the fingers, depending on the
steel grade of the shell (S355, S460 or S650MC).
- Buckling length of the fingers together with part of the shell is estimated to 66% of the free
length of the finger (between the bolts and root of the slot). Reduction of the bending
resistance of the connection due to buckling of the fingers can be calculated as the buckling
reduction factor for uniform member in axial compression according to EC3 using the
proposed buckling length. The design check is proposed for the free length of the fingers.
- Assembling tolerances can alternatively be provided by pre-bending of the fingers instead of
providing gap between the shells. With the assembling tolerance of 10 mm, analysed here,
approximately 5% higher ultimate bending resistance is achieved.
- Slots at the corners of polygonal tower shape additionally induce stress concentrations up to
90% of yield strength of the material at the roots of the slots after the stage of preloading
of the bolts. These stress concentrations need to be considered with regards the fatigue
performance in more details for such tower shapes in order to justify their application in
wind towers.
- Fatigue performance of the bolts in friction connection is approximately 4 times better when
compared to the bolts in ring flange connection. This is the addressed to the absence of the
prying tension effects in bolts in the case of friction connection.
Friction connections – ‘Finger’ tests and design guidance
The transverse connection of two segments of a wind turbine using prestressed friction connections
comprises fingers and slotted holes (Fig 5.50). The inner segment has to be produced with a
smaller diameter in order to facilitate the assembling and thus the fingers have to be bent during
the prestressing. The bending of the fingers can cause either elastic or plastic deformations
depending on the fingers length/thickness and the size of the assembling gap. In a further (global)
bending of the wind turbine, the fingers can buckle under compression loads.
Small scale buckling tests on prestressed-bolted transversal connections with fingers and long
slotted holes of wind turbines have been performed (see fig. 5.52). To evaluate the tests, they
have been simulated using the finite element programme Abaqus. The numeric model was used to
perform a parameter study and to widen the range of parameters under investigation: the finger
width (50 and 75 mm) and thickness (10 and 15 mm); the gap size (10 and 15 mm); the steel
grade (S 355 and S 460); the influence of plastic deformation during closing of the gap (controlled
by the length); the number of fingers; the finger length as a dependent parameter, i.e. the finger
length was calculated so that the elastic or plastic moment was reached at the end of the fingers
during closing the gap (Fig. 5.53). This causes for example a shorter finger length for fingers with
the same width and thickness when an S 460 is used instead of an S 355. If a finger is designed
such that the elastic moment is reached after closing the gap, the length of the finger is called
“lElastic”. If the plastic moment is reached the length is called “lPlastic”. From the dependency of the
finger length of the other parameters follows, that not only single parameters can be varied but
always a set of parameters. The tests were performed with two steel grades, S 355 and S 460. It
was assumed that the number of fingers and the finger width have a linear influence on the
bearing. For that reason the width and the number of fingers were only be varied for the steel
grade of S 355. For each test series 2 tests were carried out. Tests were performed on 48
specimens.
The tests were evaluated using the design approach of EC 3-1-1 for the design against buckling.
The aim was to derive imperfection factors and to establish a buckling curve for a safe design of
7
the fingers. As the bearing behaviour of the fingers changes depending on the state of
deformation, the tests were divided into three classes in order to derive buckling curve for fingers
in different regions of the slenderness: (i) fingers with elastic deformation only during closing the
gap (l>lElastic), (ii) fingers with full plastification during closing the gap (l<lPlastic), (iii) fingers with
partially plastic deformation during closing the gap (lPlastic < l < lElastic)
The buckling curve is depicted in Fig 5-63 to illustrate the range of slenderness under investigation.
The 48 tests were performed on fingers with slenderness’s between 0.75 and 1.3. The tests were
complemented by finite element simulation to facilitate a quick and safe design in the range of
application.
Optimization of door opening reinforcements in tubular Wind Towers
To deal with the localized effect of the door opening in tubular towers the shell thickness is
increased or stiffeners are added aroud the opening. The aim of the numerical study performed
was to investigate effectiveness of different approaches. The parameters taken into account in this
study correspond to real designs situation of steel wind turbine towers. Concerning the analysis
methodologies, firstly, linear perturbation analysis of models have been performed in order to get
elastic buckling modes and eigenvalues in ’perfect’ geometry. Secondarily, nonlinear analysis of
models with imperfection geometry from buckling mode has been considered. Comparative studies
based on varying parameters and steel grade were also carried out. Results from comparative
studies in cases of door opening under compression and tension showed that the position of door
opening did not affect significantly the ultimate strength of tower (1% of difference) in the same
conditions. Parametric study aiming at optimal designs of door opening and stiffener showed that
37 mm of stiffener and 51 mm of thickness around door opening are sufficient to obtain the same
resistance in comparison with the model without door opening and 37 mm of wall thickness. When
steel grades higher than S355 are used the thickness of door opening and stiffener decreased
significantly. Values of 22% and 38% were found when steel grades S500 and S650 were used for
stiffener respectively.
Longitudinal bolted connections – manufacturing of prototypes and feasibility

assembling
A next step towards higher hub heights and greater diameters of steel tubular towers is the
modularization in vertical direction (see Fig. 5.96). For towers with such a segmented cross
section, 4 plates overlap in the junction of longitudinal and transversal joint. This can be avoided
by cutting the edges of the two inner plates, so that the two plates are located in the same surface
(see Fig. 5.97).
In order to have evidence of the feasibility of such manufacturing approach two downscale
prototypes were built and assembled. Each prototype is composed by two main cylindrical
segments assembled with transversal friction connection. Each of these segments is obtained by
the overlapping of four curved plates connected by bolts in pure shear connections along the
longitudinal direction.
During manufacturing the rolling process (cold forming) plays the most important role due to the
dimensional and geometrical precision required. After the rolling of the segments a very good
result was observed, that is the “fingers” curvature was similar to the final curvature of the plate.
Also the assembly of the parts was achieved without major difficulties showing that the
manufacturing procedure is reliable and could be extrapolated to real size prototypes.
Though the holes had been cut using the plasma technique, they presented an acceptable quality
for the assembly of the prototypes but in this case it was used a plasma arc current of 200A. The
final results might have been improved by using an intensity of 260A. As there was not been made
a pre-hole for the starting of the plasma cutting, dirt around the holes was accumulated which had
to be removed after the cutting process. The use of the pre-holes may decrease the cleaning time
considerably which may improve the required time for the manufacture of such elements since the
cleaning time is twice of the cutting time.
8
The choice of the proper rolling pressure is a key factor for the rolling. Excessive pressures may
lead to the appearance of waviness on the plates close to the center of the roll. With low pressures
the waviness effect is more pronounced close to the edges of the plates (along the width). In this
case it can be concluded that the chosen pressure was correct as no waviness was detected in
none of the plates.
It was observed some inconstant gaps between the upper and the lower prototype segments
related with the process of rolling and/or assembly. The difference of gaps between both segments
may be related with a pre-tightening that occurred prior the measurement in order to fix all the
pieces together and to allow the full assembly which may lead to some plastic deformation of the
elements.
It was also observed some misalignments (around 1.5 mm) between the holes of two different
levels but with the tightening of the first bolts that misalignment tended to disappear.
Optimization of foundation – Tests on Micropiles
The procedures and the results of experimental tests carried out on reduced scale micropiles are
presented in order to understand their behaviour and to calibrate numerical models that allow the
extrapolation of the results to a real scale. A total of 36 reduced scale experimental tests were
carried out in order to obtain the proper resistances and stiffness to use in numerical models of the
whole hybrid foundations. The calibration of the numerical model/procedure will allow an
extrapolation for other models with longer micropiles and different (better) ground properties.
Single and groups of micropiles were tested in a container filled with sand under monotonic or
cyclic+monotonic loading. They were tested without and with pressure grout in order to access
improvements introduced by the grouting.
Main conclusions of this experimental program are:
- The grouted specimens show higher resistances than the corresponding ungrouted cases.
This improvement is more emphasized in compression than in tension tests and is due to
the increase of grout-to-ground bond strength, the micropile diameter and the
improvement of the soil characteristics through the grouting pressure.
- In the large majority of the tests the compression resistances are higher than the
correspondent tension resistance due to the influence of the tip effect. Both for single and
group tests, the differences are higher for grouted than for ungrouted specimens.
- The spacing between piles in a group does not affect significantly the results
- The values obtained for the mean unit skin friction on single micropiles, both grouted and
ungrouted, fall quite well within the analytical estimations performed ahead of the tests.
The values for the unit skin friction for the single micropiles when compared with the group
specimens were, in average, higher 132% for ungrouted tests and 29% for grouted tests.
The looser sand in the middle of the group specimens may have conducted to lower friction
angles.
- The obtained static monotonic stiffness (or post-cyclic when applicable) is, in average, 490%
higher for the grouted than for the ungrouted specimens. The improvement of the grout is
more relevant on the single micropiles where an average improvement of 630% was
obtained against an improvement of 200% for the group cases.
- In the case of the cyclic stiffness it was concluded that, for every case studied, the grouted
specimens provided higher stiffness than the corresponding ungrouted, both for
compression and tension loadings.
- On the group tests the compressive stiffness is significantly higher both for ungrouted and
grouted tests.
- As the relevance of the grouting in both strength and stiffness of the micropiles was shown,
it is therefore very important to achieve the best grouting procedure in order to optimize
use of the micropile and promote the highest ground-to-grout bond strength.
Optimization of foundation –Design recommendations for hybrid foundations
The design of hybrid foundations is very similar to the design of shallow foundations after removing
the portion of the loads resisted by the micropiles which can be evaluated using a numerical model
9
that reproduces the foundation supported by springs. These are used to simulate both the soil
reaction and the effect of the micropiles, through a soil-foundation interaction analysis. The
experimental tests carried out provide valuable information for the determination of the spring
parameters as it will be possible to extrapolate the obtained results for a full scale element and so
reproduce accurately the micropile behaviour.
The tests performed show that the skin friction on ungrouted micropiles in cohesionlles soil may be
determined using the following expression:
q s  K   'v tg ( )
where K is the horizontal stress coefficient, based on the at-rest coefficient K0 and will typically
vary from 0.8 K0 to 1.5K0 for bored and driven ungrounted micropiles, ’v is the effective vertical
stress and  is the micropile/soil friction angle, which depends on the material. On the tests
performed K=0.44 and =20º.
For the grouted micropiles, the tests showed an increase in the side friction and it is possible to
conclude that in a full size prototype the ultimate grout-to-ground resistance depends on the type
of micropile and ground conditions. The proposals of Bustamante and Doix [32] and FHWA [30] are
adequate but should be complemented with control load tests.
Regarding the micropile stiffness, the tests showed that the controlling parameters are the soil’s
deformability modulus E (or the shear modulus, G) and the grouting process. For the numerial
analysis, hyperbolic models such as the proposed by Randolph and Wroth and modified by McVay
et al. [34] are applicable and may be used to define a simpler bi-linear curve:
 0 r0   rm     rm  r0  
z  ln   
Gi   r0    rm   r0    
r0 0 R f
with  and where z is the vertical displacement, Gi is the initial shear modulus, 0 is the
f
mobilized skin-friction, f is the ultimate skin-friction, r0 is the micropile radius, Rf is the failure ratio
from the hyperbolic model and rm is the integration limit.
For the number of cycles performed, there was not a substantial loss of stiffness, but further
research is required in order to extrapolate for full scale prototypes.
The axial force obtained in each micropile must be lower than the minimum value between the
internal (structural) and the external (geotechnical) resistances of the elements. For EC7 Design
Approach 1, the external resistance must be determined both for EQU/type 1 and type 2
combinations. If the seismic loading is taken into account the external resistance must also be
determined for this load case.
The remaining portion of the loads must be supported only by the shallow foundation and the
bearing capacity under the foundation must be verified for this situation.
The Limit state of equilibrium according to EC7-1 [45] is fulfilled if the axial resistance in the
micropiles never surpass the correspondent resistance (internal and external).
Concerning the micropile spacing a distance of 2 to 3 times of the micropiles diameter should be
guaranteed in order to avoid group effect phenomena and so obtain higher individual resistances.
The base sliding under the foundation should be considered to be supported only by the friction
between the shallow portion and the soil.
Optimization of foundation – Design and LCA of hybrid foundations case studies
A detailed description of the geotechnical design of both shallow and hybrid (shallow reinforced
with micropiles) foundations is presented in the report as well as the design results of some specific
examples. A LCA analysis was carried out in order to understand the impact of each solution and to
understand the benefits of the use of micropiles.
10
A sub-set of the study cases defined to assess the LCA of towers was considered. Three different
tubular wind towers with heights of 80, 100 and 150 meters supporting turbines of 2, 3.6 and 5
MW respectively were considered for the design of the foundations. The tower in each of the three
cases is considered to be built using concrete, steel and hybrid steel-concrete tubular shell. A
feasibility assessment about the use of micropiles as a reinforcement of the actual shallow
foundations of the wind towers was carried out. The increase in the height of the towers will
consequently lead to an increase in the foundations diameter; hence; an improvement in the
foundation system was proposed.
The results of the shallow and the hybrid foundations designs showed that the micropiles
reinforcement is a very satisfactory solution both in terms of material consumption and in terms of
potential environmental impacts, according to the presented LCA.
In terms of concrete consumption, it was observed that in the case of the hybrid foundations, a
reduction between 15% and 54% was achieved when compared with the correspondent shallow
foundations. The reduction in terms of steel reinforcement is between 50% and 69% (disregarding
the micropile tube as rebar) and 30% to 53% (considering the micropile tube as rebar). It is noted
that in the case of the hybrid foundations, the materials related to the micropiles installation
(micropiles and grout) were also considered in the analysis.
The results provided by the LCA pointed for a lower environmental impact in the hybrid foundations
in comparison with the correspondent shallow foundation, for every structure geometry and
typology and for every environmental indicator considered in the analysis.
A reduction of 10% to 30% of the global warming potential was achieved by the hybrid foundations
in relation to shallow foundations, while for the primary energy demand a reduction of 9% to 25%
was obtained. The main reason for the better environmental performance is due to the reduction of
the mass achieved by hybrid foundations, in spite of the use of additional equipment.
11
2 Scientific and technical description of the results
2.1 Objectives of the project

The main objective of this project is to ensure and raise the competitiveness of steel tubular
towers, which support wind trubines, on the global market. Special focus is brought to the ability to
compete against concrete and hybrid towers higher than 100 m not only in production and
assembly but also in maintenance and life cycle analysis.
Based on research results from the previous project “High-Strength Steel Tower for Wind Turbine
(HISTWIN)” from 2007-2009 [1], modular tower assembly is investigated. A closer look is given to
the friction connection, which joins two tower segments. The design and geometry details of the
long open slotted holes are studied.
Various alternatives for cross-sections are tested and modelled in finite elements. The influence of
the number of folds in a polygonal cross-section on the stability of the tower is checked, as well as
the effect of various stiffening methods of door openings in the tower shell.
Also the foundations of such towers are looked into: A new type of hybrid slab foundation anchored
with steel micropiles will be developed based on laboratory tests and finite element analysis.
The objective of WP 1 “Alternative structural solutions for high towers” is to analyse different
structural solutions of steel tubular towers: steel tower and hybrid tower. For the sake of
completeness this original scope is extended by adding the concrete towers and rather realistic
predesign considering five most relevant design criteria. For steel alternatives traditional flange
connection is compared to alternative using the friction connection. Furthermore, the possibility of
a modularisation of the tower will be analysed in considering polygonal and circular cross-section.
Influence of the cross-section on resistance of the towers is investigated.
WP2 “Foundation of the hybrid tower vs. steel tower” has the scope of studying a new proposal of
geometry for the foundations of existing and higher wind towers. The new proposed geometry is
based on a reinforcement of the actual direct foundations with micropiles that could increase the
resistance not only in compression but also in tension. The input data for design of this novel
foundation are evaluated from experiments.
The critical details of the new structure will be examined in WP 3 “Down-scale testing of the tower
segments” with static tests on segments and their connections. Preliminary design of the
equipment for testing of connection details (“fingers”) will back up design formulae, which will be
provided in WP 5 “Design guidance”. The second objective of WP3 is to investigate alternatives for
door opening design. Experiments will be used for benchmarking of FEA. The benchmarking is done
priori testing with the nominal material data. After testing of coupon test taken from the flat parts,
circular parts and along the folds FEA will be updated, if necessary.
A feasibility study of the “short segments” of the tower, approximately 2m long and 3,5m in
diameter, will be done in Martifer workshop hall, within WP4. Specimens are designed and material
ordered. These tests will fill in a gap in knowledge of pre-stress forces in bolts due to execution
tolerances.
The accessories inside the tower are reviewed in order to check possible benefits of production
processes.
2.2 Comparison of initially planned activities and work accomplished

The major change compared to the initially planned work is the substitution of Task 3.1, Static
bending tests of circular/polygonal down-scaled tower segment by small scale “finger” tests. The
new type of tests, together with the numerical analysis, have filled the knowledge gap, as
identified in the proposal, and have covered wider range of the friction connection configurations.
Minor modification of the Task 3.2 is that no longitudinal bolted connection was tested which is a
logical consequence of changes approved in the Task 3.1. Focus is on detail material investigation
of the folds’ influence on resistance of the polygonal towers.
Tasks 1.1 and 1.2 were modified so the friction connections are thoroughly examined instead of
investigating improvements of the hybrid (concrete 2/3 and steel 1/3) towers.
13
In Task 4.1 and Task 4.2, more realistic plate thicknesses were considered in feasibility tests on
circular cross section. The circular cross section is chosen because it is more complex to assemble
segments of circular sections and has more practical relevance.
2.3 Description of activities and discussion
2.3.1 Structural analysis of various tower types
2.3.1.1 Introduction
This chapter contains a global structural analysis of steel, concrete and hybrid steel-concrete wind
towers with heights of 80, 100 and 150 m supporting multi-megawatt turbines of 2, 3,6 and 5 MW
power respectively.
In the original technical annex a comparison of steel tubular and hybrid towers was planned, but
during a preliminary investigation it was concluded that an additional comparison to concrete
towers would be also useful.
The design of the towers was made in accordance with the structural Eurocodes considering a
linear variation of thickness and diameter of the towers along the height.
For steel towers and steel segments of hybrid towers two types of connections between segments
are considered: i) the current welded flange connection (WFC) shown in Figure 2.1a) and ii) the
bolted friction connection (FrC) shown in Figure 2.1b).
Welded flange connection (WFC) Friction connection (FrC)

Figure 2.1: Type of connections used in steel tubular towers
The use of WFC assumes that the tubes are built by rolling steel plates and welding them together
with the flanges on the extremities to form the tubular segments. This is a current solution which
can be performed in factory in order to deliver tubular segments to the construction site, where
they are assembled by bolting flanges together to form the tower. The limitation for this procedure
is the diameter that should not exceed 4.5m, considered as limit for road transportation of pre-
fabricated tubes. Therefore, for larger diameters welding on site must be considered when using
this solution.
The alternative solution is the use of bolted friction connection (FrC) to assemble prefabricated
tubes and/or to assemble curved panels on site to form circular cross-sections. This avoids welding
on site when diameters of tubular section are larger than 4.5m. This alternative connection has
been developed in the previous HISTWIN-project [1] and has been recognized as a competitive
solution for steel towers (Engström et al., 2010). This type of connection allows the deconstruction
of the tower by dismantling the steel segments after the service life of 20 years.
Scenarios considered in Table 2-1 include the state-of-art situation of 20 years life time of the wind
towers whatever the type of construction and material. Besides that, the exercise of considering
doubling the lifetime of the tower by reusing the tower with new turbine and possible
14
transportation to another spot is justified by: i) enough reserve of fatigue resistance of the tower
after 20 years can be achieved with some changes in design; ii) improvements in turbine
technology and their dynamic control make them more and more ‘tower friendly’ concerning
fatigue requirements.
Table 2-1 summarizes the case studies considered in the design, taking into account two lifetimes
and two type of connection of steel segments. The designation used hereafter for each case study
includes a prefix (WFC, FrC, CT) for the type of tower and connection and a suffix (20, 40)
representing the lifetime – e.g. FrC40 is a steel or hybrid tower whose steel parts are connected on
site using friction connections and is designed for 40 years lifetime.
Table 2-1– Scenarios for tower design
Tower Height and Rated Power of Wind turbine

80m/2MW 100m/3.6MW 150m/5MW
Lifetime (years) 20 40 20 40 20 40
Steel towers and steel WFC x x x
segments in hybrid
towers FrC x x x x x x
Concrete towers and
concrete segments of CT x x x x x x
hybrid towers
2.3.1.2 Tower loading and design methodology

Typical loading can be obtained from the bibliography [2] which is considered sufficient for the
sake of performance comparison between tower types.
Five main design situations, corresponding to different load combinations, were considered:
 Extreme wind load in non-operating condition (EW); this combination includes loads on top
of tower and wind load distributed along the tower height; in this condition turbine is in
parked position;
 Extreme operating condition (EO); this combination includes loads on top of tower and wind
load distributed along the tower height;
 Earthquake loading condition; the response spectrum was defined according to Eurocode 8
for a region complying with the seismic risk defined by the peak ground acceleration of
0.255 g, behaviour factor was considered as q=1; the seismic combination was obtained
through superposition of the effect of this seismic load with 30% of the extreme wind load;
 Fatigue condition using Damage Equivalent Loads (DEL) adapted from LaNier et al [2];
 Dynamic characteristics condition that avoids resonance between rotor and natural frequency
of the tower during operation. The dynamic interaction of the soil, foundation and
superstructure need to be considered.
The wind and seismic section loads are presented in Table 2-2 and Table 2-3 respectively for
extreme situations relevant for ultimate limit state design.
15
Table 2-2: Wind section forces.
2.0MW 80m 3.6MW 100m 5.0MW 150m

Wind Load
EW EO EW EO EW EO
FTop (kN) 628 599 1087 1198 578 1065
FzTop (kN) -1093 -1080 -3155 -3129 -5000 -4879
MTop (kN∙m) 1978 1994 16767 9913 28568 14987
MzTop (kN∙m) 1213 1021 5961 1597 5834 3966
FBase (kN) 871 647 1728 1358 1545 1279
FzBase (kN) - Steel -2366 -2559 -6882 -6856 -14733 -14614
FzBase (kN) - Concrete -9197 -9184 -22594 -22568 -47332 -47213
FzBase (kN) - Hybrid -7503 -7489 -16196 -16170 -38020 -37901
MBase (kN∙m) 60149 50221 143961 108656 258446 238810
MzBase (kN∙m) 1218 1010 5961 1905 5834 3966
Table 2-3: Earthquake load, not combined.
2.0MW 80m 3.6MW 100m 5.0MW 150m

Earthquake Load
Steel Conc. Hybrid Steel Conc. Hybrid Steel Conc. Hybrid
Period (s) 2.56 1.59 1.77 2.83 1.87 2.13 4.00 2.84 2.92
S (m/s2) 1.55 3.19 2.86 1.26 2.71 2.24 0.63 1.25 1.18
FTop (kN) 256 768 820 605 1887 1717 539 1472 1638
FMid. (kN) 353 2120 1130 780 4456 2214 795 4130 2418
MMid. (kN∙m) 12093 56475 38682 34397 154799 97638 49707 203410 151197
FBase (kN) 402 2986 2185 886 6238 3692 951 6054 4592
MBase (kN∙m) 27465 163104 111335 76738 433467 25592 116729 603335 435016
Wind towers are prone to fatigue damage and must be checked for fatigue limit state. The load
combination associated with this condition is based on S-N curves describing the number of cycles
N for each load reversal amplitude S. These curves are built on a yearly basis considering all
possible load situations that may occur during operation and extrapolated for the lifetime of the
turbine, usually 20 years. Extrapolation for 40 years lifetime is straightforward.
The procedure used here consists in the approximation given by the Damage Equivalent Load
(DEL) which is related to the damage equivalent factor defined in eurocodes. This load induces in
the structure the same damage as the S-N load spectrum would induce and is obtained from the
equation
1
 n N m
DEL    Rangetm t 
 t 1 N ref 
where Rangei refers to the value of the fatigue spectrum defined for a certain load effect and the
parameters used in the calculation are m=4 and NRef = 2x108 cycles. It is worth to note that DEL
values obtained for a certain NRef can be converted in DEL values for a different NRef , e.g. the
reference number of cycles NRef = 2x106 used in Eurocode 3. Table 2-4 summarizes the Damage
Equivalent Loads for fatigue design considering 20 years lifetime.
16
Table 2-4: Damage Equivalent Loads, m = 4, NRef = 2x108 cycles.
2.0MW 3.6MW 5.0MW

DEL
80m 100m 150m
Mx (kN∙m) 372 657 1096
Tower Top
My (kN∙m) 1435 3293 5484
Tower Mx (kN∙m) 5677 13781 33048
Bottom My (kN∙m) 7378 29984 69813
The frequency lower limit corresponds to 1P (rotor rotation) and the upper limit corresponds to 3P
(blade passing frequency) including tolerances of -15% and +10%.(see Table 2-5)
Table 2-5: Turbine mass, rotor speed and admissible frequency range.
Rotor
Mass(to speed(Rpm) Admissible frequency
Turbine
n) (-15 %) (+10 range Hz
%)
2,0 MW 108 18.7 0.343 – 0.795
3,6 MW 315 13.8 0.253 – 0.587
5,0 MW 500 12.1 0.222 – 0.514
2.3.1.3 Tower and foundation design and results

Strength and stability design checks (ULS) for the steel towers were made according to the “Stress
Design” methodology defined in the code for load combinations EW, EO and Earthquake. The
analysis is elastic and the manufacturing class considered is B (high). The boundary conditions of
the cylindrical segments are of type BC1 (clamped) when using flanges, i.e. at the tower top and
bottom and in intermediate WFC and of type BC2 (pinned) in intermediate FrC. The steel grade
used is S355.
Fatigue resistance of the tower using WFC is defined by the detail 71 corresponding to both the
welded flange for segment connection and to the longitudinal but weld used to build the tubular
segments. Fatigue resistance of the tower using FrC is calculated using the detail 90 corresponding
to one sided connection with preloaded high strength bolts. In the design checks, the tower
opening located in the lower segment and the transition segment between the tower and Nacelle
were not considered in detail. The loading condition used in the design check is DEL.
The design of the concrete segments is performed according to Eurocode 2 [11]. The design
criteria are: (i) the ultimate limit state (ULS) of resistance of the cross section for maximum loads
both from the wind (EO, EW) as the Earthquake; (ii) the avoidance of decompression in the
concrete cross-section during turbine operation using load condition (EO); (iii) the fatigue of
concrete and reinforcement steel bars and pre-stressing tendons using damage equivalent loads
(DEL) and (iv) the minimum reinforcement of concrete sections imposed by the code. The concrete
class used is C40/50 and the steel grade is S500.
Geotechnical design has been verified in accordance with the recommendations of Eurocode 7 [12].
The soil characteristics are presented on Table 2-6 and Table 2-7 presents the forces applied on
foundations.
17
Table 2-6: Soil characteristics.
´k c´k γ δb E
42 º 0 kPa 18 kN/m3 42 º 675MPa
Table 2-7: Forces applied at the center of the foundation base
2.0MW 3.6MW 5.0MW

Load
80m 100m 150m
FBase (kN) – Steel 871 1728 1515

FBase (kN) – Concrete 2802 6047 6256
FBase (kN) – Hybrid WFC20 1979 3678 4128
FzBase (kN) – Steel FrC20 13699 29934 44347
FzBase (kN) – Concrete 23635 56508 98073
FzBase (kN) – Hybrid WFC20 19025 42090 74987
MBase (kN) – Steel FrC20 63633 152598 266777
MBase (kN) – Concrete 178533 494790 803576
MBase (kN) – Hybrid WFC20 131191 324291 600570
All checks of the Limit States were drawn up assuming an equivalent circular foundation. In all
cases the design driving criterion for the foundation geometry was the equilibrium verification. The
concrete class used is C30/37 and steel grade A500. The amount of steel rebars was calculated
using a finite element model in Autodesk Robot [13].
Details for the design of shallow foundations can be found in chapter 2.3.8.3.
2.3.1.4 Design results

The governing load cases and design checks of the different solutions identified after performing all
design calculations are summarized in Table 2-8. The design results related to the main
geometrical parameters of the towers are summarized in Table 2-9 to Table 2-11.
18
Table 2-8: Governing load cases and design checks.
2.0 MW/80m 3.6 MW/100m 5.0 MW/150m

1
bottom Fatigue Fatigue1 Fatigue1
WFC20 ½ Height Fatigue Fatigue Fatigue
Steel top Fatigue1 ULS (EW) Fatigue1
towers bottom ULS (EO) Fatigue1 Fatigue1
FrC20
½ Height ULS (EO) Fatigue Fatigue
FrC40
top ULS (EO) ULS (EW) Fatigue1
bottom ULS (EQuake) ULS (EQuake) Decompression
Concrete CT20
½ Height Decompression Decompression Decompression
towers CT40
top Concrete fatigue Decompression Decompression
Concrete part bottom Decompression Decompression Decompression
CT20, CT40 ½ Height Concrete fatigue Concrete fatigue Concrete fatigue
Hybrid Steel part ½ Height Fatigue Fatigue Fatigue
towers WFC20 top Fatigue1 ULS (EW) Fatigue1
Steel part ½ Height ULS (EO) Fatigue Fatigue
FrC20, 40 top ULS (EO) ULS (EW) Fatigue1
Table 2-9: Design results for concrete towers.
2.0MW 3.6MW 5.0MW

80m 100m 150m
External diameter (mm)/ bottom 6100/400/38 8900/550/54 12000/620/56
Thickness (mm)/ ½ Height 4530/290/18 6400/400/28 8400/425/42
Number of Tendons top 2950/180/8 3900/250/10 4500/300/10
Deflection (EQuake) (mm) top 580 650 890
Volume of concrete, C40/50 (m3) 322.2 790.9 1778.9
Steel rebars (Ton) 2.3 5.7 12.9
Prestressing tendons (Ton) 21.6 39.6 74.2
Table 2-10: Design results for steel towers.
2.0MW 3.6MW 5.0MW

80m 100m 150m
WFC20 bottom 4300/23 6900/42 10100/47
top 2950/10 3900/17 4500/25
External diameter (mm)/ FrC20 bottom 4300/23 6100/35 9400/44
Thickness (mm) top 2950/10 3900/17 4500/20
FrC40 bottom 4300/23 6900/38 9600/47
top 2950/10 3900/16 4500/25
WFC20 top 1770 1340 1810
Deflection (Wind) (mm) FrC20 top 1770 2030 2390
FrC40 top 1770 1470 2030
WFC20 top 850 750 930
Deflection (EQuake) (mm) FrC20 top 856 907 1050
FrC40 top 856 766 979
WFC20 122.7 414.0 1025.0
Mass of steel shell, S355
FrC20 122.7 333.5 871.9
(Ton)
FrC40 122.7 384.7 987.6
WFC20 4.65/226 6.99/337 21.43/911
Flange/Bolts
FrC20 0/1845 0/2383 0/6280
(Ton/Number)
FrC40 0/1845 0/2452 0/6323
19
a) b)
Figure 2.2: Coordinate system used for the design of wind turbines;
a) Hybrid tower; b) Steel or concrete tower
Table 2-11: Design results for hybrid towers.
2.0MW 3.6MW 5.0MW

80m 100m 150m
Concrete part (CT20, CT40)
External diameter (mm)/ bottom 5700/400/32 8700/450/38 11800/600/60
Thickness (mm)/ ½
CT20 4700/370/32 7000/350/38 9200/400/60
Number of Tendons Height
&
Concrete C40/50 (m3) 233.3 488.3 1187.5
CT40
Steel rebars (Ton) 1.7 3.5 8.6
Prestressing tendons (Ton) 16.9 25.1 59.3
Steel part
½
3630/17 5350/27 7300/36
WFC20 Height
top 2950/10 3900/17 4500/25
½
External diameter (mm)/ 3630/17 5000/26 6950/32
FrC20 Height
Thickness (mm)
top 2950/10 3900/17 4500/20
½
3630/17 5400/27 7050/36
FrC40 Height
top 2950/10 3900/17 4500/25
WFC20 top 750 740 1000
Deflection (EQuake) (mm) FrC20 top 750 780 1050
FrC40 top 750 750 1000
WFC20 44.0 136.4 342.5
Mass of steel shell, S355 (Ton) FrC20 44.0 120.7 284.1
FrC40 44.0 129.6 334.8
WFC20 2.62/103 3.71/145 9.41/368
Flange/Bolts
FrC20 0/1012 0/1302 0/2090
(Ton/Number)
FrC40 0/1012 0/1325 0/2099
20
The dimensionsresulting from the foundation design are present in chapter 2.3.8.3.
2.3.2 LCA of various tower types

This section reports the comparative study concerning life cycle assessment of steel, concrete and
hybrid steel-concrete wind towers with heights of 80, 100 and 150 meters supporting multi-
megawatt turbines of 2, 3,6 and 5 MW power respectively.
2.3.2.1 Model for Life-Cycle Assessment

The Life-Cycle Assessment includes the production of materials, their recycling and the excavation
for the foundations.
A model for the LCA of wind towers was developed, which is illustrated in Figure 2.3, in order to
compare different alternative construction solutions for the tower supporting the wind turbine. The
Life-Cycle model includes the production of materials, their transportation to the construction site,
the excavation of soil for the foundations, the erection of the structure (fuel consumed by cranes),
the maintenance, the deconstruction of the towers and aftermost their recycling. The fuel
consumption of cranes is based on [3] and adjusted for different powers and heights.

Equipment
Production
Transport Rehabilitation
of
materials
Site
Construction
preparation
Recycling
Transport
Deconstruction
Landfill
Figure 2.3: Life cycle analysis of a wind tower.
Moreover, it was assumed that the support structure of the wind turbine generator does not need
any maintenance work with significant environmental impact during the service life of 20 years.
When considering the life span of 40 years, some rehabilitation work is considered as described
further down in the text.
During the deconstruction process, steel structures are disassembled in the reverse order they
were initially built and transported for recycling, in the first and second scenarios, or to the new
place where they are re-assembled, in the third scenario. For the concrete tower and concrete part
of the hybrid tower, the structure is demolished and sent for recycling. Therefore, in the third
scenario, new concrete elements are produced and sent to the new location for the assemblage of
each tower.
Each scenario is further detailed in the following sections.
The life cycle analysis was carried out according to international standards ISO 14040 [4] and ISO
14044 [5] and the CML methodology [6]. The environmental categories selected for the analysis
are indicated in Table 2-12.
21
Table 2-12: Environmental indicators for LCA
Indicator Unit
Abiotic Depletion (ADP fossil) MJ
Acidification Potential (AP) kg SO2-Equiv.
Eutrophication Potential (EP) kg Phosphate-Equiv.
Global Warming Potential (GWP) kg CO2-Equiv.
Ozone Layer Depletion Potential (ODP) kg R11-Equiv.
Photochemical Ozone Creation Potential (POCP) kg Ethene-Equiv.
2.3.2.2 Analysis of environmental impacts for a life span of 20 years

Comparison of the life cycle results for the three towers
In the first scenario, it is considered that in the end-of-life stage, in year 20, the structure of the
supporting tower is deconstructed and materials are sent to their final destination. The scope of the
analysis comprehends the production of materials, the transportation of materials to construction
site, the assemblage of the structure in the construction site, the deconstruction of the tower and
the transportation of the waste to a recycling plant or a landfill.
According to the procedure described in the previous section, all pieces of the structure are
transported to the construction site and assembled there by the use of a crane. The consumption of
diesel is taken into account in the calculation.
During the use stage, the structure is regularly maintained in cycles occurring two to three times
per year [7], which usually include bolt adjustments, painting and substitution of spare parts as
established by maintenance handbooks. However, most maintenance actions occur in the nacelle
and rotors, which are outside the scope of the analysis; therefore the use stage (for the period of
20 years) was considered to be negligible.
For the end-of-life stage it is considered that all materials are recycled with the following recycling
rates: steel structure with a recycling rate of 90%, steel reinforcement with a recycling rate of
70%, and concrete structure and foundations with a recycling rate of 70%. The remaining parts are
sent to a landfill of inert materials.
The allocation of recycling materials, considered in this paper, takes into that the use of secondary
material displaces the use of virgin material [5]. Therefore, for the recycling of steel products, the
close-loop approach provided by World Steel Association [8] is taken into account. According to
this approach, scrap recycling avoids primary production of steel and thus, credits are provided to
the system producing the recycling materials. It is noted that credits are calculated based on the
net scrap produced in the end of the life cycle, i.e., the difference between the scrap recovered in
the end of life and the initial scrap content. Likewise, in relation to concrete, it is considered that
concrete waste is crushed and valorised to replace aggregates, thus avoiding the need to produce
virgin material. However, before being valorised it is assumed that concrete waste has to be
treated.
The transportation distances from the production plant to the construction site and from the
demolition site to the recycling plants were considered to be 100 km for all materials. The
transportation distance from the demolition site to the landfill was considered to be 20 km.
The results of the LCA for this scenario are indicated in Figure 2.4 for the environmental category
of global warming. The analysis was performed by the software [9]. The bill of materials for each
tower was presented in previous towers design (Task 1.1 and Task 1.2).
The global warming potential aims to quantify the emission of greenhouse gases, such as CO2 and
CH4, to the atmosphere. Due to its major influence in climate change, the results for the life cycle
analysis, focusing on global warming, are illustrated in Figure 2.4 for the three types of towers
(steel, concrete and hybrid tower) and for the three heights (80 m, 100 m and 150 m).
Furthermore, for the steel towers and steel part of hybrid towers, two types of assemblage were
considered, taking into account the welded flange connection (WFC) and the bolted friction
connection (FrC). In all cases, earthquake design was taken into account for the quantification of
materials.
22
Hybrid tower FrC (150 m)
Hybrid tower WFC (150 m)
Concrete tower (150 m)
Steel tower FrC (150 m)
Steel tower WFC (150 m)
GWP [kg CO2 eq.]
0,00E+00 5,00E+05 1,00E+06 1,50E+06 2,00E+06
Figure 2.4: Life cycle analysis of wind towers (1st scenario).
Observing the global results it is worth noting that the increase of tower height from 80 to 150
meters more than triples the amount of CO2 emitted. This increase must obviously be balanced by
the increase in efficacy of the turbine, taking advantage of the higher wind speed, lower wind shear
and more powerful turbine.
Concerning the type of material, for the 80 m height towers, the concrete tower has the worst
performance (highest value of emissions). In this case, the steel tower has a slight advantage in
relation to the hybrid tower. The 100 m concrete tower has still the worst performance when
compared with the other towers, although in this case, the difference to the steel tower with WFC
is lower. On the other hand, for the 150 m height towers, the steel tower with WFC has the worst
performance, followed by the concrete solution.
Focussing on the use of steel segments in the towers, the use of bolted friction connections (FrC)
has clearly an advantage in relation to the use of welded flange connections (WFC). The use of FrC
enables to reduce the amount of steel used in the structure. In addition, the benefit by the use of
FrC is enhanced by steel towers in comparison to hybrid towers.
To summarize, steel towers with WFC are penalized by the height of the tower and for the height of
150 m, the steel tower has the worst performance. On the other side, steel towers with FrC have
the best performance in all cases and the advantage of FrC is enhanced by higher towers.
The results of the life cycle analysis are indicated in Table 2-13 for the towers with a height of 80
m. The lower values in each environmental category are highlighted in bold.
23
Table 2-13: Environmental indicators for the steel tower with 80 m (1st scenario)
Environmental Concrete Steel tower Hybrid tower

category tower WFC FrC WFC FrC
- -
ADP fossil [MJ] 3,62E+06 3,63E+06 3,58E+06 1,4% 2,85E+06 2,82E+06 1,0%
- -
AP [kg SO2-Eq.] 8,52E+02 8,96E+02 8,84E+02 1,3% 7,84E+02 7,77E+02 0,9%
- -
EP [kg PO4-Eq.] 1,24E+02 1,13E+02 1,12E+02 1,0% 1,10E+02 1,09E+02 0,6%
- -
GWP [kg CO2-Eq.] 4,21E+05 3,32E+05 3,28E+05 1,2% 3,69E+05 3,66E+05 0,6%
- -
ODP [kg R11-Eq.] 1,13E-03 6,42E-03 6,25E-03 2,6% 2,97E-03 2,88E-03 3,2%
- -
POCP [kg Ethene-Eq.] 8,73E+01 1,32E+02 1,30E+02 1,4% 9,33E+01 9,23E+01 1,1%
From Table 2-13 the hybrid tower gets the best performance in three environmental categories;
while, the concrete tower gets the best performance for two. Steel towers using FrC achieve the
best performance in one category.
The benefit by the use of FrC in relation to WFC is highlighted by column , representing the
reduction of emissions in relation to the latter. The use of FrC has advantage in relation to WFC for
all impact categories. This benefit is generally higher for the steel tower in comparison with hybrid
tower.
In the case of the towers with a height of 100 m, the results of the life cycle analysis are
summarized in Table 2-14.

-
ADP fossil [MJ] 7,13E+06 9,28E+06 8,00E+06 13,8% 6,58E+06 6,30E+06 -4,2%
-
AP [kg SO2-Eq.] 1,84E+03 2,42E+03 2,07E+03 14,3% 1,82E+03 1,75E+03 -4,1%
-
EP [kg PO4-Eq.] 2,76E+02 2,96E+02 2,58E+02 12,8% 2,53E+02 2,45E+02 -3,2%
-
GWP [kg CO2-Eq.] 9,56E+05 8,48E+05 7,48E+05 11,8% 8,33E+05 8,12E+05 -2,6%
-
ODP [kg R11-Eq.] 2,31E-03 2,05E-02 1,65E-02 19,7% 8,14E-03 7,27E-03 -10,7%
-
POCP [kg Ethene-Eq.] 1,82E+02 3,75E+02 3,14E+02 16,3% 2,25E+02 2,12E+02 -5,8%
The same conclusions can be drawn from Table 2-14. However, in this case, the benefits achieved
by the use of FrC, given in column , are much higher than in the previous case, both for the steel
towers and for the hybrid towers. Likewise, steel towers achieved higher reductions in comparison
with hybrid towers.
Finally, in the case of the towers with a height of 150 m, the results of the life cycle analysis are
indicated in Table 2-15.
24

-
ADP fossil [MJ] 1,14E+07 2,00E+07 1,75E+07 12,7% 1,32E+07 1,22E+07 -7,5%
-
AP [kg SO2-Eq.] 3,07E+03 5,22E+03 4,54E+03 13,1% 3,66E+03 3,39E+03 -7,3%
-
EP [kg PO4-Eq.] 4,83E+02 6,14E+02 5,40E+02 12,1% 5,07E+02 4,78E+02 -5,7%
-
GWP [kg CO2-Eq.] 1,67E+06 1,70E+06 1,50E+06 11,7% 1,63E+06 1,56E+06 -4,7%
-
ODP [kg R11-Eq.] 3,15E-03 5,00E-02 4,20E-02 16,0% 1,89E-02 1,57E-02 -16,6%
-
POCP [kg Ethene-Eq.] 2,87E+02 8,53E+02 7,33E+02 14,1% 4,64E+02 4,18E+02 -10,0%
In this case, the concrete tower achieves the best performance in four environmental categories,
while, the steel structure and the hybrid tower achieve the best performance in one environmental
category each. The advantage by the use of FrC is still highlighted for the steel tower and the
hybrid tower. The benefits achieved by the hybrid tower have slightly increased in relation to the
previous case.
2.3.2.3 Analysis of environmental impacts for a life span of 40 years

Extension of the service life in the same location (2nd scenario)
In the second scenario, it is considered that after the initial period of 20 years, it is decided to
extend the service life of the towers for another period of 20 years in the same wind farm. In this
scenario, only bolted friction connections (FrC) are considered for the steel and hybrid tower. Since
no transportation on public roads is necessary, tubular segments do not need to be dismantled and
therefore no significant differences exist between FrC and WFC tower types.
In this scenario, the scope of the analysis comprehends the production of materials, the
transportation of materials to construction site, the erection of the tower, the rehabilitation of the
tower (in year 20), the demolition of the tower (in year 40) and the transportation of the waste to
a recycling plant or a landfill. In order to extend the service life of the structures, the following
rehabilitation assumptions were taken: (i) complete replacement of the coating of the steel tower
and the steel part of the hybrid tower; and (ii) rehabilitation of the surface of the concrete tower
and the concrete part of the hybrid tower.
According to this new scenario, the global warming potential for the three types of towers (steel,
concrete and hybrid tower) and for the three heights (80 m, 100 m and 150 m) is illustrated in
Figure 2.5.
25
Hybrid tower FrC40 (80 m)
Steel tower FrC (80 m) GWP [kg CO2 eq.]
0,00E+00 5,00E+05 1,00E+06 1,50E+06 2,00E+06

Figure 2.5: Life cycle analysis of wind towers (2nd scenario).
In this scenario, the steel tower has the most beneficial performance (lower value of emissions), in
all cases, followed by the hybrid tower. On the other hand, the concrete structure has the worst
performance. Comparing Figure 2.6 (considering only FrC) and Figure 2.5 it is noted that the
results are very similar. The same conclusion would be achieved when comparing the results of the
remaining environmental categories.
Therefore, the extension of the service life of the wind towers does not significantly change the
results obtained from the first scenario, since the difference results from maintenance works, which
have negligible effect.
Extension of the service life in another location (3rd scenario)
In the third scenario, it is considered that, in the end-of-life stage (in year 20), the structure of the
wind tower is deconstructed and reused in another location for another period of 20 years. The new
location is considered to distance 50 km from the initial location.
Likewise, in this scenario, only bolted friction connections are considered for the steel tower and
hybrid tower, which enables the complete deconstruction of the towers.
In this case, the scope of the analysis comprehends the production of materials, the transportation
of materials to construction site, the erection of the tower, the deconstruction of the tower (after
20 years), the transportation of the tower to the new location, the re-assemblage of the structure,
the rehabilitation of the tower, the demolition of the tower (after 40 years) and the transportation
of the waste to a recycling plant or a landfill.
The three types of towers are assembled in different ways, which affect the way they are
deconstructed. Therefore, the three types of towers have different scenarios for the deconstruction
stage and the following stages. However, in all cases it is considered that after the first period of
20 years the foundations are demolished and the materials are recycled. Thus, in year 20 new
foundations are built in the new location for all types of towers.
The steel structure is entirely demountable due to the novel bolted connection referred before, the
bolted friction connection. Therefore, the structure is transported to the new location, the same
way it was initially transported from the production plant to the construction site.
Likewise, the steel part of the hybrid solution is demountable and transported the same way to the
new location. The concrete tower and the concrete part of the hybrid tower is demolished and
recycled according to the recycling rates indicated in the previous sub-section.
In order to extend the service life of the structures, the steel tower and the steel part of the hybrid
tower are assumed to have the coating system totally replaced in year 20.
26
New elements for the concrete tower and for the concrete part of the hybrid tower are produced in
year 20 and transported to the new location, as described in the first scenario.
According to this new scenario, the global warming potential for the three types of towers (steel,
concrete and hybrid tower) and for the three heights (80 m, 100 m and 150 m) is illustrated in
Figure 2.5.
Hibrid tower (150 m)
Steel tower (150 m)
Hibrid tower (100 m)
Steel tower (100 m)
Hibrid tower (80 m)
Steel tower (80 m) GWP [kg CO2 eq.]
0,00E+00 1,00E+06 2,00E+06 3,00E+06 4,00E+06
Figure 2.6: Life cycle analysis of wind towers (3rd scenario).
In this scenario, the steel tower has the most beneficial performance (lower value of emissions), in
all cases, followed by the hybrid tower. On the other hand, the concrete structure, due to higher
consumption of materials (namely cement), has the worst performance.
The results of the life cycle analysis, taking into account the remaining environmental categories
are indicated in Table 2-16. The lower values in each environmental category for the three types of
towers (80 m, 100 m and 150 m) are highlighted in bold.
Table 2-16: Environmental indicators for the three towers (3rd scenario)
Tower 80 m Tower 100 m Tower 150 m

Steel Concrete Hybrid Steel Concrete Hybrid Steel Concrete Hybrid
tower tower tower tower tower tower tower tower tower
ADP fossil [MJ] 5,35E+06 7,24E+06 4,99E+06 1,18E+07 1,43E+07 1,09E+07 2,38E+07 2,27E+07 2,10E+07
AP [kg SO2-Eq.] 1,27E+03 1,70E+03 1,38E+03 3,01E+03 3,68E+03 3,05E+03 6,02E+03 6,14E+03 5,83E+03
EP [kg PO4-Eq.] 1,69E+02 2,48E+02 1,99E+02 3,92E+02 5,52E+02 4,40E+02 7,46E+02 9,65E+02 8,51E+02
GWP [kg CO2-Eq.] 5,14E+05 8,42E+05 6,82E+05 1,17E+06 1,91E+06 1,49E+06 2,12E+06 3,35E+06 2,84E+06
ODP [kg R11-Eq.] 6,75E-03 2,26E-03 3,69E-03 1,97E-02 4,62E-03 9,31E-03 4,86E-02 6,30E-03 2,05E-02
POCP [kg Ethene-Eq.] 1,70E+02 1,75E+02 1,53E+02 4,23E+02 3,65E+02 3,43E+02 9,14E+02 5,74E+02 6,67E+02
From Table 2-16 it is observed that for towers with 80 m and 100 m, the steel tower has the best
performance (lower values in three out of the six environmental categories), followed by the hybrid
tower and in last the concrete tower.
In the case of the towers with 150 m, each tower has the best performance in two environmental
categories.
The advantage of steel towers in case of reuse is emphasized by the comparison between scenario
1 and scenario 3. The reuse of steel towers for another period of 20 years increases the
environmental impacts of about 50%; while, for the concrete towers, the values are duplicated.
27
2.3.2.4 Comments and Conclusions
The design results of a set of tubular onshore wind towers and respective foundations performed in
Task 1.1 and Task 1.2 were herein used to assess the life-cycle performance of nine case studies,
defined by different tower heights (80 m, 100 m and 150 m) and type of structural solution: steel,
concrete and hybrid steel-concrete.
Focussing on steel towers with bolted friction connection (FrC) and welded flange connection
(WFC), the former is clearly more beneficial than the latter for wind towers. The save of materials
by the use of FrC enables to reduce environmental impacts up to 16%. The reduction of impacts
increases with the height of the tower. Moreover, the use of FrC enables for the complete
dismantling of the tower and its reuse somewhere else. The reuse of towers (partially or
completely) is a strong advantage of steel towers in comparison with other types of towers,
particularly in relation to the environmental category of global warming.
The results presented in this work are based on the design of the towers considering seismic
loading. In order to assess the influence of seismic design on LCA, a comparison between the
design of towers with and without the consideration of seismic loading was performed for the first
scenario. It was concluded that without seismic design the results of LCA may be reduced up to
27%. The reduction is more significant for concrete towers than for steel or hybrid towers due to
the higher relative importance of foundations in the former case.
In terms of global life-cycle performance of the wind towers, and considering equal importance for
all environmental categories, it may be concluded that for heights up to 100 m hybrid towers with
FrC are the most efficient solution. For higher heights, the concrete tower becomes more efficient.
Transversal to all type of analysis and material is the conclusion that the increase of tower height
from 80 meters to 150 meters multiplies by more than 3 times the global warming indicator. The
most penalized situations are those represented by scenario 1 and 2. The possible reuse of parts or
of the entire tower, foreseen in scenario 3 when using steel towers built with friction connections,
may deem this drawback.
2.3.3 Steel tubular towers: Alternatives for cross-sections

Polygonal cross sections already have been used in the past in many applications. A good example
is numerous lighthouses which have been constructed using folded plates, Figure 2.7a). Compared
to towers with circular cross sections, the fabrication of towers with polygonal cross sections can be
done in an even more automated way and less expensive as no highly skilled workers are needed
to control the rolling process. Other application areas, such as modern mobile cranes take benefit
from the higher buckling resistance of polygonal cross sections (Figure 2.7b)).
The main reason that polygonal shaped towers haven’t become the standard solution in the past is
the elastic resilience effect of the plate during the bending process. This effect is mainly dependent
on the yield strength of the steel plate, which normally varies from plate to plate. But as this
problem already has been solved technically by many fabricators, e.g. in the ski-lift tower industry,
the use of polygonal shaped towers for wind turbines is considered as practical and realistic
solution.
28
a) b)
Figure 2.7: a) Lighthouse; b) Mobile crane
2.3.3.1 Buckling resistance of plated structures vs. shell structures

The elastic bucking stress of a perfect shell is much higher compared to those of a polygonal tube
with comparable dimensions. Furthermore, circular cross sections are more sensitive to
imperfections than plated structures. As the critical buckling stress for shells is mainly dependent
on its thickness to radius-ratio, a slight variation in the roundness of the circular cross section
leads to a local increase of the radius R and thus to a decrease of the critical buckling stress cr of
the shell, see also Figure 2.8 leading to the higher reduction factor of the circular cross-section
compared to the polygonal cross-section.
Figure 2.8: Cross section imperfection: out-of-roundness
The resistance and critical stress of cylinders with a polygonal cross-section is different compared
to the circular cross-section. The folds increase the local stiffness of the cylinder. It is reflected in a
buckling shape where all edges represent “fixed-nodes”, see Figure 2.9. This edge-effect allows for
calculation of the whole cross section acc. to EN1993-1-5 where each flat side of the polygonal
tube can be handled as one single flat plate with lateral supports along all edges.
29
Figure 2.9: First Eigenmode of a polygonal (dodecagonal) cross section (left hand side)
and circular cross-section (right hand side) [14]
2.3.3.1.1 Parametric studies

FEM-Analysis, 4 point bending
The main goal of a parametric study shown below is to check the buckling resistance of tubes with
respect to an economic optimisation of the tower cross-section. The finite element program
MARC/Mentat [15] is used.
The 4 point-bending model has a constant bending moment between two loads. The advantage
compared to a cantilever loading, which actually better reflects tower behaviour, is a fact that no
disturbing load introduction or stiffening effects exist at the location of the maximum bending
moment. Thus the FEA results can be compared to the hand-calculation results according to
EN1993-1-5 and EN 1993-1-6 more easily.
The FE calculations have been performed due to two reasons.
1. to check if the edge-effect of polygonal tubes can be taken into consideration for the hand-
calculation methods (using both linear buckling analysis and GMNI-analysis),
2. to verify that the critical buckling stresses for tubes under pure bending determined by the
hand calculation formulas according to EN1993-1-6 do not vary from the FE results (using
linear buckling analysis).
Figure 2.10 shows that all edges of longitudinal folds of polygonal cross-sections considered remain
“fixed” in the first eigenmode where the local buckling occured.
30
Quadrangular cross section Octagonal cross section
20-angular cross section Circular cross section

Figure 2.10: First eigenmode of tubes under bending with different cross sections [15]
The results of the parametric study are shown in Figure 2.11. To allow for comparability, the
moment of inertia I and shell thickness t has been kept constant for all calculations. The values
have been defined to
t = 22 mm
I = 0.553 m4
whereas the chosen dimensions correspond to the mean dimensions of the MM92 wind tower given
in [29].
31
MRd.eff [MNm]
120.0
100.0
80.0
60.0
40.0
Polygon (sigma.crit from FEA)
Polygon (sigma.crit from hand-calculation)
20.0
Circle (sigma.crit from FEA)
Circle (sigma.crit from hand-calculation)
0.0
0 4 8 12 16 20 24
Number of edges
Figure 2.11: MRd,eff(number of edges): Comparison between polygonal and circular cross
section (reduced stress method)
The points in the diagram in Figure 2.11 represent the GMNI-FE calculations. All results from hand-
calculations according to Method 2 (reduced stress method) are represented by solid lines. From
these calculations the following conclusions can be drawn:
1. In all cases the GMNI-FE calculation leads to a higher bending resistance than the hand-
calculation according to EC3.
2. By determining the critical buckling stress cr,p by means of a FEA, the bending resistance
according to the hand-calculation can be increased. This is mainly due to the reason that for
the hand-calculation all plate edges of the sup-panels are assumed to be perfectly hinged,
whereas the discrete FE-model considers the actual clamping effects at the edges.
3. The bending resistance of the tubes with polygonal cross section increases with increasing
number of edges.
4. For the given plate thickness t and moment of inertia I a number of 16 edges is sufficient to
reach a higher bending resistance with a polygonal cross section compared to a circular cross
section (blue lines).
5. There is no big difference between the critical buckling stress determined by means of a FEA
compared to those obtained by analytic hand-calculation formulas given in EN 1993-1-6
(compare both blue lines).
To check the buckling resistance of tubes with respect to an economic optimisation, the bending
resistance as a function of the material consumption has been checked in the next step. Figure
2.12 shows the bending resistance as a function of the gross cross section area. All results are
shown as relative values whereas the reference values for all parameters are the bending
resistance MRd,eff,circle and the area Acircle of the circular cross section. As shown in Figure 2.11 for a
tube with a dodecagonal cross section (12 edges) the bending resistance is equal to the bending
resistance of the tube with a circular cross section. The amount of steel needed is almost the same
(just 1% higher). With increasing number of edges the buckling resistance increases whereas the
gross cross section decreases.
These good results can be even improved by the use of Method 1 (Effectivep Cross Sections
Method) instead of Method 2 (Reduced Stress Method).
32
160%
MRd.eff.i / MRd.eff.Circle
Polygon (sigma.crit from FEA)
140% 24 Polygon (sigma.crit from hand-calculation)
16 Circle (sigma.crit from FEA)
120% Circle (sigma.crit from hand-calculation)

12
100%
8
80%
4
60%
40%
20%
0%
99% 100% 101% 102% 103% 104% 105% 106% 107% 108%
Ai / ACircle
Figure 2.12 MRd,eff/MRd,eff,circle(Apolygon/Acircle): Comparison between polygonal and circular

cross section (Reduced Stress Method)
Conclusions
The parametric studies have shown that in general the application of polygonal cross sections
allows for a material reduction compared to circular cross sections due to better buckling
characteristics. For example if 24 edges are applied, the moment capacity of the section can be
increased by 20% compared to that one of a circular hollow section with the same mass. However,
these calculations are all based on “perfect” closed sections; effects due to discrete longitudinal
bolts connecting the plates among each other are not considered. Therefore further parametric
studies as well as tests are necessary to take these effects into consideration.
FEM-Analysis, axial compression
A preliminary study of a cantilever column using a circular cross-section with three different
radiuses of 2, 4 and 6 meter and with a length of 5 m is performed using ABAQUS. The thickness
of the column is 8 mm thick. For the sake of comparison a polygonal cross section a column with
radiuses of 2 and 4 meter, thickness of 8 mm and 5 meters length is evaluated. The polygonal
cross-section consists of 16 folds. Tower segments with two and four overlapping parts are
considered. The material properties corresponding to nominal steel grade S355 is used for in all
FEA. For nonlinear analysis various geometrical imperfections are used, harmonic, discrete
symmetrical and discrete unsymmetrical, see Figure 2.13.
harmonic discrete symmetric discrete un-symmetric
Figure 2.13 Imperfection types used in non-linear analysis
Design calculation according to EN 1993-1-6 is performed using flow chart shown in Figure 2.14.
33
Figure 2.14 Flow chart of the design stress method
For the sake of illustration, the ultimate load resistance for a cylinder of 4 m diameter using
harmonic shape of imperfections is shown in Table 2-17 and Table 2-18.
34
Table 2-17 Ultimate load of 4 m diameter cylindrical cross section.
Imperfections Analytical
[mm] EN 1993 1-6 [kN]
3,2 17264
5,1 14329
7,9 10016
Table 2-18 Ultimate load of 4 m diameter polygonal cross section.
Imperfections 2 segments 4 segments
Disp [mm] Load [kN] Disp [mm] Load [kN]

[mm]
3,159 8,02 25841 7,44 29830
5,055 8,02 25336 7,40 27836
7,898 8,01 24642 7,563 25680
2.3.3.2 Influence of longitudinal folds on the buckling stress

In order to study the influence of longitudinal joints on the buckling behaviour, parametric studies
are performed using two different approaches:
 buckling behaviour under bending load (RWTH)
 buckling behaviour under compression load with different kinds of imperfection (LTU)
The different parametric studies are discussed separately in the following subparagraphs. In
general, two failure modes can occur due to buckling:
 global buckling of the cross section
 local buckling of the plates between bolts
where the first failure mode is influenced by the overlapping of the plates and the latter is
influenced by the distances and spacing of the bolts.
By varying the geometry and the spacing of the bolts, the influences of these parameters can be
evaluated. The knowledge gained within this WP is used to plan the down scaled tests which are
then used to develop design criteria for the connection.
Buckling under bending loading
To study the influence of the bolt layout, a tower segment is modelled using Ansys 13.0. The model
consists of shell elements and represents a section of 20 m with a diameter of 4.2 m. The
numerical analysis is a non-linear investigation. Figure 2.15a) shows the whole FE-model and
illustrates the deformation. On the left side, the tower is fixed and on the right side it is loaded.
The bolts are realized by rigid coupling of corresponding nodes. This assumption holds for the
elastic range, were slip does not occur.
35
Figure 2.15 a) Overview of the finite element model and b) detail of the overlapping area
with node coupling
The contact in the overlapping area is modelled with contact elements. Figure 2.15b) illustrates the
area of contact. In the figure, the blue colour denotes areas where no contact can occurs, yellow
means a possible contact orange slip + pressure and red clamping + pressure.
In Figure 2..16 the deformed shape with global buckles in the area under compression is shown.
One can see that plate buckling occurs between the folds of the plates.
Figure 2..16 deformed shape of the finite element model
In the finite element calculations the following parameters are varied:

 closed and segmented cross section,
 different steel grades (from higher-strength to high-strength steel),
 different layout of bolts (one row, two rows, two staggered rows, different interspaces),
 different surface treatment (different friction coefficients),
 opportunity to save material (shell thickness, exterior radius).
In the following the results of investigations where an effect on the buckling performance is of
major effect are shown.
Comparison of closed and segmented cross section
Figure 2.17 shows the load displacement curves from a closed (green) and a segmented (purple)
model. As a result of the additional material in the segmented model, resulting from the
overlapping, the elastic load-bearing capacity of the segmented section is comparably higher. It
can be concluded that the segmentation does not have a negative effect on the buckling
performance if the bolts are designed and placed with short distances.
36
180
160
140
Biegemoment [MNm]
120
100
80
60
40
SQS
20
VQS
0
0 50 100 150 200 250 300 350 400
Weg [mm]
Figure 2.17 Load displacement curve for cross section with (SQS) and without (VQS)
longitudinal joint
Comparison of different bolt layouts: variation of number of bolt

rows
Figure 2.18 shows the load displacement curves of a segmented model with one row of bolts
(purple line), two rows (green line) and staggered bolts (red line). By the application of a second
row of bolts the stiffness of the section can be increased. The cross section with staggered bolts
nearly reaches the maximum stiffness of the section with two rows of bolts. Therefore under
economic consideration, staggered bolts could be a possibility to save production as well as
maintenance costs.
160
140
120
Biegemoment [MNm]
100
80
60
40
1 Reihe
20 2 Reihen
2 versetzte Reihen
0
0 50 100 150 200 250 300 350 400
Weg [mm]
Figure 2.18 Load displacement curve for cross section with different bolt distances
Comparison of different bolt layouts: variation of bolt distances

Figure 2.19 shows the load displacement curves from segmented models with three different
distances between the bolts – 110 mm (purple), 250 mm (green) and 500 mm (red). 250 mm and
37
500 mm are bigger than the maximum distance allowed according to Eurocode 3. As Figure 2.19
demonstrates, there is no significant decrease of load-bearing capacity between 110 mm and
250 mm. For these ‘small’ distances, no local buckling occurred. For 500 mm, there is a change in
the failure mode, which results in local buckling between the bolts. However, this behaviour needs
to be verified by tests.
160
140
120
Biegemoment [MNm]
100
80
60
40
p1= 110 mm
20 p1= 250 mm
p1= 500 mm
0
0 50 100 150 200 250 300 350 400
Weg [mm]
Figure 2.19 Load displacement curve for cross sections with different bolt distances
2.3.3.3 FEA of polygonal vs. circular tower segments

A parametric study of a tubular section designed as the lower part of a tower with two different
cross-sections (polygonal and circular) is performed. The analysis is focusing on the angle between
the adjacent edges for polygonal cross-sections compared with a circular tube. The number of
edges for polygonal cross-section varies from 8 to 22 edges. The tube has the same length of 5000
meter and a thickness 8 mm for all cross-sections. The thickness of the cross-section is used as the
minimum thickness to obtain the cross-section “4” acc. to the rules EN 1993 part 1-1.
Two methods were used in order to compare the results between analytical and numerical
analyses. Analytical analyses is performed using the rules of the EN 1993 part 1-5 for the
polygonal shapes considered the adjacent edges as simple supported plates and EN 1993 part 1-6
for circular tube shell; calculations for polygonal shapes were done using the program Matlab and
for the circular cross-section with Mathcad . In order to compare the results and start the
investigation of the influenced of the adjacent angle a numerical analysis is done with the help of
the commercial software for Finite Element method (FEM) ABAQUS.
38
Figure 2.20: Polygonal cross-section section with number of edges ranged from 8 to 22
with increment by 2
Table 2-19 shows the geometry properties and the angle of the adjacent edges of the cross-
sections shown in Figure 2.20. One of the main interests is to study the tower segment as the thin
shells cross-sections. Therefore, it was necessary to evaluate the classification of the cross-section
for plastic global analysis acc. to EN 1993 part 1-1 chapter 5.6 table 5.2 (Annex1), for each loading
case: bending and axial compression.
Table 2-19: Geometry and properties of polygonal tubes with 5000 mm length and 8 mm
thickness.
Angle Width of Class-section Class-section

the edge Iyy
Number Deg Area Compressive Bending
of folds (b) 4 2 1011
(∝ 10 (mm ) EN 1993 part EN 1993
(mm4)
(mm) 1-1 part 1-1
8 135 1527.67 9.777 1.7575 4 4
10 144 1266.6 9.868 1.8407 4 4
12 150 1233.6 9.919 1.888 4 4
14 154.3 888.36 9.949 1.916 4 4
16 157 778.8 9.969 1.935 4 3
18 160 693.20 9.982 1.948 4 3
20 162 624.48 9.991 1.958 4 3
22 163.6 568.20 9.999 1.965 4 3
Circle 4m diameter 10.03 1.999 4 4
The widths of the polygonal cross-section are taken as the width of the adjacent edge which means
that every adjacent edge is considered as a single thin plate for polygonal cross-sections.
39
L
b L=5000 mm
Single supported plate
t=8mm
Imperfection b/200, acc. EN 1993 part 1-5.
Figure 2.21: Tower segment, 4m in diameter, considered in FEA
Figure 2.22: a) Buckling mode and b) failure shape of a polygonal cross-section made of
three segments with 8, 18 and 20 edges and a circular cross-section under axial
compressive. All columns have a 4m diameter.
Figure 2.22 shows the deformation shape of the columns with different number of folds under axial
compressive conditions, the buckling mode shows local buckling between the folds. Comparison
between the circular cross-section and the polygonal shapes shows that after the 20 folds the
deformation of the local buckling and ultimate shape deformation are almost the same for the two
columns.
40
8 16 20 Circular
(a)
(b)
Figure 2.23: a) Buckling mode and b) failure shape of a polygonal cross-section made of
three segments with 8, 18 and 20 edges and a circular cross-section under bending. (All
tubular sections have 4 m diameter).
Figure 2.23 shows the deformation shape of the tower segment with different number of fold under
the pure bending. The buckling mode shows local buckling between the folds. Comparison between
the circular cross-section and the polygonal shapes shows that after the 20 adjacent edges the
deformation of the local buckling and ultimate shape deformation are almost the same for the two
columns.
FEA results are compared with results obtained from hand-calculation method used in Eurocodes,
in axial compression and due to bending moment.
Comparison of results of FEA show lower values compared with design results. The scattering
between results is around 10 % as the maximum for tubes with 8 folds. For the rest of the tower
segment the differences are between 0.2 to 2 %.
Table 2-20: Analytical vs. numerical analysis results for critical stresses and critical loads
under compressive loading.
Numerical analysis Analytical Analysis

Number of σcrit (MPa)(Numerical)/
(FEM) (EN 1993 part 1-5)
folds σcrit (MPa)(Analytical)
σcrit (MPa) Nc, crit (kN) σcrit (MPa) Nc, crit (kN)
8 22,35 2185 24,37 2383 -0.09
10 33,77 3333 35,28 3482 -0.04
12 47,63 4724 48,77 4837 -0.02
14 63,94 6343 64,78 6445 0.008
16 82,76 8234 83,28 8301 -0.007
18 104,02 10384 104,24 10406 -0.002
20 125,28 12517 127,72 12762 -0.02
22 152,17 12216 153,66 15364 -0.009
Circle 4m 488,53 49000 509,21 51089 -0.042
41
FE and results obtained according to Eurocodes show good agreement for polygonal cross-sections
A rather big difference is shown for the circular cross-section under bending resistance analysis,
comparing with results obtained from EN1993-1-6, about 24%. However this difference is not
investigated further because it is out of the scope this project. The results from the FEA on circular
cross-section are used here for the sake of comparison with results obtained on polygonal cross-
section.
Table 2-21: Numerical and analytical results for the compressive resistance load NcR and
bending resistance moment for polygonal cross-sections ranged from 8 to 20 edges and
comparison with a circular cross-section of 4 m diameter.
NcR NcRch MRch MR/MRch

Number MR
(kN) (kN) NcR/NcRch (MNm)
of edges (MNm) FEA
FEM EC EC
8 8801 8592 0.02 12.9 14.1 -0.09
10 9754 10270 -0.05 14.2 15.12 -0.06
12 12201 11998 0.02 17.6 17.35 0.01
14 13719 13646 0.005 18.8 18.52 0.01
16 15027 15315 -0.02 20.6 19.82 0.04
18 17191 16872 0.02 22.3 21.20 0.05
20 18710 18800 -0.05 23.24 22 0.05
22 19436 21148 -0.08 25.27 24.08 0.04
Circle 4 m
18012 17264 0.04 24, 85 18,915 0.24
Diameter
Figure 2.24: Ultimate axial resistance N.cR for polygonal cross-sections ranged from 8
to 22 edges and one circular cross-section. Columns have a diameter of 4 m and 5 m
length.
42
Figure 2.25: Ultimate stresses for polygonal cross-sections ranged from 8 to 22 edges
and one circular cross-section. Columns have a diameter of 4 m and 5 m length.
Figure 2.26: Comparison of the Moment between analytical and numerical analysis for
tubes ranged with 8 adjacent edges to 22 compared with a circular tube.
43
Table 2-22: Bending resistance moment for analytical and numerical analysis with three
imperfections measurements acc. to EN 1993 part a-6.
Analytical Numerical
Imperfection Analytical/N
Class Bending Moment Bending Moment
(mm) umerical
(MN*m) (MN*m)
A 3,159 18,91 24,93 -0.32
B 5,055 15,7 24,71 -0.36
C 7,898 10,97 24,41 -0.56
Tower segment under axial compression shows good agreement between numerical and eurocode
calculations. The results indicated that the number of folds for optimal resistance compared to a
circular cross-section is for polygonal cross-sections with 18 to 20 fold. Difference between results
did not exceed the 10%.
The angle of the adjacent plates 1600 to 1620 degrees is sufficient to ensure resistance under
compression so the fold behaves as joint.
FEA results show that the optimum angle compared to circular cross-sections was allocated
between 1620 and 163.60 for number of folds between 20 and 22. For analytical results the bending
resistance moment, the optimum angle of the adjacent edges was 1570 for 16 edges.
2.3.3.4 Optimum spacing of bolts in single-lap connections of a hexagonal

element consisting of four folded steel sheets
Various FE models are used to investigate the most suitable model in terms of computational time
and accuracy of results to be achieved. Two different models are shown below: the “complete
tower segment model” with realistic modelling of the discretely distributed bolts along the
overlapping plates of four folded steel sheets creating the segment model and simplified model
where only overlapping parts are modelled.
Figure 2.27: General view and a cross-section of an analysed 5-meter long

hexadecagonal element (dimensions are given in millimetres)
44
2.3.3.5 Bolt spacing and geometry of the overlapping parts on a simplified
model
The analysis was performed on simplified shell models consisting of two overlapping plates cut out
from complete model as shown in Figure 2.27. The following range of parameters is analysed:
 different cases of the geometry of the overlapping in which axis of the row of bolts was
shifted from the starting position at 125mm see Figure 2.27 an then distances 191, 3 mm,
257.6 mm, 323.9 mm and finally at 390.2 mm which is in the middle between the two
longitudinal edges of the model.
 different bolt spacings: 416.67mm, 500mm, 625mm, 833.33mm and 1000mm
Figure 2.28: First two out of 5 analysed positions of axis of row of bolts
The analysis is done in two steps. In the first step the imperfections were generated. These
imperfections were then introduced in the second step, in which the models were subjected to
boundary conditions shown in Figure 2.28. The imperfections were generated by putting pressure
on overlapping parts that forced them apart Figure 2.29. The imperfection is generated so that
buckles are obtained of equal amplitudes in all models. Amplitudes of two middle buckles were
(regardless of the position of the axis of the bolt row) always equal 7.80 mm for a wider plate and
2.50 mm for the narrower (h1 and h2 see Figure 2.29). These values correspond to 1/100 of width
of the wider and narrower plate (dimensions a and b in Figure 2.29). Amplitudes of the
intermediate buckles were smaller by 5% and of the outermost by 10%.
Figure 2.29: Imperfections in simplified models
In the analyses both elastic and elasto-plastic material models are used with Young’s modulus
equal 210GPa and Poisson’s ratio 0.3. Curve representing elasto-plastic model definition and the
characteristic points of the curves are is shown in Figure 2.30.
45
stresses [N/mm2] 0 560 585 700 900
elastic strains [%] 0 0,2666667 0,27857143 0,3333333 0,4285714
plastic strains [%] 0 0 3,80142857 12,866667 59,571429
total strains [%] 0 0,2666667 4,08 13,2 60
Figure 2.30: Elasto-plastic material model definition
Influence of geometry of the overlapping on the connections behaviour is analysed using a

prefectly elastic material and elasto-plastic hardening materail.
Figure 2.31 shows behavior of models with constant bolt spacing but for different positions of bolt
axis. It can be observed that shifting the axis towards the middle deteriorates the models
resistance. Therefore, making the narrower part of the overlapping as samll as possible is the best
alterantive for the connection’s geometry. It can be additionally observed that the position of the
axis of the row of bolts does not influence longitudinal stiffness of the connection.
Figure 2.31: Elasto-plastic material model definition Force-displacment curves for

models under compression with constant bolt spacing and different distance of axis of
row of bolts from the longitudinal edge of the model (material model: elastic)
46
Figure 2.31 shows behavior of models under compression for different bolt spacing but with
constant 125mm-distance of bolt axis from the longitudinal edge. Also a prefectly elastic material
definition was used in all models. Curves in the figure show a clear relation between bolt spacing
and models’ behaviour - the bigger the bolt spacing, the smaller the longitudinal stiffness. The
same conclusions can be drawn for other positions of axis of bolts and for elasto-plastic definition
of material. The only additional observation for models with elasto-plastic material is that points at
which curves bend in case of models with elastic material are points of limit loads, see Figure 2.33.

models under compression with different bolt spacings and with constant 125mm-
distance of axis of row of bolts from the longitudinal edge of the model (material model:
elastic)
47
models under compression with different elastic and elasto-plastic material models.
2.3.3.6 Analysis of the “complete tower segment model” resistance under

compression and bending
The analysis was performed for models with two types of imperfections. Imperfections type 1 were
imperfections generated by applying pressure pushing the overlapping parts apart as it is described
in section above whereas the imperfections type 2 were in the form of buckling mode
corresponding to the first eigenvalue. The second type of imperfections were scaled in such a way
so as to obtain the value of the amplitude of the greatest buckle equal 7.80mm.
Figure 2.34: Elasto-plastic material model definition Comparison of force-displacement

curves for models with different bolt spacings and 2 types of imperfections.
In Figure 2.34 dashed lines representing models with geometric imperfections in form of first
buckling modes (imperfection type 2) show that this type of imperfections is more disatvantageous
for the resistance in compression. Firstly, stiffnesses are slightly smaller in comparison to
stiffnesses of contiuous curves showing results of models with imperfections generated by applying
pressure that pushes the overlapping parts apart (imperfection type 1). Secondly, the ultimate
loads are smaller.
First of these observations probably results from the fact that imperfection type 2 concerns only
the overlapping parts of the models. Thus, longitudinal stiffness in the other intermediate plates
remains unaffected. Imperfection type 1 generates buckles on both intermediate and overlapping
plates and therefore reduces stiffness of whole models. Due to this difference between both types
of imperfections it cannot be stated that imperfetion type 1 should not be considered in further
analyses. Probably if the imperfectino type 1 was enriched by introducing buckles also on
intermediate plates it would be more disadvatageous for the models as it is right now in the current
form.
In Figure 2.34 and in Figure 2.37 it can be seen that increasing bolt spacing from 416.67mm to
625mm does not influence the models’ resitance in contrast to the increase from 625mm to
1000mm. Therefore, it would be worth investigating models with bolt spacing euqal 833.33 mm.
48
a) b)
Figure 2.35: Elasto-plastic material model definition Stress distributions (effective stress
range 0-500MPa) for peak loads: a) bolt spacing 416.67mm and 625mm; b) bolt spacing
1000mm
a)
b)
range 0-500MPa) for the post-buckling range of loads: a) bolt spacing 416.67mm and
625mm; b) bolt spacing 1000mm
Analysis of the complete tower segment model under pure bending is shown below.
49
Figure 2.37: Elasto-plastic material model definition Comparison of force-displacement
curves for models with different bolt spaceings and 2 types of imperfections
a) b)
range 0-500MPa) for peak loads: a) bolt spacing 416.67mm and 625mm; b) bolt spacing
1000mm
50
a) b)
range 0-500MPa) for the post-buckling range of loads: a) bolt spacing 416.67mm and
625mm; b) bolt spacing 1000mm
2.3.3.7 Segment Tests

The following paragraphs study in more detail the effect on the strength of steel tower of door
opening and respective stiffening countermeasure like thickness increase or stiffeners. The
parameters taken into account in this study correspond to real designs situation of steel wind
turbine towers. Concerning the analysis methodologies, firstly, linear perturbation analysis of
models have been performed in order to get elastic buckling modes and eigenvalues in ’perfect’
geometry. Secondarily, nonlinear analysis of models with imperfection geometry from buckling
mode has been considered. Comparative studies based on varying parameters and steel grade
were also carried out and included in these paragraphs.
Finite element model
The lower segment of steel tower of real wind turbine structure was modelled using Abaqus. Three
types of FE models of the cylindrical shell were considered: model without opening, models with
door opening surrounded by thicker steel shell (models with varying thickness) and models with
door opening surrounded by welded stiffener (models with stiffener). Geometries of the tower
segment are as follows: 6666 mm of height and diameters of 4150 mm on the bottom and 3919
mm on the top. Thickness of wall of models is 37 mm. Thickness of door opening and stiffener
range from 37 mm to 60 mm. Details of geometries are showed in Figure 2.40. Boundary
conditions were determined according to EN1993-1-6 and referred to case of open tank with
anchors: BC3 (radially - W, meridionally - U and rotation - Rφ are free) and BC1f (radially - W and
meridionally - U are restrained and rotation - Rφ is free) on the top and bottom of the structure
respectively, (see Figure 2.41). External load was applied on the model through a reference point
[16] coupled with the cross section on the top of model.
51
Figure 2.40: Geometry of Models.
The commercial finite element analysis program Abaqus [16] was used in the analyses of the
models. S4R (shell element with 4 nodes and reduced numerical integration is appropriate for large
strain of buckling and riks analyses. The quality and symmetry of mesh was especially considered
in order to get more accurate results. The FE mesh was divided in two parts: the wall part and the
door opening part including varying thickness or stiffener, (see Figure 2.41).
Figure 2.41: Boundary conditions.
A total of 35 FE models of lower segments of steel tower were analysed and compared in the
parametric studies. Models with thickness of door openings or stiffener ranging from 37 mm to 60
mm were analysed. Thickness of tower wall was kept constant equal to 37 mm and the total height
was kept at 6666 mm. In order to compare results, the model without door opening has the same
height and 37 mm of wall thickness. All the models had the same diameters. In addition,
characteristics of Steel S500 and S650 were considered for comparative analyses.
Results and discussion
Figure 2.42 presents the results of the first buckling mode of the models. Buckling mode shape of
the model without door opening is characterized by local buckling waves. The magnitude of
deformations decreases gradually from top to bottom of the model. In model with varying
thickness and model with stiffener, buckling occurs at door opening with different deformations.
52
Figure 2.42: First buckling mode of the models: (a) Model without door opening, (b)
Model of a door opening with varying thickness, (c) Model of a door opening with
stiffener.
The eigenvalues presented in Table 2-23 increase gradually among modes of each model. It is
interesting to emphasize that the first eigenvalue of model with varying thickness and of the model
with stiffener decrease to 47.6 % and 69.8 % respectively in comparison with the model without
door opening. Eigenvalues of model with varying thickness are smaller than model with stiffener
(68.2 % of first mode and 74.2 % of third mode).
Table 2-23: Comparison of eigenvalues between models.
Model withoutModel of a door openingModel of a door opening

door opening with varying thickness with stiffener
Mode 1 -1.000 0.476 0.698

Mode 2 1.000 0.487 0.712
Mode 3 1.002 0.580 0.782
After performing the buckling analysis, geometrically and materially nonlinear analyses with
imperfection modes (GMNIA) were carried out in order to predict ultimate strength of the structure.
Figure 2.43 shows moment-rotation curves of models with the same primary input parameters, in
which, the half of tower with door opening is under compression. Thickness of both door opening
and stiffener is 60 mm. These values are close to real for design of wind towers. The curves consist
of two parts: the linear behaviour before buckling point and nonlinear behaviour in post-buckling.
In first part, three curves are linear and almost coincident. In second part, the curves behave
nonlinear up to ultimate strengths of 175.7 MNm, 193 MNm and 189.7 MNm corresponding
respectively to model without door opening, model with varying thickness and model with stiffener
respectively. In conclusion, values of ultimate strength show that both stiffening solutions are
sufficient to strengthen the structure. However the big differences in the ultimate strengths justify
further studies to optimize stiffening solutions.
53
Figure 2.43: Relation between moment and rotation of models - door opening under
compression.
Figure 2.44 presents moment-rotation relationship of models which have the half of tower with
door opening under tension. Ultimate strength of model with varying thickness and model with
stiffener are 194.9 MNm and 192.0 MNm, respectively. In comparison with models that have a half
of tower with door opening under compression, the difference is just 1%. It may lead toconclusion
that the position of a door opening (under compression or tension) does not affect the ultimate
strength of towers with the same conditions.
Figure 2.44: Relation between moment and rotation of models - door opening under
tension.
Figure 2.45 presents the distribution of normal stresses at ultimate load in all models. Distribution
of normal stresses in the model with varying thickness shows big difference between wall area and
door opening area. Stress concentration occurs around door opening. Normal stresses in tension
and compression in the model with varying thickness are 382.4 MPa and -457.0 MPa respectively.
As mentioned above, the ultimate load of model without door opening is lowest (175.7 MNm).
However, its normal stress in tension is the highest (412.6 MPa). Besides, that normal stress in
compression of model with stiffener is at minimum (-495.8 MPa). It should to be noted that Steel
54
S355 was used in analyses of these models. It proves that yield phenomenon appeared in parts of
the models.
Figure 2.45: Distribution of normal stress at ultimate load: (a) Model without door
opening, (b) Model of a door opening with varying thickness, (c) Model of a door opening
with stiffener.
Figure 2.46 shows distribution of reaction forces of the models for 0.0002 rad rotation applied on
the top of the top of the models. The three curves almost coincide. They are typical shapes of
distribution of reaction forces in linear part. It is interesting to emphasize effects of door opening
on distribution of reaction forces. Outside the area of door opening, the curves of reaction forces
are smooth. The results also show that distribution of reaction forces of models is approximate in
pre-buckling part.
Figure 2.46: Distribution of reaction forces of the models at 0.0002 rad of rotation.
Figure 2.47 shows distribution of reaction forces of the models corresponding to 0.013 rad rotation
applied on the top of the models. It should be noted that the figure presents distribution of reaction
forces in nonlinear part. The curves of model with varying thickness and model with stiffener are
almost coincident. The iInfluence of buckling on the distribution of reaction forces is significant.
Minimum reaction forces at door opening area of the model with varying thickness and the model
with stiffener are 921.6 kN and 895.1 kN respectively. However, at the same position, reaction
force of model without opening is only 242.2 kN.
55
Optimal designs of door opening and stiffener
In this study, varying thickness around door opening and of stiffener were considered. These
results were analysed and compared with the results of model without opening in order to op-
timize the thickness of door opening and stiffener as well. The thickness ranges from 37 mm to 60
mm. These values are appropriate to design the stiffening of door opening. Table 2-24 presents
results of relations between thickness, moment and rotation at ultimate load of the models. As
mentioned above, the ultimate moment of model without opening is 175.7 MNm. In comparative
studies of the door opening with varying thickness and of the door opening with stiffener, 51 mm
of thickness of door opening and 37 mm of stiffener are sufficient to strengthen the structures.
Table 2-24: Relation between thickness, moment and rotation at ultimate load of the
models.
Door opening Stiffener
Thickness Moment Rotation Moment Rotation
(mm) (MNm) (rad) (MNm) (rad)
37 163.910 0.01464 182.203 0.01357

39 163.092 0.01869 182.920 0.01357
41 163.531 0.01186 183.772 0.01088
43 165.130 0.01186 184.107 0.01357
45 167.618 0.01015 184.668 0.01357
47 170.473 0.01015 185.252 0.01357
49 173.347 0.01015 185.864 0.01357
51 176.392 0.01015 186.631 0.01186
53 179.654 0.01015 187.703 0.01186
55 183.084 0.01015 188.555 0.01186
57 187.160 0.01037 189.078 0.01186
59 190.870 0.01079 189.520 0.01186
60 192.976 0.01058 189.719 0.01186
56
Use of higher strength steels
In practice, design of wind towers is based on Steel S355. In this research, higher strength steels
S500 and S650 were used for door opening and stiffener in order to study decrease of thickness
without affecting the strength of towers.
Figure 2.48 presents moment-rotation curves of the tower when steel S355, S500 and S650 were
used for the stiffener. The curves are similar in shape. Ultimate load is approximately 182 MNm.
However, the thickness of stiffener significantly decreased as higher strength steels were used. It
should be noted that in case of steel S355, the thickness of stiffener is 37 mm. When S500 and
S650 are used, the thickness of stiffener decreases to 29 mm and 23 mm respectively. These
correspond to approximate to 78% and 62% of the thickness of stiffener when steel S355 is used.
Figure 2.49 shows curves of moment versus rotation of the tower as Steel S355, S500 and S650
were used for the door opening. The shape of curves changed significantly as higher strength
steels were used. Ultimate load is approximately 177 MNm. However, difference between pre-
buckling and post-buckling is detected. The thickness of door opening significantly decreasedas
higher strength steels were used. It decreased to 41 mm and 35 mm for steel grades S500 and
S650 respectively. It decreased approximately 20% and 33% in comparison to the thickness of
door opening (51 mm) when steel S355 was used.
Figure 2.48: Comparison of using higher grade steels for stiffener.
57
Figure 2.49: Comparison of using higher grade steels for door opening.
Conclusions
In this paper, the influence of door opening and respective stiffening were investigated.
Comparative studies between model without opening, models with door opening surrounded by
thicker steel shell and models with door opening surrounded by welded stiffener were presented.
Linear and nonlinear moment-rotation behaviours of the thin walled structures were analysed.
Results from comparative studies in cases of door opening under compression and tension showed
that the position of door opening did not affect significantly the ultimate strength of tower (1% of
difference) in the same conditions. Parametric study aiming at optimal designs of door opening
and stiffener showed that 37 mm of stiffener and 51 mm of thickness around door opening are
sufficient to obtain the same resistance in comparison with the model without door opening and
37 mm of wall thickness. In addition, results from studies when higher strength steels are used,
have shown that thickness of door opening and stiffener decreased significantly when higher
strength steels are used. Thickness of door opening just equalled approximately 80% and 67% of
its value as steel S355 was used. Thickness of stiffener decreased by 22% and 38% as steel S500
and S650 were used for stiffener respectively.
2.3.4 Transversal friction connections

The transverse connection of two segments of a wind turbine using prestressed friction connections
comprises fingers and slotted holes, see Figure 2.50. The inner segment has to be produced with a
smaller diameter in order to facilitate the assembling and thus the fingers have to be bent during
the prestressing. The bending of the fingers can cause either elastic or plastic deformations
depending on the fingers length/thickness and the size of the assembling gap. In a further (global)
bending of the wind turbine, the fingers can buckle under compression loads. Although bending
tests have already been performed in the HISTWIN 1 project [1], a formula for a safe design of the
fingers is still missing.
Figure 2.50: Prestressed friction connection with fingers and slotted holes
2.3.4.1 Finger tests

Small scale buckling tests on prestressed-bolted transvers connections with fingers and long slotted
holes of wind turbines have been performed. To evaluate the tests, they have been simulated using
the finite element programme Abaqus. The numeric model was used to perform a parameter study
and to widen the range of parameters under investigation. During the closing of the gap between
adjacent tower segments, the fingers are bended and deform depending on their length either
elastic, plastic or partially plastic. It could be seen, that this initial stress state affects not only the
load bearing behaviour but also the sensitivity to the gap width. A recommendation for the
analytical design according to EC 3 could be derived.
Principal test set-up
In a real tower, the bending moment in the cross section will cause a nearly constant membrane
stress in the fingers of the compression zone. During the buckling of a single finger, the centres of
gravity of the adjacent tower segments will remain on each other and the load introduction point of
58
the finger will not change. These assumptions correspond to a displacement controlled load
introduction and a buckling mode as depicted in Figure 2.51. This type of load introduction can be
achieved by restraining the movement of the test specimen with a linear bearing.
Figure 2.51: Loading of the fingers in wind turbines (left) and static system of the
fingers (right)
The test set up is shown in Figure 2.52. The test specimen consists of a finger and an adapter
plate. This adapter plate can be bolted to four linear bearings which enable vertical displacements
of the test specimen. The linear bearings are attached to a substructure (Pos. 1). The test
specimen will be fixed to the substructure at its lower end with a prestressed bolted connection
using an adapter plate (Pos. 2). This adapter plate can be changed in order to test different finger
widths and number of fingers. In addition, spacer plates can be introduced between adapter plate
and substructure in order to control the gap size . Further filler plates are used to obtain a
symmetric connection between finger and adapter (Pos. 3). These filler plates have the same width
and thickness like the fingers and are used to prevent a bending of the bolts.
Figure 2.52: General Test set up
59
Figure 2.53 shows the test set up of the small scale tests. Due to the high number of tests, the
measurement instrumentation was kept to a minimum in order to facilitate for a fast conduction of
the tests. To assess the overall behaviour of the test specimen, the stroke load and the vertical
displacement of the test specimen at the bottom of the plate was measured. The horizontal
displacement of the finger was measured with three lvdts and one lvdt at the bottom of the finger
controlled the slip in the friction connection.
Figure 2.53: Test set up of the small scale buckling tests
A preliminary finite element analysis was conducted with ANSYS Workbench 13.0. The results were
used to identify influences of different parameters qualitatively and to get an impression of the
bearing behaviour of the fingers.
The following parameters were studied:
- The finger width (50 and 75 mm)
- The finger thickness (10 and 15 mm)
- The gap size (10 and 15 mm)
- The steel grade (S 355 and S 460)
- The influence of plastic deformation during closing of the gap (controlled by the length)
- The number of fingers
- The finger length as a dependent parameter
The length of the finger was varied as a dependent parameter, i.e. the finger length was calculated
so that the elastic or plastic moment was reached at the end of the fingers during closing the gap
(Figure 2.54). This causes for example a shorter finger length for fingers with the same width and
thickness when an S 460 is used instead of an S 355. If a finger is designed such that the elastic
moment is reached after closing the gap, the length of the finger is called “lElastic”. If the plastic
moment is reached the length is called “lPlastic”. From the dependency of the finger length of the
other parameters follows, that not only single parameters can be varied but always a set of
parameters.
60
Figure 2.54: Closing of the gap (left); moment distribution in the finger (right)
The tests were performed with two steel grades, S 355 and S 460. It was assumed that the
number of fingers and the finger width have a linear influence on the bearing. For that reason the
width and the number of fingers were only be varied for the steel grade of S 355. For each test
series 2 tests were carried out. The test programme is shown in Table 2-25
Table 2-25: Performed tests.
Name Width Thickness Gap size Steel No. of

series Length
[mm] [mm] [mm] strength fingers
a 50 10 441 10 S 355 1
b 50 10 441 10 S 355 1
c 50 10 441 10 S 355 1
d 50 10 421 10 S 355 1
e 50 10 391 10 S 355 1
1
f 50 10 361 10 S 355 1
g 50 10 331 10 S 355 1
h 50 10 301 10 S 355 1
i 50 10 271 10 S 355 1
j 50 10 241 10 S 355 1
a 75 10 419 10 S 355 1
2
b 75 10 389 10 S 355 1
a 50 15 516 10 S 355 1
3
b 50 15 486 10 S 355 1
a 50 10 516 15 S 355 1
4
b 50 10 486 15 S 355 1
a 50 15 632 15 S 355 1
5
b 50 15 602 15 S 355 1
a 50 10 419 10 S 355 3
6
b 50 10 389 10 S 355 3
8 a 50 10 344 10 S 355 1
61
b 50 10 314 10 S 355 1
a 75 10 344 10 S 355 1
9
b 75 10 314 10 S 355 1
a 50 15 421 10 S 355 1
10
b 50 15 391 10 S 355 1
a 50 10 421 15 S 355 1
11
b 50 10 391 15 S 355 1
a 50 15 516 15 S 355 1
12
b 50 15 486 15 S 355 1
a 50 10 344 10 S 355 3
13
b 50 10 314 10 S 355 3
a 50 10 390 10 S 460 1
14
b 50 10 390 10 S 460 1
a 50 15 453 10 S 460 1
15
b 50 15 423 10 S 460 1
a 50 10 453 15 S 460 1
16
b 50 10 423 15 S 460 1
a 50 15 555 15 S 460 1
17
b 50 15 525 15 S 460 1
a 50 10 322 10 S 460 1
18
b 50 10 272 10 S 460 1
a 50 15 370 10 S 460 1
19
b 50 15 340 10 S 460 1
a 50 10 370 15 S 460 1
20
b 50 10 340 15 S 460 1
a 50 15 453 15 S 460 1
21
b 50 15 423 15 S 460 1
Sum: 48
Normally, in a parameter study, the parameter under investigation is changed while all the other
parameters are kept fix. In this case, this approach causes the problem that a fixed length for
instance will cause different bending moments (and hence initial stress states) if the thickness is
varied. For that reason, the length was changed in some cases to cause the same bending moment
in the finger after closing the gap. In some cases, however, the length was kept fixed to see how
the ultimate load would change if, for example, the gap would change due to assembly tolerances.
The results of the tests and the finite element calculations were evaluated altogether using the
buckling analysis procedure of EC 3.
62
Influence of the kind of deformation during closing the gap
During the assembly the bolts are tightened to close the gap. In this process the finger deform and
its length determines whether the deformation will remain elastic or plastic, see Figure 2.55. In
order to save material, a short finger desirable.
The tests were modeled and tested with different lengths of the fingers in order to obtain elastic or
plastic deformations during closing the gaps. The results for two simulations with different lengths
are depicted in Figure 2.55. The necessary length to avoid plastic deformation in the finger is
377 mm (when MEl is reached) and the deformation of a finger with a length of 308 mm is fully
plastic. The load bearing behavior differs significantly between the finger, for which MEl had not
already reached and the other specimen where MPl was already exceeded. The load-displacement
curve indicates that a finger, for which MEl is not already reached during closing the gap behaves
more in a column-like (i.e. it buckles) manner whereas the shorter finger does not show such an
abrupt loss of load. Anyway, the increase in load is only approximately 6 % although the smaller
length would imply a significantly higher bearing load. This is due to the assumed change in the
bearing behavior of an already plasticized finger.
Figure 2.55: Load displacement curves for tests with elastic and plastic deformations
Figure 2.56: Ultimate load over length of the finger for the test series 1 and 8 and
corresponding Abaqus simulations
63
Figure 2.56 summarizes all tests and simulations with the same thickness, width and gap size. It
has to be noticed that the length of the finger seems to have a linear influence on the ultimate
load. This is surprising as the length is a parameter which was considered to influence the results
in a quadratic function.
Influence of the gap size
In a real tower, the gap size can vary due to assembly tolerances or due to the tightening process
when one side of the tower is tightened before the other. For that reason, the gap size was varied
for fixed lengths of the finger. The results of the simulations for two different finger lengths with
different gaps are summarized in Figure 2.57 and Figure 2.58.
In this example the following parameters are compared, see Table 2-26:
Table 2-26: Compared tests.
Width Thickness Gap size Steel No. of

Length
5, 10,
Figure 2.57 50 10 350 S 355 1
15
5, 10,
Figure 2.58 50 10 500 S 355 1
15
By a comparison of the results, it can be observed that the gap size can have a strong influence on
the ultimate load. This can be traced back to the influence of the deformation during closing the
gap. Table 2-27 lists the length of the finger where MEl or MPl is reached during closing the gap for
the given dimensions of the fingers.
Table 2-27: Elastic and plastic length for the different gap sizes
Elastic Plastic
gap size
length length
[mm]
[mm] [mm]
5 271 221
10 383 313
15 469 383
The series in Figure 2.57 has a fixed length of 350 mm. Hence, the finger is longer than lElastic for a
gap size of 5 mm, between lElastic and lPlastic for a gap size of 10 mm and shorter than lPlastic for the
gap size of 15 mm. That means, an increase of the gap from 5 to 10 mm during the assembly
would in this case cause a decrease of the ultimate load of about 15 %.
64
Figure 2.57: Load displacement curves for tests with different gaps and fixed length of
350 mm
The results of the series with a length of 500 mm, see Figure 2.58, indicate a smaller influence of
the gap size on the ultimate load. This is due to the fact that in all calculations the length of the
finger exceeds lElastic, which means that the ends of the finger did not deform plastic yet and a
pure buckling occurs. With an increase of the gap from 5 to 10 mm the ultimate load only
decreases by 4 %. In this case, the structure would be very insensitive to assembly tolerances.
Figure 2.58: Load displacement curves for tests with different gaps and fixed length of
500 mm
The results demonstrate that the gap size is a crucial parameter if the transversal connection is
designed with short fingers. Although the ultimate load of a shorter finger will increase, compared
to a longer finger, the structure will be very sensible to imperfections and assembly tolerances. A
further disadvantage that arises from shorter fingers is the higher loss of pretensioning of the bolts
during closing the gap and possible plastic deformation, which can affect the fatigue strength of the
structure.
Influence of the thickness of the fingers
In the design of a wind-energy tower, the thickness of the structure –and hence the fingers- is
determined in the global analysis, i.e. stability, stiffness and fatigue design. For that reason, the
65
thickness is a given parameter in the design of the connection. The load bearing behavior of fingers
with different thicknesses does not differ qualitatively. However, one has to keep in mind that a
higher thickness means a higher stiffness of the finger and demands therefore a greater length if
plastic deformations shall be avoided.
In this chapter, the following parameters are compared, see Table 2-28:
Table 2-28: Compared tests
Width Thickness Gap size Steel No. of

Length
Figure 2.59 50 10, 15, 20 350 10 S 355 1
Figure 2.60 50 10, 15, 20 500 10 S 355 1
The fingers were modeled with thicknesses of 10, 15 and 20 mm. Table 2-29 lists the length of the
finger where MEl or MPl is reached during closing the gap for the given dimensions of the fingers and
the measured steel strength.
Table 2-29: Elastic and plastic length for the different thicknesses
Elastic Plastic
thickness
length length
[mm]
[mm] [mm]
10 383 313
15 469 383
20 542 442
The results for three simulations with different thicknesses are depicted in Figure 2.59. In this case
the length was not changed, which means that all fingers have the same total length of 350 mm.
For the length of 350 mm, the finger with a thickness of 10 mm is in the transition area between
lElastic and lPlastic whereas the thicker fingers are shorter than lPlastic. This explains the different
behavior of the three tests with the more buckling like loss in load of specimen L350-T10 and the
other two specimens.
Figure 2.59: Load displacement curves for tests with a thickness of 10, 15 and 20 mm
with a fixed length
66
For a length of 500 mm the fingers with thicknesses of 10 and 15 mm are longer as lElastic and the
finger with the thickness of 20 mm now is in the transition are between lElastic and lPlastic. Again, one
can notice the different load bearing behaviour of the two configurations.
Figure 2.60: Load displacement curves for tests with a thickness of 10, 15 and 20 mm
with a fixed length of 500 mm
The finger´s thickness has a major influence on the bearing behaviour of the finger. Although the
load bearing behaviour does not change qualitatively with changing thicknesses, one has to keep in
mind that a thicker finger necessitates a longer finger to keep the deformation in an elastic range
during closing the gap. Also, a thicker finger will increase the necessary force to bend the finger to
the lower section thus increasing the loss of pretensioning.
Influence of the width of the fingers
The fingers were modeled and tested with two different widths of 50 and 75 mm. The width of the
finger has a linear influence on the bearing behavior. The following series were used to compare
the influence of the width of the finger, see Table 2-30.
Table 2-30: Tests to compare influence of finger width
Specimen Width Thickness Gap size Steel No. of

Series
1 d, e 50 10 10 S 355 1
2 a, b 75 10 10 S 355 1
8 a, b 50 10 10 S 355 1
9 a, b 75 10 10 S 355 1
The average increase in ultimate between series 1 and 2 was 1.54 and between series 8 and 9 was
1.44. This corresponds to an increase of the thickness of 1.5.
Influence of the number of fingers
The test specimens were modeled with different numbers of fingers. The number of the finger has
a linear influence on the bearing behavior. In this case a number of 3 fingers leads to an increase
in ultimate load of 300 %. As the fingers have the same stiffness, the fingers have the same
length.
67
Influence of the steel grade
The test specimens were modeled and tested with different steel grades. The steel grade dos not
change the qualitative behavior of the fingers. An increase of the steel strength leads to a higher
elastic moment and thus to shorter finger length. This of course changes the slenderness
positively.
2.3.4.2 Derivation of modified buckling curve

The tests were evaluated using the design approach of EC 3-1-1 for the design against buckling.
The aim was to derive imperfection factors and to establish a buckling curve for a safe design of
the fingers. As the bearing behaviour of the fingers changes depending on the state of
deformation, the tests were divided into three classes:
 Elastic deformation during closing the gap (l>lElastic)
 Full plastification of the finger during closing the gap (l<lPlastic)
 Transition area (lPlastic < l < lElastic)
Buckling curve for fingers with elastic deformation only during

closing the gap
As no plastic deformation occurs during closing the gap, the buckling length can be calculated
under the assumption of a double-sided clamped beam. Hence, a  factor of 0.5 was used to
calculate the buckling length. The plastic resistance of the cross section is calculated without safety
factors and using real material properties gained from material testing. As the bending of the finger
causes high initial stresses, the resistance of the cross section has to be reduced by the reduction
factor . Therefore, the imperfection factor  has to be fitted to gain good correspondence between
test results and analytical evaluation. The parameter fitting resulted in an  value of 0.7. The
results of the buckling analysis are visualized in Figure 2.61 by means of utilization ratios of
ultimate load to design load. Values higher than 1 denote a safe design as the experimental load is
higher than the characteristic (calculated) load. The experimental data scatters only between 1 and
1.3 although a high number of parameters were tested. The blue markings indicate the test results
and the red dots the results of the finite element calculations.
Figure 2.61: Utilization ratio of the tests and Abaqus calculations over slenderness with
only elastic deformation during closing the gap
It can be seen that the tests were performed in a range of slenderness of l=0,8-1,3. This was due
to the fact that the specimens were designed to be close to MEl or MPl after closing the gap. In order
to broaden the field of investigated slenderness´s, finite element calculations were performed with
a higher and lower slenderness (red dots).
68
Buckling curve for fingers with plastic deformation during closing
the gap
As explained in the background document of these tess, the plastic deformation during closing the
gap causes plastic hinges on both ends of the finger. During the subsequent buckling and the
change in curvature, one plastic hinge has to be closed whilst the opposite hinge has to be opened
further. For that reason, it is assumed that the static system behaves like a one side clamped and
one side pin ended beam. The b factor for the calculation of the buckling length for this case is
0.75.
The imperfection factor a remains unchanged with 0.7. The results of the buckling analysis are
visualized in Figure 2.62 by means of utilization ratios of ultimate load to design load. Again, values
higher than 1 denote a safe design as the experimental load is higher than the characteristic load.
The experimental data scatters on the safe side between 1.3 and 1.6. This is due to the fact that
the buckling length is smaller than 0.75∙ℓ as the “pin-ended” side of the finger has still a resistance
due to hardening. A reduction of the buckling length, however, is not recommended as the short
fingers react very sensitive on changes of the gap size. In addition, one should consider if plastic
deformation really should be allowed in a wind-turbine as the fatigue behaviour of the finger was
not part of the investigations.
Figure 2.62: Utilization ratio of the tests and Abaqus calculations over slenderness for
fingers with plastic deformation during closing the gap
Buckling curve for fingers with partially plastic deformation during

closing the gap
For fingers with lengths in between lElastic and lPlastic the b factor to calculate the buckling length has
to lie between 0.5 and 0.75. A b factor of 0.6 gained good results for the evaluation of the tests
using the unchanged imperfection factor a of 0.7. The results of the buckling analysis are
visualized in Figure 2.63 by means of utilization ratios of ultimate load to design load. Again, values
higher than 1 denote a safe design as the experimental load is higher than the characteristic load.
The experimental data scatters on the safe side between 1.05 and 1.4 meaning better agreement
with the analytical analysis than the fully plastic deformed fingers.
It has to be stressed that the short fingers react very sensitive to changes of the gap size. In
addition, one should consider if partially plastic deformation should be allowed in a wind-turbine as
the fatigue behaviour of the finger was not part of the investigations.
69
Figure 2.63: Utilization ratio of the tests and Abaqus calculations over slenderness for
fingers with plastic deformation during closing the gap
The buckling curve is depicted in Figure 2.64 to illustrate the range of slenderness under
investigation. 48 tests were performed on fingers with slenderness’s between 0.75 and 1.3. The
tests were complemented by finite element simulation to facilitate a quick and safe design in the
range of application.
Figure 2.64: Buckling curve and tested slenderness’s
2.3.4.3 FE analysis of transversal friction connections for circular and polygonal

wind turbine towers
Down-scale tests of friction connections in wind tower structures have been conducted previously
[1], but still no tests or field measurements on pilot structures are available with real dimensions of
a wind tower with common height of 100 m or above. Diameters of such towers are large e.g. 3000
– 4000 mm and thicknesses of shells can go up to 40 mm, depending on the strength of steel that
is used. It is hard to expect that tests up to failure of such large specimens, even on short
segments of tower adjacent to the connection, are ever to be realised. FEA of one bolt row
segments and down-scale friction connections have also been conducted previously [17].
Behaviour at failure of friction connections in wind tower structure of real dimensions is considered
in this study using beyond-state-of-the-art Finite Element Analysis (FEA). Behaviour of friction
connection is compared to the ring flange connections of the tower segment with same dimensions
as shown in Figure 2.65. Diameter of the shell of 3374 mm (steel grade S460) with thickness of 24
70
mm and high-strength bolts M48 (grade 10.9) are used in both cases. Dimensions of the tower
segment, as well as the design and fatigue loads are taken from a real design of a tower with
classical ring flange connection, provided by Vestas.
a) friction connection b) ring flange

connection
Figure 2.65: Layout of the tower segment and connection considered in the study.
In order to provide verification for the FE modelling technique used for the real tower dimensions,
FE models of down-scale four point bending tests have been made. Results from [18] are used
where friction and ring flange connections with bolts M20 and shell diameter of 1000 mm (8 mm
thickness) were tested.
New approaches of polygonal tower shapes have been studied recently [19]. Therefore, friction
connection is here analysed in a parametric study, as well, considering: different shapes of tower
shells (circular and polygonal), assembling tolerances (gap between the shells) and length of
fingers. Moreover, an alternative method for providing assembling tolerances by pre-bending of
fingers is proposed and examined.
Finite element models
Different cases analysed in this study are shown in Table 2-31. Parameters that are varied are:
type of the connection, shape of the tower shell, length of the fingers, gaps between the shells and
imperfections and different methods to provide assembling tolerances.
71
Table 2-31: Cases considered in the study.
Tower Gap between Length of the Number Dimple

shape the shells (mm) fingers (mm) of the bolts imp. (mm)
Ring flange 0
circular - - 96
connection 13
0 450
10 450
circular
10 550 3x56=168 0
10 650
Friction 0 (bended fingers) 450
connection 0 450
10 450
polygonal 10 550 3x4x14=168 0
10 650
0 (bended fingers) 450
Friction connection is analysed in two cases of tower shapes (see Figure 2.66): circular shape and
polygonal shape and compared to each other. The polygonal tower shape is composed out of 14
segments. The dominant criterion for such division was to avoid the local buckling of the segments,
i.e. keeping the slenderness of the segment close to the limit for the cross section of class 3 (see
Eq. 1 and Eq. 2), according to EN 1993-1-1.
cseg / t  42  cseg  42  0.71 24  716 mm 1

D
nseg   14.7 2
cseg
a) circular tower shape b) polygonal tower shape
Figure 2.66: Different tower shapes considered for the friction connection.
Imperfections of the shells are introduced both to the models of ring flange and friction
connections, but in different ways. The aim was to obtain comparable results. The case with the
ring flange connection was analysed as perfect (without imperfections) and with dimple
imperfection in the compression zone of the shell in order to be comparable with the friction
connections with gaps between the shells. Dimple imperfection of 13 mm was calculated according
to EN1993-1-6 with fabrication tolerance quality class B. The cases with friction connection were
analysed without dimple imperfection (no welding) but with gaps between the shells, as shown in
Table 2-31 and Figure 2.89.
Apart from the initially analysed cases with perfect fitting of diameters of outer and inner shell (see
Figure 2.89a), two methods for provision of assembling tolerances for friction connection are
considered and their influence of the behaviour of the connection is assessed. The first case is the
classical solution with the gap between the outer and inner shell, as shown in Figure 2.89b). The
second one is an alternative solution with pre-bending of the fingers on the inner shell and no gap
between the outer and inner shell, as shown in Figure 2.89c). The first case was analysed with the
gap between the shells of 10 mm while the length of the fingers were varied in a parametric study.
In the second case, fingers were assumed to be pre-bended 10 mm at the top of the fingers by
heating (see Figure 2.89c).
72
a) perfect case, no gap b) 10 mm gap between shells c) bended fingers, no gap between
shells
Figure 2.67: Different approaches for provision of assembling tolerances.
a) LF = 450 mm b) LF = 550 mm c) LF = 650 mm

Figure 2.68: Different lengths of the fingers in the case of circular tower shape with gap.
Gaps between the shells and length of the fingers in the case of friction connection were considered
in different models as it is show in Table 2-31 and Figure 2.68. In the case of friction connection
with polygonal tower shape in all cases with the gap between the shells, indentions were made at
the corners to allow the unrestrained deformation of the fingers (see Figure 2.69). The width of
those indentions is 24 mm, while the length is 50 mm longer then the fingers. No indentions are
made in the initial case (perfect case) without the gap between the shells. Also, in the case of
circular tower shape no indentions are needed. Overlapping lengths of the friction connections are
50 mm longer than the fingers in all cases. Distances between the bolts are 150 mm, while the
cover plate is 12 mm thick, 150 mm wide and 450 mm high. The width of the slotted hole between
the fingers is 56 mm (see Figure 2.65a).
73
a) no gap – LF = 450 mm b) 10 mm gap – LF = 450 mm
Figure 2.69: Indentions at the corners of the polygonal tower shape with gap between
the shells.
Geometry, boundary conditions and mesh
An example of the model geometry and boundary conditions is shown in Figure 2.70a). It consists
of 180° segment of the connection, modelled in symmetry boundary conditions for bending. The
segment length of 3000 mm on each side of the connection was modelled in order to avoid over-
constraining of the model in means of buckling and redistribution of the stresses in the shells.
Upper and lower surfaces of the shells are fully kinematically constrained to reference points.
Appropriate rotations were applied to those reference points to define the bending loading up to
failure. The upper reference point was released for the vertical translation while all the other
degrees of freedom were restrained. Free side surfaces of the shells and flanges are assigned with
symmetry boundary conditions.
Realistic geometry of the bolt and nut are shown in Figure 2.70b). Bolts and nuts are modelled with
threads in order to achieve preloading of the bolts by turn-of-nut method. Tetrahedron solid
elements (C3D4) are used to form the mesh. Global element size is 11 mm, while in the zone of
threads it is reduced to 5.5 mm.
Eight node hexahedron solid elements with reduced integration (C3D8R) were used for the shells,
cover plates and flanges mesh, since they offer good results for a reasonable calculation time.
Global element size of 20 mm has been used for shells and cover plates with four elements through
the thickness in order to properly account for bending stiffness (see Figure 2.69).
General contact interaction procedure was used in ABAQUS/Explicit with “Hard” formulation of
normal behaviour and “Penalty” friction formulation of tangential behaviour for the most contact
pairs in the model. Friction coefficient of 0.14 was set for bolts and nuts threads surface pairs, 0.3
for nuts and washers surface pairs and 0.45 for all other surface pairs in the model, except the
friction connection interface between the two shells which is assumed to have treated surfaces. For
this contact pair, “Static-Kinetic Exponential Decay” friction formulation is used with 0.60 static
friction, 0.55 friction at slip rate of 0.15 mm/s and 0.5 friction at the infinite slip rate. Those values
are based on measurements and experience from the verification FEA of down-scaled tests for
friction connection made in RWTH Aachen, using the same modelling technique as presented here.
74
upper reference point
Ux=Uz=Rx=Ry=0
Z symmetry BC
Z symmetry BC
lower reference point

Ux=Uy=Uz=Rx=Ry=0
a) boundary conditions b) geometry and mesh of the

bolts
Figure 2.70: Geometry and boundary conditions.
Loading and analysis method
Two loading steps were analysed for each model: preloading of the bolts and bending loading up to
failure. Preloading of the bolts was applied with turn-of-nut method. Hexagon edges of nuts are
coupled to the reference points in the centre line of each nut, as it is shown in Figure 2.94a). Those
reference points are turned by applying changes in boundary conditions as rotations around axis
parallel to the shank of each bolt. Preloading force of 960 kN was calculated according to EN1993-
1-8 for M48 bolts, grade 10.9.
Rotations of the nuts were calibrated for each case shown in Table 2-31 to achieve the desired
preloading force. Values of achieved preloading forces that are obtained by integrating node forces
in cross section of the bolt are shown in Table 2-32. Stresses in the bolt, shells and cover plate in
the case of friction connection with circular tower shape, are shown in Figure 2.94b). In the case of
circular tower shape larger rotation of the nut is needed when compared to the case with polygonal
tower shape (see Table 2-32) in order to bend the cover plate and achieve its contact with the
inner shell.
75
Table 2-32: Rotations of the nuts and achieved preloading forces.
Grip Gap Nut Achieved

length (mm) rotation preloading
(mm) (rad) force (kN)
Ring flange 170 0 1.00 939
connection
Friction connection 60 0 1.00 905
(polygonal tower 60 10 13.60 964
shape)
Friction connection 60 0 3.83 960
(circular tower shape) 60 10 16.40 1024
Loading up to failure was applied as displacement controlled by rotations at upper and lower
reference points of the shells (see Figure 2.70a).
a) the turn-of-nut method b) stresses in the bolt, shells and cover plate
Figure 2.71: Preloading of the bolts.
ABAQUS/Explicit dynamic solver was used to solve FE models presented here. Explicit solver is
known to be robust for those kinds of analyses incorporating complicated contact interactions
coupled with large deformations and nonlinear behaviour. It does not have usual convergence
issues such as Standard static solver. On the other hand, analyses can last for certain long period
of time if the real dynamic response of model is required. Loading in this model, and tests as well,
is supposed to be quasi-static. Dynamic solvers can be used to efficiently solve quasi-static
problems by using either time or mass scaling techniques. Variable non-uniform mass scaling
technique, offered in ABAQUS was used as it has large advantages when compared to usual time
scaling method. Desired time increment was set to a value of t = 5.0 x10-5 s. Artificial step
durations were adopted for the loading steps: preloading - 5 s and loading up to failure – 10 s.
Care must be taken in selection of desired time increment to avoid significant inertia influences on
analysis results. Smooth amplitude functions were used for all loading steps (rotations of reference
points of nuts and shells) to roll out the impact behaviour and excitation of the model with
unnecessary inertia forces.
Material models
Nominal material properties were used for the bolts and shells while measured data was used for
the flange (grade S355) because of the reduction of strength due to its large thickness. Isotropic
plasticity with initial modulus of elasticity of E0=210 GPa, and Poisson’s ratio of v=0.3 was used for
bolts, flanges shells and flanges. Steel material S460 was set for the shells with yield point fy = 460
MPa and ultimate strength fu = 550 MPa. Nominal stress-strain curves shown in Figure 2.72 were
transformed to true stresses and true strains for input in ABAQUS plasticity model.
Bolts, washers and nuts were set to have nominal values of stress at the yield point fy=900 MPa
and ultimate strength fu = 1000 MPa, with elongation after failure A = 10% according to [23] for
bolt material grade 10.9. A parabolic shape of the nominal stress-strain curve was assumed in
order to enable the definition of damaged plasticity material model.
76
Figure 2.72: Stress-strain curves used in FE models.
Ductile damage model in ABAQUS was used to account for failure and element removal of the bolt.
Parameters of ductile damage initiation criterion and damage evolution law were derived observing
basic behaviour of the standard tensile test coupon and implementing principles of progressive
damage model described in [16]. Standard (round bar) tensile test model of a coupon was built
and material parameters were calibrated by comparing numerical results to corresponding nominal
curve shown in Figure 2.72. Good match of numerical and nominal curve was found as shown in
Figure 2.73. Subsequently this material model, and the same size and mesh type was used for the
one bolt model and the complete model of the ring flange connection.
Figure 2.73: Nominal and numerical tensile tests results
First, damage initiation criterion needs to be defined as equivalent plastic strain at the onset of
damage  0pl in function of stress triaxiality  (strain rate is rolled out). For uniaxial tension
(  1 / 3 ), corresponding to standard tensile test, equivalent plastic strain at the onset of damage
can be defined as  0pl   0pl   npl , where  npl is defined in Figure 2.74a) as uniaxial true plastic
strain at onset of necking obtained from experimental results of standard tensile tests (artificial
nominal curve in this case). Function of equivalent plastic strain at the onset of damage on
triaxiality, will be defined based on experimental and theoretical findings of some authors. Trattnig
et al. [24] conducted series of tests with different triaxiality on austenitic steels. Based on
experimental results they proposed exponential dependency of equivalent plastic strain at fracture
77
 fpl on triaxiality, as given by Eq. 3 in function of material constants  and . Same fracture line
was theoretically derived by Rice and Tracey [25] defining exponential dependency of the void
growth rate on triaxiality.
 fpl    exp(    ) 3
Divided by the same expression written in Eq. 4 for uniaxial strain state, the ratio of equivalent to
uniaxial strain at fracture  fpl /  fpl is obtained in Eq. 5.
 fpl    exp(  1 / 3) 4
 /   exp   (  1 / 3)
f
pl pl
f 5
It is assumed that ratio of equivalent and uniaxial strain at fracture and at the onset of damage are
the same:  fpl /  fpl   0pl /  0pl . Therefore equivalent plastic strain at the onset of damage  0
pl
is
given in Eq. 6, in function of triaxiality, based on uniaxial plastic strain at the onset of damage  0 .
pl
 0pl ( )   0pl  exp    (  1 / 3)  6

Material parameter = 1.5 is adopted as proposed by Rice and Tracey [25]. Finally with
 0pl   npl  0.0301 damage initiation criterion according to Eq. 7 is shown in Figure 2.74b).
 0pl ( )   npl  exp  1.5  (  1 / 3) 7
1400
Critical
Damage initiation, damage,
1200 "necking" point "rapture"
(Dn=0) point
(Dr=Dcr)
1000
True stress (MPa)
800
600
uniaxial
400 plastic strain Total damage,
at the onset "fracture"
of damage point
(Df=1)
200
accumulated plastic strain
0
0 0.1 0.2 0.3 0.4 0.5
Localized true plastic strain (-)
a) damage extraction procedure and input plasticity curve for ABAQUS
78
0.20
at the onset of damage (-)

Equivalent plastic strain
0.15
uniaxial
tension
0.10
0.05
0.00
-1.00 -0.67 -0.33 0.00 0.33 0.67 1.00 1.33 1.67 2.00
Stress triaxiality (-)
b) damage initiation criterion
0.0
Damage variable (-)
0.2
Dcr
0.4
0.6
0.8
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Equivalent plastic displacement (mm)
c) damage evolution law
Figure 2.74: Plasticity and ductile damage parameters for bolt materials.
Once the damage initiation criterion is defined, plasticity curve and damage evolution law for use in
ABAQUS material models can be extracted from experimental results of standard tensile tests
(artificial nominal curve in this case). The procedure shown here is based on engineering approach
and is presented in recursive form, practical for use in spreadsheet calculations and processing of
raw tensile tests data. Following characteristic points of nominal and true stress strain curves need
to be identified for further manipulation: p – onset of plasticity; n – onset of necking (damage
initiation); r – rapture point (critical damage); f – fracture point (total damage). Those points are
shown in Figure 2.73 and Figure 2.74a) and c) for bolt material.
After onset of necking, longitudinal strains of test coupon start to localize in the necking zone [26],
leaving other parts of coupon at the same strain as they were at the onset of necking. To account
for strain localization, initial gauge length l0 (100 mm in this study), is fictively reduced to length
lloc = 7.0 mm (for the case shown here) representing average necking zone length, as illustrated in
Figure 2.74a). Therefore, variable gauge length li is defined by Eq. 8 at every loading (elongation)
stage “i” as function of elongation  li . Rate of gauge length reduction i.e. strain localization is
governed by power law through localization rate factor L, calibrated to a value of L = 0.9 (for the
case shown here).
l 0 , i  n
li   8
l 0  (l loc  l 0 )( li  ln ) /( lr  ln )  L , i  n
Further, nominal strains  i
nom
are obtained by Eq. 9, following previous assumption that
increments of elongation after onset of necking are applied only to localized zone of test coupon.
79
  li / l i , i  n
 inom   nom 9
εi 1  ( li - li 1 )/li , i  n
Based on well-known relations, localized true strains i and true stresses i in necking zone are
obtained in Eq. 10 and Eq. 11 respectively. Those are shown in Figure 2.74a) with dashed lines as
damaged material response (section p-n-r-f).
 i  ln(1   inom ) 10
i   inom (1   inom ) 11
Undamaged material response is defined by Eq. 12 assuming perfectly plastic behaviour after onset
of necking (point “n” in Figure 2.74a)). Together with true plastic strains obtained in Eq. 13 it was
used as input data for plasticity curves in ABAQUS as shown with solid lines in Figure 2.74a)
(section p-n-r’-f’). Dashed extensions beyond point “f’” were made to solve issues of discretising
extremely high strains in the necking zone by finite element method.
 i , i  n
 i   nom 12
 n (1   i ), i  n
nom
 ipl   i -  p 13
Damage variable is obtained as dimensionless difference between undamaged and damaged
response of material as defined in Eq. 14. It can be noticed in Figure 2.74a) and c) that at rapture
point “r” material undergoes critical value of damage Dcr immediately followed by fracture point “f”
with total degradation of stiffness. This behaviour is also noticed by Lamaitre [26], defining value
of critical damage as 1-r/n in range of Dcr = 0.2-0.5 for most steels. Nevertheless this is macro
scale measure of damage variable, as average value across entire cross at which fracture occurs.
In the numerical analyses conducted here, significant nonuniform distribution of damage variable
was noticed at the cross section at which fracture occurs, affected by higher equivalent plastic
strains in the core of the cross section. Some other authors, such as Bonora et al. [27] also
observed that real values of critical damage for steel materials are higher (0.55-0.65). For this
purpose, damage eccentricity factor D was introduced in Eq. 14, with value D = 1.7. With this
value, good match of nominal and numerical rapture points were obtained.
(1   i /  i ) D , n  i  r
Di   14
1, i  f
Damage evolution law was inserted in ABAQUS in tabular form as damage variable Di in function
of equivalent plastic displacement u ipl . Values of u ipl corresponding to Di are defined by Eq. 15,
as proportional to evolution of plastic strains in the necking zone.
uipl  ufpl ( ipl   npl ) /( fpl   npl ), i  n 15

Total equivalent plastic displacement at fracture ufpl can be defined by Eq. 16 as characteristic
element length Lchar multiplied by plastic strain accumulated during the necking (damage) process
i.e. difference between plastic displacement at fracture  fpl and at the onset of necking  npl , see
Figure 2.74a). Finite element size factor S is introduced in Eq. 16 to take into account influence of
actual mesh density used LE in relation to refined mesh density LR which could be considered as
reference mesh. Different element sizes (mesh densities) were tried in order to establish this
factor. It has been found that element size factor follows the rule: S  3 LR / L E . The value used
in this analysis is S = 0.41 for the element size LE=5.0 mm in the complete model. (S = 0.55 for
the element size LE=3.6 mm in the one bolt model).
80
ufpl  S Lchar ( fpl   npl ) 16
Characteristic element length Lchar is dependent on finite element type and size [16]. It is here
defined in Eq. 17 as element size LE = 5.0 mm in threaded zone of the bolt multiplied by element
type factor E. It has been found that value of E = 1.0 is good match for tetrahedron C3D4
elements used for bolts in this study.
Lchar=ELE 17
For the procedure shown here and adopted set of parameters, good match of numerical coupon
test and nominal curve was found as shown in Figure 2.73. Therefore this material model, with the
same mesh type (C3D4) and corresponding size (5.0 mm) was used for the bolts in models of the
ring flange and friction connection.
Verification FEA with down-scale tests
Layout of the four point bending tests (7 m span) reported in [1], which were conducted in RWTH
Aachen, is shown in Figure 2.75. The ring flange connection with 32 M20 bolts and friction
connection with 24x3 M20 bolts were tested with shell diameter 1000 mm and thickness 8 mm. FE
models conforming to those tests are shown in Figure 2.76. The same modelling technique which is
presented for the real dimensions of a wind turbine tower analysed in this study is used for the
verification FE models. Appropriate material models were calibrated in the same manner as
presented previously using the data provided with test results [18].
a) ring flange connection b) friction conneciton
Figure 2.75: Four point bending tests of down-scaled wind turbine tower connections
[18].
a) ring flange connection b) friction conneciton
Figure 2.76: FEA models of down-scaled wind tower connection tests.
81
Comparisons of the results of down-scaled tests and verification FE analyses are shown inFigure
2.77 and Figure 2.78. High level of agreement between tests and FEA is achieved which justifies
further application of the presented modelling technique in all means of analysis type, mesh,
boundary conditions and material models that are used.
2,000 500
Buckling
1,800 resistance - EC3
1,600 400
1,400
1,200 300
Bolt force (kN)

Load (kN)
1,000
800 200
600
FC1 - Test FC1 - Test
400 100
FC1 - FEA FC1 - FEA
200
0 0
0 10 20 30 0 300 600 900 1,200 1,500 1,800
Deflection (mm) Applied load (kN)
a) moment-rotation curves b) evolution of the bolt force

Figure 2.77: Comparison of the results of verification FEA of ring flange connection tests.
2,000
1,500
Load (kN)
1,000
500 FJ - FEA
0
0 5 10 15 20 25 30 35
Deflection (mm)
Figure 2.78: Comparison of the results of verification FEA of friction connection tests.
Results and discussion
Comparison of friction connection and ring flange connection behaviour
Moment-rotation curves tor the three cases of different type of the connection and shape of the
tower are shown in Figure 2.79. Initial stiffness for all three cases are identical while ultimate
bending moments that are shown in Table 2-33 are different. In all cases the dominant failure
mode is buckling of the shell in compression, which is shown in Figure 2.80.
82
a) without imperfections / gaps
b) with imperfections / gaps
Figure 2.79: Moment-rotation curves for ring flange connection and friction connection.
Figure 2.79a) shows the results of analysis without dimple imperfection on the compression side of
the ring flange connection and without assembling tolerance gaps in case the of friction
connections. It can be noticed that buckling resistance in the case of ring flange connection is
higher than in the cases with friction connections. This is expected as there is no initial eccentricity
in the case of ring flange connection as it is present even in the prefect case of friction connection
(single-lap joint). If the imperfections and assembling tolerances are considered the friction
connection can resist higher bending moments (see Figure 2.79b). Ultimate bending resistance of
circular tower with friction connection is approximately 8% higher when compared to the case with
ring flange connection. Also the ductility of the friction connection is higher, since no failure of the
bolts occurs in this case, as it is present in the case of ring flange connection, especially in the case
without the imperfections. Influence of the imperfections and assembling tolerance gaps on
ultimate bending moment resistances are shown in Table 2-33. Load bearing capacity decreases
14.3% in the case of ring flange connection, while in the cases with friction connection this
decrease is lower (2.4 – 7.1%). Circular tower shape showed better performance when compared
to the polygonal tower shape with assembling tolerance gap of 10 mm. The reason is that in the
83
case of polygonal tower shape indents at the corners (see Figure 2.69) additionally reduce the
cross section area.
Table 2-33: Comparison of results for ring flange connection and friction connection.
Ultimate bending moment (kNm)
Without With Influence of the

imperfections/gaps imperfection/gap imperfection
Ring flange connection 106,649 91,659 -14.3%
Friction connection
101,835 99,309 -2.4%
(circular tower shape)
Friction connection
95,669 88,866 -7.1%
(polygonal tower shape)
a) Ring flange conn. b) FrC, circular tower shape c) FrC, polygonal tower shape
Figure 2.80: Buckling of the tower shell at the ultimate bending moment.
Evolutions of the membrane m and bending stresses b in the shell near the connection are
shown for the ring flange (RFC) and friction connection (FC) with circular tower shape in Figure
2.81. Membrane and bending stresses are defined as in Eq. 18, where 1 and  2 are edge streses.
1   2 1   2
m  ; b  18
2 2
It can be seen that membrane stresses are similar in both cases, while higher bending stresses on
the compression side are produced in the case of ring flange connection which leads to earlier
buckling failure. A possible reason could be increased stiffness in the connection zone of friction
connection due to double thickness of the material (overlapped shells).
84
Figure 2.81: Evolution of membrane and bending stresses in the cylindrical tower shell
with friction and ring flange connection.
Prying effects are present in the case of ring flange connection which leads to the increase of the
forces in bolts (see Figure 2.82a) and opening of the gap on the tension side (see Figure 2.82b).
There are no prying effects in the case of friction connection. Slight decrease of forces in bolts on
the tension side and increase on the compression side in the case of friction connection is
consequence of change in thickness of the shell material due to Poisson’s effects.
a) bolt forces b) slip and gap opening

Figure 2.82: Comparison of behaviour of ring flange and friction connection.
Assessment of the fatigue performance of the bolts for both types of the connection is shown in
Table 2-34. Stress ranges in bolts are given at fatigue damage equivalent load (DEL)
M eq  27700 kNm which is calculated according to Vestas data for Nref = 2x108 number of cycles
and slope of the Wöhler curve m = 4. Due to the prying effects in bolts in the case of ring flange
connection fatigue performance is much lower (approximately 3 times) when compared to the case
of friction connection.
85
Table 2-34: Fatigue performance of the bolts in the ring flange and friction connection.
Range of bolt force Bolt stress

at DEL level (kN) range (MPa)
Ring flange connection 50.7 34.4
Friction connection
-11.3 7.7
(bolt in tension)
Friction connection
15.4 10.5
(bolt in compression)
Friction connection showed good performance in means of serviceability since no significant slip
occurs at the design bending moment even at the ultimate limit state, as it can be seen in see
Figure 2.82b. Lover deformability of the friction connection when compared to the ring flange
connection (see Figure 2.82b) leads to the increased endurance of such connections.
Influence of length of the fingers in friction connection
Influence of the length of the fingers on ultimate bending resistance is presented in Figure 2.83
and Table 2-35. Comparison is made with reference to the cases without assembling tolerances as
gaps between the shells (the perfect cases). Relative slenderness of fingers F are calculated and
presented in Table 2-35 with respect to free finger length Lfree according to Eq. 19, assuming that
fingers are fixed at the top by the bolts and shell at the bottom.
0.5 L free  12 / t 0.5( LF  400mm)  12 / t

F   19
 E / fy  E / fy
For both tower shapes significant reduction of ultimate bending resistance is present only with
length of fingers of LF = 650 mm. In this case local buckling of the finger occurs (see Figure 2.84c),
because the relative slenderness is higher than the limit for buckling behaviour   0.2 according
to EN1993-1-1. In all other cases relative slenderness of the fingers is lower than the EC3 limit,
and no local buckling of the fingers occurs (see Figure 2.84a,b).
In both cases of tower shapes ultimate bending resistance is higher in the case of 550 mm long
fingers when compared to the case with 450 mm long fingers it the assembling tolerances are
provided. This is addressed to reduced initial stresses induced by bending of the shell due to
closure of the gap during preloading of the bolts.
In the case of polygonal tower shape additional reduction of ultimate bending resistances in cases
with assembling tolerances when compared to the perfect case is consequence of reduction of cross
section of the shell by indentions at the corners (see Figure 2.83b and Table 2-35). Additionally
stress concentrations are present at the roots of indentions at the corners, as it shown in Figure
2.85 at the stage after preloading of the bolts. Clearly this can raise fatigue issues which need to
be assessed in more details in order to justify application of polygonal tower shapes for wind
turbines.
86
Figure 2.83: Moment-rotation curves for different length of fingers in friction connection.
Table 2-35: Influence of the length of the fingers on ultimate bending resistance.
Tower Gap Total Free Relative Buckling Cross Ultimate

shape between length length slenderness reduction section bending
the of the of the of the factor acc. reduction moment
shells fingers fingers fingers to EC3 by indents
(mm) (mm) (mm)
0 450 50 0.054 1 1 101,835
circular 10 450 50 0.054 1 1 99,309
10 550 150 0.161 1 1 102,857
10 650 250 0.269 0.93 1 90,585
0 450 50 0.054 1 1 95,669
10 450 50 0.054 1 0.967 88,866
polygonal
10 550 150 0.161 1 0.967 89,769
10 650 250 0.269 0.93 0.967 76,635
a) LF = 450 mm b) LF = 550 mm c) LF = 650 mm

Figure 2.84: Post-buckling shape for different lengths of fingers.
87
a) no gap – LF = 450 mm b) 10 mm gap – LF = 450 mm
Figure 2.85: Stress concentration at the indention root of the polygonal tower shape.
Comparison of different methods to provide assembling tolerances
Provision of assembling tolerance by gaps between shells with different diameters leads to certain
reduction of ultimate bending resistance, as it is shown above. If the provision of assembling
tolerance is achieved by bending of the fingers (see Figure 2.67), better performance is achieved,
as it is shown in Figure 2.86. This is expected as lower imperfections are introduced to the tower
segment around the connection in the absence of the gap between the shells.

Figure 2.86: Moment-rotation curves for different methods to provide assembling
tolerances.
Additionally, if the gap between the shells is present, appropriate tightening sequence must be
adopted [1] and special care must be taken during preloading of the bolts, Preloading of the bolts
often need to be done in several steps of retightening. The method of provision of assembling
tolerances by bending of the fingers in advantageous in these concerns since the preloading of the
bolts can be done in one step if it starts from the top bolt row.
Conclusions
Advanced FEA of the real scale friction connection for wind turbine tower have been conducted
relaying on verification FEA of down-scaled tests of such connections. Friction connection is
88
compared to the ring flange connection of the same tower segment and parametric study is made
considering circular and polygonal tower shapes, assembling tolerances and length of the fingers.
Additionally, alternative method to provide assembling tolerances is proposed and examined.
Following conclusions are drawn:
1. Friction connection is less sensitive to imperfections when compared to the ring flange
connection. If necessary imperfections are considered friction connection has 8% higher
bending resistance than the comparable ring flange connection.
2. Fatigue performance of the bolts in friction connection is approximately 3 times better when
compared to the bolts in ring flange connection. This is the addressed to the absence of the
prying effects on the bolts in friction connection.
3. The deformability of friction connection is low when compared to the ring flange connection
(the slip is less than 0.15 mm at the design bending moment for ULS). Therefore, better
endurance can be achieved in the case of friction connection.
4. If the lengths of the segments of polygonal tower shape are chosen to satisfy the
slenderness limit of class 4 cross section the full effectiveness is achieved and no local
buckling occurs.
5. Approximately 12% higher bending resistance is achieved with circular tower shape when
compared to the polygonal tower shape of the same outer diameter. The necessary intents at
the corners in the case of polygonal tower shape reduce cross section of the tower in the
vicinity of the connection.
6. If the length of the fingers is such that the relative slenderness of the free part (between the
root and the lower bolt) is higher than 0.2, the local buckling of fingers will occur leading to
the reduction of ultimate bending moment. In the case analysed were, with relative
slenderness of the fingers 0.267, the ultimate bending moment is reduced by 9%, which is
close to the buckling reduction factor of compressed member according to Eurocode 3.
7. Indentions at the corners of polygonal tower shape are needed to provide assembling
tolerances. Stress concentrations up to 90% of yield strength of the material are present at
the roots of the indentions after the stage of preloading of the bolts. These stress
concentrations need to be considered for the fatigue performance of such tower shape.
8. An alternative method for providing assembling tolerances, by pre-bending of the fingers is

found to be advantageous when compared to the method by providing gaps between the
shells since less imperfection is introduced to the connection zone.
2.3.5 Longitudinal bolted connections

A next step towards higher hub heights and greater diameters of steel tubular towers is a
modularization in vertical direction.
For towers with such a segmented cross section, 4 plates overlap in the conjunction of longitudinal
and transversal joint, see Figure 2.87a). This can be avoided by cutting the edges of the two inner
plates, so that the two plates are located in one level (see Figure 2.87b)).
89
Figure 2.87: a) Conjunction of longitudinal and transversal joint; b) cut of an inner plate
Further information on this is given in chapter 2.3.7.3.
The plates of each segment have to overlap in order to be assembled by bolts.

This overlapping can be realized in two different manners, see Figure 2.88.
Figure 2.88: Possible overlapping possibilities, left asymmetric; right; symmetric
The symmetric version has to be prefered as two opposing plates lie on the same diameter which
indicates a better torsional resistance compared to the unsymmetric version.
2.3.6 Design Recommendations for modularized steel tubular towers

Based on the findings concerning the segment connections described in chapters 2.3.4 and 2.3.5,
design recommendations are given. These follow the flow chart in Figure 2.89.
90
Figure 2.89: Flow chart of the program of the transversal friction connection, including
the design resistance checks according to EN 1993
The design recommendations are implemented in two excel tools, which are described in brief in
the according background document.
Excel-Tools Tool 1a: “Optimal design of transversal friction connection” and the Tool 1b: “Pre-
Design of longitudinal friction connection” are tools, which can be used to predesign the transversal
(Tool 1a) and the longitudinal (Tool 1b) joints with the new HISTWIN connection [1] in advance to
further finite element calculations.
Tool 1a can be used for designing a transversal connection with the friction joint ‘HISTWIN’, where
all design verifications of the fingers are studied and the automatically selected assembly of bolts is
91
tested on plausibility. The Tool is divided into six parts. Every part is subdivided in other parts,
which can be displayed by opening the respective group. Under ‘0 assumptions and explanations’
the made assumptions are explained and general explanations are given. In part ‘1 internal forces’,
the global internal forces acting on at the top of the tower have to be provided by the user and the
local internal forces in the fingers are calculated automatically. ‘2 Material data and geometry’
includes the entries for defining the characteristics of the tower mostly by drop-down-lists. ‘3
Design verification of bolted connection’ contains the design verifications of the bolts. In section ‘4
Design of the fingers’, the design verifications of the fingers are carried out. The output of this step
is a listing of the results if the selected connection provides a sound design. The last group ‘5
Design verification of the tower for global buckling’ is a validity check, if the verification of the
global buckling surely does not become main design criteria.
Tool 1b serves to design the longitudinal connection of the HISTWIN tower. With its help, the
assembly of bolts (in one row) can be verified according to Eurocode and thus, a starting step for
further finite element calculations can be identified. The bolt loads can be calculated using FEM or
by estimating the shear flow in the connection due to the internal shear force similar to composite
structures. Both methods are available in the tool. The tool uses dropdown list in order to facilitate
the user to change parameters like bolt diameter, friction coefficient and bolt material among
others. After calculating the internal forces in the tower structure, the user can determine the
necessary number of bolts, the distance between bolts, their diameter and the friction coefficient,
which has to be kept. Figure 2.90 gives an overview of the input mask of tool 1.b.
Figure 2.90: Overview of the Excel-Tool 1B: ‘Pre-Design of longitudinal friction

connection’
2.3.7 Feasibility study

As in the previous HISTWIN project, the manufacturing of those two prototypes was based on two
main segments or set of segments produced on structural steel and with a final shape close to a
cylinder.
Conceptually, these prototypes are composed by two main cylinders assembled by the overlapping
of four circular plates connected by bolts in pure shear connections in the longitudinal direction
according to Figure 2.91b. That is the main difference of the previous concept were each segment
was composed by only one curve plate welded longitudinally according to Figure 2.91a.
92
a) Welded segment (HISTWIN) b) Bolted segment (HISTWIN2)
Figure 2.91 - Different solutions for friction connections
This HISTWIN2 concept is a modularized solution which will allow the transportation of larger (in
diameter) segments for higher wind towers. In this case, the rolling process (cold forming) plays
an important role due to the increased dimensional and geometrical precision required for this
system and it assembly process.
2.3.7.1 Preparation Phase

Raw Material
The purchase and delivery of the raw material used in these prototypes was carried out under the
recommendation of the standards:
EN 10025-2 - Hot rolled products of structural steels — Part 2: Technical delivery conditions for
non-alloy structural steels.
EN 10163-2 :2004 - Delivery requirements for surface condition of hot rolled steel plates , wide
flats and sections - part 2 : Plate and wide flats.
EN 10029:1991 – Hot rolled steel plates 3mm thick or above – tolerance on dimension, shape
and mass.
EN 10204:2004 - Metallic products – Types of inspection documents.
EN 10027-1: 2005 – Designation systems for steel – part 1: Steel names.
As it can be checked in the Deliverable D8, annex C1, it was used 151 and 202 mm thick plates with
S355J2+N steel grade.
It was not defined any requisites for the raw material reception and so it was defined, by default,
the reception criteria3 4 defined by the standards:
EN 10163-2: Class A1
EN 10029: Thickness: Class A, Width: According table 2, Length: According table 3
Design
1
The plate thickness was commissioned following the recommendations / specifications received by FCTUC
2
The plate thickness was commissioned following the recommendations / specifications received by FCTUC
3
The technical reception records are attached to this document
4
These requirements should be reviewed in the future in order to minimize eventual defects in the process, reduce waste
and encourage better use of raw materials
93
The design of the modular plates was performed by taking into account the recommendations
received by FCTUC in an interactive process to obtain a technically satisfactory solution.
The 3D model and the production drawings were executed with Solidworks™ and the files to be
imported and used by the numerical control of the thermal cutting machines were executed by
LANTEK™
The latest revision of all drawings used in the manufacture process of the prototypes, is available in
the document.
2.3.7.2 Feasibility Tests

In the previous concept, the prototype was built with two different steel cylinders (according to
Figure 2.92).
Figure 2.92: HISTWIN prototype assembly
The final geometry of the plates was obtained by using a thermal cutting technique and the round
and open holes used for the connection of the plates were also drilled using this technique
according to the Figure 2.93.
a) Steel plate on the b) Long open slotted

c)Round holes
cutting holes
Figure 2.93: Cutting techniques (HISTWIN prototype)
As it can be checked in Figure 2.94, the open holes where only partially opened before the rolling
due to the possibility of not properly bend those areas during the process and so they were “pre-
cutted”, then rolled. In the end they were totally opened.
94
Gaps after
the cutting
Steel to bear the long

slotted holes during the
Figure 2.94: Long open slotted holes
After the rolling procedure the plates were longitudinally welded using submerse arc technique and
finally with a manual thermal cutting equipment were removed all the joint points to form the long
slotted holes.
In order to reduce the required time to execute the long open slotted holes, it was focused the
attention in the final manual cutting process, so were formulated two hypotheses to reduce this
process time for the manufacturing of the new prototypes:
1st Hypothesis
Check if it is possible to execute the complete cutting process without interruptions with the plain
plate. This will allow to avoid the final manual cutting required in the technique adopted for on the
previous HISTWIN and also to avoid the constant rotations required in order to allow an easy
cutting.
2nd Hypothesis (only if 1st hypothesis does not work)
The strategy will be to perform follow the next steps:
 Partial cut of the holes, leaving a strip of 10 mm around the top of the holes. This strip will
be removed after the plate rolling with a cutting torch applied to a robot arm (Figure 2.95)
Cutting zone after

rolling
Figure 2.95: Indication of the cutting zone
 The robot will create a cutting trajectory as function of the piece radius (Figure 2.96).
Cutting
t j t
Figure 2.96: Robot cutting trajectory and robot
In order to check the feasibility, both techniques were tested in order to check possible
deformations and misalignments in the “fingers” during the rolling process.
95
Two types of plates were then created and tested:
 Type 1: fully cutted long open slotted holes;
 Type 2: long open slotted holes partially cutted for posterior robot cutting after rolling
Note: For the rolling of the plates it was used:
 Equipment: Rolling machine PROMAU DAVI-LEONARDO MCB3028 WT
 Process: Multi-step
 Pressure: 60 bar
Figure 2.97 shows the rolling process of the plates of both types tested as well as the check of the
obtained plate curvature using moulds specially manufactured for these tests.
Figure 2.97: Rolling process
2.3.7.3 Manufacturing Process

After receiving the steel plates, the Martifer´s QSA department executes the reception (please
check the attached C2). The steel sheets will be delivered to our factory, the documents to prove
the origin and technical properties will be compared with the purchase documentation and, if there
are not any situations to report, all records are archived for a minimum period of five (5) years.
The raw material stock will be performed taking into account the thickness and outside dimensions
of the steel plates.
The steel plates are stocked above the ground level upon platform (square profiles or other
supports appropriate to this purpose).
Feasibility
Two specimens were manufactured, with two different plate thicknesses in order to evaluate and
compare the behaviour of plates with different lengths.
In this work, it was considered one specimen with 15mm plates and another with 20 mm thick
plates.
The global view of the prototypes with overlapping friction joints can be found in Figure 2.98.
96
Figure 2.98: Global view of the model
In some parts of the connection is required the superposition of the plates and in that cases it was
adopted a solution with round corners in order to avoid any stress concentration which improves
the fatigue behaviour.
Figure 2.99 shows some details of the pieces superposition. In the overlapping area of the connection
where the longitudinal shear connection and the transversal friction connection meet, there is a
superposition of 3 plates, while in the rest of the segments it can be found only a superposition of 2
plates. Figure 2.100 shows a detail of that overlapping joint.
Figure 2.99: Superposition of the plates
Figure 2.100: Overlapping joint of shear and friction connection
In the friction connections, cover plates will be placed above the long open slotted hole plate in
order to allow a uniform load transfer between the bolts and the plates.
Figure 2.101 shows the drawing for those plates with 10 mm of thickness for both models.
97
Figure 2.101: Cover plates
Production drawings and tolerances – 15 mm thick prototype
Figure 2.102 show the global geometry of the 15 mm thick plate connection.
In Figure 2.103 and Figure 2.104 is showed the production drawings from Martifer for the 15 mm
thick plates. In this case it was considered a gap of 10 mm for that prototype which means that the
outer diameter of the inner plate is 20 mm lower than the inner diameter of the outer plate.
Figure 2.102: 15 mm thick prototype
98
Straight corners (2 plates)
Round corners (2 plates)
Figure 2.103: 15 mm thick connection parts: Upper part (long open slotted hole plate)
99
Figure 2.104: 15 mm thick connection parts: Lower part (normal round hole plate)
Production drawings and tolerances – 20 mm thick prototype
Figure 2.105 shows the global geometry of the 20mm thick plate connection.
Figure 2.106 and Figure 2.107 show the production drawings from Martifer for the 20 mm thick
plates. In this case it was considered a gap of 5 mm for that prototype which means that the outer
diameter of the inner plate is 10 mm lower than the inner diameter of the outer plate.
100
101
Manufacturing process
The manufacturing process of the prototypes is illustrated in the flowcharts presented in annex A1
and A2.
Transport (Phase A1)
All the process to unload the trucks or load the line production will be performed with automatic or
semi-automatic lifting equipment available in the factory facilities (Figure 2.108).
Figure 2.108: Lifting equipment
Thermal cutting (Phase A2)
102
The Cutting process was performed with a CNC thermal cutting machine (using the plasma
technology) as presented in Figure 2.109. The specifications of the cutting machine can be found
on Figure 2.110.
Figure 2.109: Thermal cutting process
Cutting machine :
 ESAB SUPRAREX SXE-P 8000
Cutting parameters
 Shield retainer: 37081

 Shield: 0558006141
 Nozzle retainer: 37082
 Diffuser: 21944
 Nozzle: 0558006020
 Electrode: 0558003914
 Baffle: 0558002533
 Holder: 0558003924
 Amperagem: 200 Amp
 Voltagem: 148 V
 Kerf: 3,5 Figure 2.110: Print screens - obtained of the
 Plasma Speed: 1380mm/min CNC program
In each segment it was carried out some dimensional control measurements such as:
 Holes: random control of, at least,

10 holes in each plan plate
 Open holes: check of the width of

the open holes (Figure 2.111)
Figure 2.111: Width of the open

holes
 Diagonal measurement of each plan

plate
 Height of the connection region

(Figure 2.112)
Figure 2.112: Height of the conn.

region
Process time for all the plates
103
 Plate cutting (effective cutting time) – 280 minutes
 Transportation (movement and placing of the plate) – 15 minutes
 Setup – 5 minutes
Cleaning of the plates (Phase A3)
Due to the presence of dirt around the holes, due to the cutting process, it was needed to remove
it using manual grinders.
As it can be checked in Figure 2.113 the dirt is more prominent around the holes of the round holes
plate.
Figure 2.113: Dirt around the holes (after thermal cutting)
Process time for all the plates
 Process time – 560 minutes
 Movements and setup – 288 minutes
Rolling of the plates (Phase A4)
Due to the thermal cutting of the previous phase of the process, some steel hardening may occur
which may be the source of cracks. It is almost impossible to avoid the crack propagation during
the rolling process.
To avoid this type of flaws (mainly for thicknesses higher than 25mm) it may be foreseen bevers in
the edges of the plates.
In this case, as the maximum thickness is 20mm and the loading applied is very lower than in a
real situation, it was adopted not to manufacture the referred bever.
It was considered that the dimensions of the plan plate were kept unchanged during the rolling and
that the length of the plan plate is the same as the perimeter of the final piece. It was also
assumed that the rolling was achieved by a pure bending moment and the neutral axis of the
deformation is coincident with the medium line of the segment5.
For the rolling of the plates, it was used the equipment:
 Rolling equipment - PROMAU DAVI-LEONARDO MCB3075 WT
 Process - Multi-step
 Pressure - 60 bar
5
For plates with higher thickness or lower curvature radius, the neutral axis is located in
a distance between ¼ and ½ of the thickness of the plate, measured from the interior of
the segment. The location varies with the curvature radius and the material mechanical
properties
104
The initial radius was determined taking into account the elastic recovery of the plates and it
estimated using Figure 2.114.
Legenda :
 Re = Rolling radius
 Rf = Final radius
 h = Thickness of the plates
 E = Young modulus ( 210.000Mpa)
 σe = Yield stress
Figure 2.114: Method for the estimation of the rolling radius
The values required for the expressions presented on Figure 2.114 can be found in the product
certificate presented on annex.
In order to check the curvature of the plates it were manufactured four different pieces in steel
with the desirable radius (Figure 2.115). The pieces were cleaned and the radius was previously
checked.
Figure 2.115: Curvature check pieces
Process time for all plates:
 Process time – 560 minutes
 Movements and setup – 240 minutes
Notes:
 For rolling with curvature radius higher than 100 times the thickness of the plates, it is
assumed that the maximum strains obtained in the plates are lower than 0.005 and during
the rolling it will be formed an elastic deformation core in the central area of the plate.
 In the case of curvature radius lower than 100 times the plate thickness it is assumed that
full plasticity is likely to occur and it is distributed along all the plate.
Assembly (Steps B and C)
The assembly of both segments was accomplished with a 16ton crane, chains with 2 arms and
elevation claws. The position of the claws was fundamental to ensure the levelling of the pieces.
105
In both prototypes, the assembly started for the lower segment by placing together all four pieces.
In order to keep them together and close to the final shape, some bolts were used in the
longitudinal connection.
After the assembly of the lower segment, the installation of the upper segment follows two
different paths.
For the 15 mm thick connection (Annex A2 of Deliverable D4), the upper level was assembled by
placing individually all four pieces above the lower level, as presented in Figure 2.116.
Figure 2.116: 15 mm thick connection assembly
In the case of the 20 mm thick prototype (Annex A1), the assembly of both segments was carried
out individually according to Figure 2.117 and Figure 2.118.
Figure 2.117: Assembly of the lower segment (20 mm thick connection)
Figure 2.118: Final assembly of both segments (20 mm thick connection)
Post-assembly measurements/checks
Segment shape geometry (20 mm thick prototype)
It was measured the geometry of both segments (upper and lower) on both prototypes. The
measurement was made in the inner side of the prototype according to Figure 2.119.
106
Figure 2.119: Measurement points on the prototype
The measurements were made after the tightening of some bolts in the longitudinal shear
connection. It were conducted four measurements in order to obtain the distance between the
tightening locations.
The measurements where obtained in the following points:
 Upper level – CS
 Intermediate level - Cm
 Lower level – CI
The results obtained are presented in Table 2.36.
Table 2.36 – Shape measurements
Measurement point Value (mm) Observations

Cs0-Cs1 2140
Cs3-Cs2 2150
Upper level
Cs1-Cs2 2111
Cs0- Cs3 2119
Cm0-Cm1 1986
Cm3-Cm2 1992
Intermediate level
Cm1-Cm2 2128
Cm0- Cm3 2132
Ci0-Ci1 1985
Ci3-Ci2 1990
Inferior level
Ci1-Ci2 2128
Ci0- Ci3 2125
The curvatures of the upper and lower segments were different, with 11 mm gap between both
segments.
Gap measurement
This check was made in order to obtain the gaps between both segments (upper and lower) in both
segments along the perimeter. It was made using an analogic calliper for external measurements
as presented in Figure 2.120.
107
Figure 2.120: Gap measurement with analogic calliper
The results obtained in the measurements, in the inner and outer faces of the prototypes, are
presented on Table 2.37. As it can be observed, 27 check points (F0 to F26) were measured along
the perimeter of the prototypes. The measurement points were marked in the prototype walls as
presented in Figure 2.121.
Figure 2.121: Gap measurement points
108
Table 2.37 – Gap measurements
External Gap Internal Gap External Gap Internal Gap

Point Point
(mm) (mm) (mm) (mm)
F0 2.1 1.6 F0 11.6 13.6
F1 3.1 2.7 F1 11.8 9.7
F2 1.5 0.9 F2 12.4 9
F3 0.9 1.3 F3 6.7 2.2
F4 1.6 2.1 F4 3.9 0.8
F5 0.7 1.5 F5 4.5 2.5
F6 3.2 3.3 F6 4.8 3.1
F7 3.9 2.2 F7 6.6 6.9
F8 2.7 1.1 F8 13 12.1
F9 6.4 2.4 F9 12.1 8.8
F10 7.6 3.3 F10 9.6 4.9
F11 6.2 3.1 F11 3.3 1.7
F12 3.3 2 F12 6.4 5.7
F13 4.1 2.6 F13 5.1 5.1
F14 3.1 2.4 F14 6.4 8
F15 3.4 3.1 F15 11.1 9.8
F16 6.8 4.9 F16 7.7 2.9
F17 5.3 1.6 F17 7.4 1.2
F18 5.9 3.5 F18 1.7 1.9
F19 0.8 1.3 F19 2.8 1.1
F20 3.5 0.9 F20 2.2 12.3
F21 1.9 0.9 F21 10.9 13.3
F22 4.9 1.7 F22 12.2 11.3
F23 11.6 7.8 F23 7.6 6.5
F24 9.8 6.8 F24 7 3.3
F25 5.5 3.4 F25 1.9 4
F26 1.7 0.8 F26 4.3 1.4
20 mm thick prototype 15 mm thick prototype
2.3.7.4 Conclusions
After the rolling it was observed that both types of pieces tested fulfil the requirements foreseen
which means that in both cases the “fingers” curvature is similar to the final curvature of the plate.
According to the exposed, it was adopted the type 1 process for the manufacturing of the final
prototypes as this process lead to a reduction of the amount of work time to produce the final
pieces.
Though the holes had been cut using the plasma technique, they presented an acceptable quality
for the assembly of the prototypes but in this case it was used a plasma arc current of 200A. The
final results might had been improved by using an intensity of 260A.
As there was not been made a pre-hole for the starting of the plasma cutting, it was accumulated a
lot of dirt around the holes that had to be removed after the cutting process. The use of the pre-
holes may decrease the cleaning time considerably which may improve the required time for the
manufacture of such elements since the cleaning time is twice of the cutting time.
The choice of the proper rolling pressure is a key factor for the rolling. Excessive pressures may
lead to the appearance of waviness on the plates close to the centre of the roll. With low pressures
the waviness effect is more pronounced close to the edges of the plates (along the width). In this
case it can be concluded that the chosen pressure was correct as no waviness was detected in
none of the plates.
There was observed some problems in the assembly of the two first pieces due to the differences
between both curvature radius (it had to be made some tightening and untightening in order to get
the proper shape of the connection). Due to that difference, it is recommended to develop some
kind of moulds to help in the assembly stage, in order to obtain quickly the desired final shape and
diameter for the segment.
The assembly technique used for the 15mm thick connection was found to be slightly quicker but
the technique of the 20 mm thick prototype is likely to be the most suited for the in-situ
assemblies.
109
It was observed some inconstant gaps between the upper and the lower segments related with the
process of rolling and/or assembly. The difference of gaps between both segments may be related
with a pre-tightening that occurred prior the measurement in order to fix all the pieces together
and to allow the full assembly which may lead to some plastic deformation of the elements.
It was also observed some misalignments (around 1.5 mm) between the holes of two different
levels but with the tightening of the first bolts that misalignment tended to disappear.
Due to the fact that there was no gap between the adjacent pieces of different levels, it was
difficult to install properly the pieces and close the segment. It was required to tighten and
untighten the bolts some times. For future prototypes it is important to foreseen some gaps in this
region (in the round corners).
The use of automatic wrenching machines will significantly improve the time for bolts tightening
process.
There are two possible manufacturing options for in-situ assembling: bolted friction connection and
welded technology, the first of which is considered in the report.
2.3.8 Foundations
2.3.8.1 Introduction
The increase in the height of the new generation of wind towers raises some problems in the
structural components of those systems as described previously.
One of the problems related to the increase in the height is the foundations dimensions. The
current steel wind towers with a height of 80 meters require an octagonal shallow foundation of
about 17 meters of diameter, with a concrete consumption of about 400 m3 which represents a
cost of about 20% of the full tower budget. The general geometry considered for the wind towers
foundation is presented on Figure 2.122.
Figure 2.122: Geometry and geometric parameters of the octagonal foundation
In this chapter, an analysis of the feasibility of the use of micropiles as a reinforcement of the
current shallow foundations will be explained and presented.
A detailed description of the geotechnical design of both shallow and hybrid (shallow reinforced
with micropiles) foundations will be presented as well as the design results of some specific
examples. A LCA analysis was carried out in order to understand the impact of each solution and to
understand the benefits of the use of micropiles.
The procedures and the results of experimental tests carried out on reduced scale micropiles will
also be presented to understand their behaviour and to calibrate numerical models in order to
extrapolate the results to a real scale.
110
2.3.8.2 Alternatives for tower foundations
Micropiles are a specific type of small diameter (usually lower than 300 mm) deep foundations
which allow the mobilization of both tensile and compressive resistance. This elements are used
both as a new foundation solution and as a retrofitting of existing foundations.
Usually the design of the current shallow foundations for wind towers is governed by the
overturning failure mode since those systems do not allow the mobilization of tensile resistance.
The use of the micropiles as a reinforcement will allow an improvement in the overturning
behaviour of the foundation with the mobilization of the tensile resistance but also with the
compression resistance improvement. The micropiles will also improve the dynamic behaviour of
the foundation by reducing the vibrations which may benefit the energy production equipments.
Some authors/manufactures already evaluated this situation (or similar) by proposing some new
solutions for the foundations of wind towers.
Patrick and Handerson, Inc. proposed two solutions for the reinforcement of wind towers
foundations both for cohesive as non-cohesive soils respectively P&H Tensionless Pier [42] Figure
2.123.a, which is a large pipe with corrugated sheets inserted in the soil and filled with concrete
and P&H Rock or Pile Anchor Foundation [42] Figure 2.123.b, which is a solution very similar to the
use of micropiles.
a. P&H Tensionless b. P&H Rock or Pile Anchor
Pier Foundation
Figure 2.123: Patrick and Handerson deep foundations solutions [42]
Ruuki also conducted some studies on the behaviour of micropiles as a reinforcement of wind tower
foundations and concluded that by using eight steel piles with 600 mm of diameter and 10 m of
depth and injected in the bedrock for a few meters they achieved a reduction in the concrete
consumption of about 2/3 in the shallow portion of the foundations (600 to 250 m3) and the total
cost of the foundation was reduced in about 10% [43].
Aschenbroich [44] also studied the effect of using deep foundations on the reinforcement of
shallow foundations for wind towers (Figure 2.124) and concluded that it may lead to a reduction
on the foundation area of about 75%, on the concrete consumption of about 40% and on the steel
consumption of about 70%. The final cost of the foundation may be reduced on 20% to 30%.
111
Figure 2.124: Micropile application on wind turbine foundations [44]
2.3.8.3 Shallow foundations

The traditional shallow systems with geometry similar to the presented on Figure 2.122 and
considered for the actual wind towers must fullfill some design requirements and verifications.
For the sake of this study, the European design guides and design recommendations were
considered are presented as follows.
The design of the shallow foundations must fullfill the following requirements:
 Pressure on the foundation for non-factored loads – in this verification it was estimated the
maximum applied stress in the base of the foundation (circular foundation);
 Verification of the compressed area - for characteristics loads, at least 50% of the base area
must be compressed which can be fullfiled when e < 0.59R. This verification is only applied
to the wind action.
 Limit state of equilibrium according to EC7-1 [45] – in this verification was checked if the
resistant moment due to the vertical forces is enough to ensure the equilibrium of the
foundation when subjected to the overturning moment;
 Soil bearing capacity according to EC7-1 [45] – in this case it was guaranteed that the
maximum applied pressure in the base of the foundation is lower than the bearing capacity
of the soil. This verification was considered for three types of combinations according to the
EC7-1 [45] (ULS-GEO, ULS-STR and accidental combination for seismic loads);
 Soil bearing capacity according to EC8-5 [46] – this part of Eurocode (Annex F) presents a
formulation for foundation stability taking into account the soil inertia (kinematic effect). In
Pender (2010) is presented that the drained response is much more sensitive to inertia
loading then undrained one;
 Soil bearing capacity according to DNV [47] for extremely eccentric loading – This
verification reflects the failure of the soil under the unloaded part of foundation area
(Rupture 2).
 Base sliding – in this point it was checked that the friction resistant forces between the base
of the foundation and the soil beneath (related to the vertical forces) are higher than the
acting horizontal forces on the tower and turbine structures. In this verification is also
considered unfavourable contribution of the torsional momente, Mz.
The design of the foundations was conducted for 3 diferent heights of wind towers (80, 100 and
150 meters supporting multi-megawatt turbines of 2, 3.6 and 5 MW respectively) with 3 different
structural solutions (steel, concrete and hybrid (concrete+steel)). The connections between the
steel segments is assumed to be friction connections. It was considered a C30/37 concrete class
and S500 rebar steel grade.
The loads considered for this study are presented on Table 2.38. The horizontal forces (FH) and the
bending moment (M) are not related to the system of the foundation considered. The values for the
112
vertical forces (FZ) are related to the foundation system considered because of the foundation self-
weight as the dimensions differ from each case.
In Table 2.38 it is only presented the load values for the governing load cases. For the horizontal
forces and bending moment the governing load case for the steel towers is the Extreme Wind
Model (EWM) according to the design guide IEC 61400-1 [48] while for the concrete towers the
governing load case is the Earthquake (EQ) according to EC8-1 [49].
Table 2.38: Load table (base of foundation)
2.0MW 3.6MW 5.0MW

Load
80m 100m 150m
FH (kN) – Steel (EWM) 871 1728 1515
FH (kN) – Concrete (EQ) 2802 6047 6256
FH (kN) – Hybrid (EQ) 1979 3678 4182
FZ (kN) – Steel (Shallow) (EWM) 13699 29934 44347
FZ (kN) – Concrete (Shallow) (EWM) 23635 56508 98073
FZ (kN) – Hybrid (Shallow) (EWM) 19025 42090 74987
M (kN.m) – Steel (EWM) 63633 152598 266777
M (kN.m) – Concrete (EQ) 178533 494790 803576
M (kN.m) – Hybrid (EQ) 131191 324291 600570
Mz (kN.m) – Steel (EWM) 1218 5961 5834
Mz (kN.m) – Concrete (EWM) 1218 5961 5834
Mz (kN.m) – Hybrid (EWM) 1218 5961 5834
Considering a torsional moment in addition to the horizontal and vertical forces to the tower
structure and therefore to the slab it is important to take into account the interaction between
these forces, made in accordance with the methodology proposed by Hansen (1978) (cit [47]).
The soil considered complies with the soil type B for the seismic analysis according to EC8-1 [49]
and it properties are presented in Table 2.39.
Table 2.39: Soil characteristics
´k c´k γ E
3
42 º 0 kPa 18 kN/m 675MPa
The dimensions and the material quantities obtained for the shallow foundations of the tower
geometries considered are presented on Table 2.40. The geometric parameters designation of the
foundations are according to Figure 2.122.
113
Table 2.40: Octagonal shallow foundations design results
80m 100m 150m

2.0MW 3.6MW 5.0MW
Tower Type
Concret Concret
Steel Concrete Hybrid Steel Hybrid Steel Hybrid
e e
Hf 2.00 2.50 2.50 3.00 3.50 3.50 3.50 4.50 4.50
Governing
EW EQuake EQuake EW EQuake EQuake EW EQuake EQuake
load case
Governing
EQU
design criteria
H1 0.95 0.95 0.95 1.20 1.20 1.20 1.20 2.00 2.00
H2 2.00 2.50 2.50 3.00 3.50 3.50 3.50 4.50 4.50
H3 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50
I 7.04 7.04 6.21 8.28 9.11 7.87 8.70 9.53 8.28
Beq 17.46 17.46 15.41 20.55 22.60 19.52 21.57 23.63 20.55
17.0
B 17.00 15.00 20.00 22.00 19.00 21.00 23.00 20.00
0
B0 5.60 8.00 7.20 8.00 11.00 10.80 11.00 14.00 13.80
C 18.40 18.40 16.24 21.65 23.81 20.57 22.73 24.90 21.65
3
Concrete (m ) 359.0 458.9 373.6 729.3 1058.6 831.4 981.9 1664.1 1324.0
Rebar (Ton.) 30.5 54.4 44.35 53.0 114.0 62.7 65.4 152.6 62.8
Excavation
394.9 504.8 410.9 802.2 1164.4 914.5 1080.0 1830.6 1456.4
(m3)
2.3.8.4 Foundations with steel micro-piles

In this chapter, a detailed description of the design of hybrid foundations (shallow reinforced with
micropiles) will be presented. The geometry considered for the foundation is the same as presented
in Figure 2.122 but with micropiles placed close to the foundation edges in order to have the
largest lever arm available.
The design of the hybrid foundations has some similarities with the shallow foundations design,
however the contribution and the behaviour of the micropiles must be taken into account. This
must be done using a finite element model.
Such as in shallow foundations, the bearing capacity of the hybrid foundations is verified.
 Soil bearing capacity according to EC7-1 [45], EC8-5 [46] and DNV [47] - In this case, the
bearing capacity was estimated by determining the percentage of load transferred by the
shallow portion of the foundation and by the micropiles elements. It was estimated the
percentage of both the bending moment and axial force resisted by the micropile elements
and it was considered that the remaining load was transferred by the shallow portion of the
foundation; The forces applied in each micropile, obtained with the numerical model, were
compared with the individual load capacity of each element, both for combination 1 and 2.
For this analysis, the adopted micropiles were 88.9 mm x 9.5 mm (O.D. x t) pipe micropiles (As =
2370 mm2), in N80 steel (fy = 562 MPa), installed in 200 mm drilled boreholes and injected with
grout (fck,grout = 25 MPa). The internal (structural) resistance of the element is 1066 kN in tension
and 1550 kN in compression.
For those elements, it was obtained for EQUilibrium and Combinations type 1 an external
(geotechnical) compression resistance of 1508kN and 1206kN for the tension resistance
considering Eurocode 7 design approach 1 (D.A.1). For Combinations type 2 the compression
resistance value is 1160kN and the tension resistance is 942kN. In Seismic combination the
compression resistance is 1311kN and the tension resistance 1160kN. In this case it was
considered boreholes with 200 mm diameter, a bond length of 12 m and an ultimate grout-to-
ground shear resistance of 300 kPa.
114
 Limit state of equilibrium according to EC7-1 [45] – in this verification was checked if the
applied forces in the micropiles, obtained with the numerical model, never surpassed their
resistant capacity;
 Micropile spacing – it was checked that the spacing between adjacent micropiles was higher
than 2 to 3 times of the micropiles diameter in order to avoid group effect phenomena;
 Base sliding – it was assumed that the sliding resistance was only provided by the shallow
portion of the foundation and ensuring that it was enough to resist the applied horizontal
forces, neglecting the effect of the micropiles.
In order to estimate the portion of the loads resisted by the micropile elements, a numerical model
must be used and in this case it was considered a numerical model performed with Autodesk
Robot® [50].
In this case the behaviour of the micropiles was introduced by considering a spring with a bi-linear
behaviour with plateaus as defined in Figure 2.125. Depending on the the combination considered
for the design, the plateau was defined by the minimum value between the internal and the
external resistances.
R
Rtension
u
Kz
Displacement
u
Rcompression
Figure 2.125: Bilinear micropile springs
In the following examples, the soil-springs where defined as compression-only, linear-elastic, with
a coefficient of vertical reaction computed with Es=675 MPa and ν=0.30, resulting on different kz
for each foundation dimension according to Vesic’s expression.
In this case, the number of micropiles used was 32, 64 or 96 according to Figure 2.126.
115
a. 32 micropiles b. 64 micropiles
c. 32+32 d. 64+32
Figure 2.126: Micropile locations
Considering the procedure presented and the geometries considered, the results of the design of
the hybrid foundations are presented on Table 2.41.
Table 2.41: Octagonal hybrid foundations design results
80m 100m 150m

Tower type 2.0MW 3.6MW 5.0MW
Steel Concrete Hybrid Steel Concrete Hybrid Steel Concrete Hybrid
Hf 2.00 2.50 2.50 3.00 3.50 3.50 3.50 4.50 4.50
Governing load case EW EQuake EQuake EW EQuake EQuake EW EQuake EQuake
Typology (according
a b a c b b b d b
to Figure 2.126)
H1 1.59 1.81 1.74 2.40 2.25 1.95 2.58 2.83 2.69
H2 2.00 2.50 2.50 3.00 3.50 3.50 3.50 4.50 4.50
H3 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50
I 4.14 4.97 4.56 4.97 7.04 5.80 6.21 8.28 7.04
Beq 10.27 12.33 11.30 12.33 17.46 14.35 15.41 20.55 17.46
B 10.00 12.00 11.00 12.00 17.00 14.00 15.00 20.00 17.00
B0 5.60 8.00 7.20 8.00 11.00 10.8 11.00 14.00 13.80
C 10.82 12.99 11.91 12.99 18.4 15.15 16.24 21.65 18.40
Concrete (m3) 164.8 297.9 246.9 360.9 707.4 497.0 655.5 1410.6 1067.5
Rebar (Ton.) 14.1 19.2 13.8 20.3 56.3 21.0 30.7 66.3 31.2
Micropiles Steel
7.14 14.29 7.14 14.29 14.29 14.29 14.29 21.43 14.29
(Ton.)
Micropile Grout (m3) 11.17 22.34 11.17 22.34 22.34 22.34 22.34 33.50 22.34
Excavation (m3) 181.3 327.7 271.5 397.0 778.1 546.7 721.1 1551.6 1174.3
116
2.3.8.5 Micropiles experimental tests
In order to better understand the behaviour of the hybrid foundations, 36 reduced scale
experimental tests were carried out in order to calibrate a numerical model for the behaviour of the
micropiles and so obtain the proper resistances and stiffness to introduce in the spring behaviour
(as in Figure 2.125) of the numerical models of the whole hybrid foundations.
The calibration of a numerical model/procedure will allow an extrapolation for other models with
longer micropiles and different (better) ground properties.
Single and groups of micropiles were tested under monotonic or cyclic+monotonic loading as
described inTable 2.42. They were tested without and with pressure grout in order to access about
the grouting improvements. In the designations M stands for single micropile while G stands for
group.
Table 2.42: Experimental tests description
Micropil
Layout Loading Grout
e
Compression No
Tension No
M3 Cyclic +
Yes
Compression
Cyclic + Tension Yes
2
Compression No
Tension No Layout Group Loading Grout
M4
Compression Yes Cyclic +
No
Compression
Tension Yes
Cyclic + Tension No
Cyclic + 5 G1 (3B)
No Cyclic +
Compression Yes
Compression
Cyclic + Tension No
M5 Cyclic + Tension Yes
Cyclic +
Yes Cyclic +
Compression No
Compression
3 Cyclic + Tension No
Cyclic + 6 G2 (4B)
No Cyclic +
Compression Yes
Compression
Cyclic + Tension No
Cyclic +
Yes Cyclic +
Compression No
Compression
Cyclic + Tension No
Cyclic + 7 G3 (5B)
No Cyclic +
Compression Yes
Compression
Cyclic + Tension No
Cyclic +
Yes
Compression
4
Cyclic +
No
Compression
Cyclic + Tension No
M8
Cyclic +
Yes
Compression
Single tests Group tests
The loading protocol was based on control of displacements. The cycles had a period of 800 sec.
each and amplitude of ±1 mm. Five cycles were applied on all but the tests on micropile M8, where
10 load cycles were adopted to evaluate the effects of the additional cycles on micropile behavior.
117
In the monotonic tests (or after the cyclic loading) the displacement rate was 0.01 mm/s for the
compression tests and 0.005 mm/s for the tensile tests.
The micropiles were installed in a cylindrical soil container with a 2m diameter and 3.5m high. Due
to space constraints related to the available height inside the laboratory, the placement of the
micropiles into the experimental layout was carried out before the filling of the container with sand.
The load was applied using a 20 tons Dartec actuator and a steel reaction frame. This actuator is
provided with 2 unidirectional hinges, one on each end, placed orthogonally in order to simulate
the behavior of a 3D hinge.
The assembly procedure consisted on the following steps:
 Instrumentation of the micropiles and placement of the grout exit holes protection rubber
ring according to Figure 2.127;
Rubber ring
Protection rings
Figure 2.127: Grout exit holes protection (tube à manchette)
 Placement of the micropiles into the soil container according to the positions of Figure 2.128;
DMT2 DMT1 and 3
Single layout Group layout

Figure 2.128: Micropile positions in layouts – front and top view
118
 Filling of the soil container including registration of the soil weight and medium soil volume
(estimation of the mean index density);
 Load tests of the ungrouted micropiles;
 Grout injection with the use of the prepared pressuring vessel presented in Figure 2.129,
with injection pressure close to 0.2 MPa;
Grout entrance
Air exit
Pressurized air
Grout exit
Figure 2.129: Grout pressure system
 Load tests of the grouted micropiles, 7 days after the grout injection;
 Soil discharging and storage for reuse.
The considered sand used for this purpose was characterized according to the proper standard. The
soil used in the tests is a poorly graded sand (SP). The particle size distribution curves, obtained
for several samples, are shown in Figure 2.130. Minimum and maximum values for the unit weight
were determined according to ASTM D4253-00 [51]. Specific gravity was also determined
according to ASTM 854-05 [52] and NP-83 [53]. The main physical properties obtained for the
considered sand are presented in Table 2.43. For the determination of the soil mechanical
properties, triaxial tests were conducted considering four different soil densities and the results
obtained, in terms of friction angle vs. density index obtained by the triaxial tests are presented in
Figure 2.131.
119
nº200 0,075
nº140 0,106
nº60 0,250
nº80 0,180
nº40 0,425
1 1/2'' 37,5
nº20 0,850
ASTM Sieves
nº10 2,00
3/4'' 19,0
nº4 4,75
3/8'' 9,5
1'' 25,0
2'' 50,0
Opening
(mm)
Sedimentation Sieving
100
90
80 Container nº1
Container nº2
Percentage Passing(%) 70
Container nº3
60
Container nº4
50 Container nº5
40 Container nº6
Container nº7
30
Container nº8
20
10
0
0,000 0,000 0,000 0,001 0,010
fine medium coarse fine medium coarse fine medium coarse

Clay
silt sand gravel
Figure 2.130: Sieve analysis curves
Table 2.43: Soil physical properties
d,min d,max min max

emax emin GS
(kN/m3) (kN/m3) (gr/cm3) (gr/cm3)
14.7 18.9 1.50 1.93 0.76 0.37 2.64
48
44
40
' ˚
36
32
28
0 25 50 75 100
ID(%)
Figure 2.131: Friction angle vs. density index (triaxial tests)
In these tests it was used an S355 steel circular tube with 101.6 mm of external diameter and a
wall 3.6 mm thick.
The grout exit holes position, number and diameter changed between each layout, in order to
obtain a more uniform grout distribution along the micropile wall. The diameter on the lower levels
was reduced because there was a higher grout flow on those sections. Figure 2.132 and Figure
2.133 show the considered geometry for each, respectively, single and group layout.
120
300 300 300
Level 1 (Strain Gauges) Level 1 (Strain Gauges) Level 1 (Strain Gauges)
800 800 800

Grout Exit Holes – Level 1 (6Φ6)
Grout Exit Holes – Level 2 (6Φ6) Grout Exit Holes – Level 1 (6Φ6)
800 Grout Exit Holes – Level 1 (3Φ8) 800 Grout Exit Holes – Level 3 (6Φ8) 800 Grout Exit Holes – Level 2 (6Φ6)
800 Grout Exit Holes – Level 2 (3Φ8)
800 Grout Exit Holes – Level 6 (6Φ8) 800 Grout Exit Holes – Level 5 (6Φ8)
Level 4 (Strain Gauges) Level 4 (Strain Gauges)
Level 2 (Strain Gauges)
300 Grout Exit Holes – Level 3 (3Φ8) 300 Grout Exit Holes – Level 7 (6Φ8)
300 Grout Exit Holes – Level 8 (6Φ8)
Layout 2 (M3 and M4) Layout 3 (M5 and M6) Layout 4 (M7 and M8)
Figure 2.132: Single specimens geometry
300 300 300

800 800 800
Grout Exit Holes – Level 1 (6Φ6) Grout Exit Holes – Level 1 (6Φ6) Grout Exit Holes – Level 1 (6Φ6)

Layout 5 (G1) Layout 6 (G2) Layout 7 (G3)
Figure 2.133: Group specimens geometry
Flat Dilatometer Tests (DMT) were performed in three layouts in order to better characterize the
sand placed in the soil container. The DMT1 and DMT2 tests were performed in layouts 7 and 4
respectively, both after the grout injection. The DMT3 test was performed on layout 5 after the
ungrouted tests on the G1 group and before grouting took place.
From what was observed during the tests and by the analysis of the results presented on Figure
2.134, it is possible to state that the sand properties improved after the grouting process. Test
DMT3 was performed before grouting and presents lower readings (p0 and p1) and soil parameters
than DMT1 and 2.
121
Corrected readings, p0 (kPa) Corrected readings, p1 (kPa)
0 20 40 60 80 0 50 100 150 200 250 300
0.0 0.0
0.5 0.5
1.0 1.0
Depth (m)
Depth (m)
1.5 1.5
2.0 2.0
2.5 DMT1 2.5
DMT1
3.0 DMT2 3.0 DMT2
DMT3
3.5 3.5 DMT3
Friction angle, ' (º) Constrained Modulus, M (MPa)
20 25 30 35 40 0 5 10 15
0.0 0.0
0.5 0.5
1.0 1.0
Depth (m)
Depth (m)
1.5 1.5
2.0 2.0
2.5 DMT1 2.5 DMT1
3.0 DMT2 3.0 DMT2
DMT3 DMT3
3.5 3.5
Horizontal Stress Index, Kd Material Index, Id
0 2 4 6 8 0.1 1 10
0.0 0.0
0.5 0.5
1.0 1.0
Depth (m)
Depth (m)
1.5 1.5
2.0 2.0 DMT1
2.5 DMT1 2.5 DMT2
3.0 DMT2 3.0 DMT3
DMT3
3.5 3.5 Clay Silt Sand
Figure 2.134: DMT results
The DMTs carried out on grouted layouts allowed a measurement of the properties along all the
height because of the confining pressure caused by the sand in the blade membrane. On the
ungrouted layout that did not happen. It can be observed in Figure 2.134 that at some soil depths
it was not possible to obtain readings because of the low confinement obtained.
For these experimental tests, the grout composition was kept constant between each layout.
The following mix proportions were adopted: water/cement ratio 0.4, with type II: 32.5N Portland
cement, 1% of modified polycarboxylate admixture (high range water reducer); and 1% of
expansive admixture.
Tests were carried out in order to control some properties of the grout such as fluidity, exudation,
volume variation and compression resistance. In Figure 2.135 are presented the results obtained
for each test considered in each layout according to EN445 [54] as well as the correspondent limit
provided by the proper standard (EN 447 [55]).
122
30 5
Flow Time (sec.) 4
Exudation (%)
20
3 Exudation
Limit
2
10
Cone (Imediately After)
Cone (30 Minutes After)
1
Cone Limit
0 0
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Layout Layout
50 6
Compression Strength (MPa)
40 4
Volumetric Variation (%)
30 2
20 0
Compression (7 days)
1 2 3 4 5 6 7
10 Compression (28 days) ‐2
Compression Limit (7days) Volumetric Variation
Compression Limit (28 days) Volumetric Variation Limit
‐4
0
1 2 3 4 5 6 7
‐6
Layout Layout
Figure 2.135: Grout control tests results
Some of the problems observed with the exudation and volume variation results may be due to the
low rotation speed of the electrical mixer, which should be higher to achieve a better mixing of the
grout components.
Finally, the compression resistance was determined, according to EN 447 [55] and EN 196-1 [56],
for all the layouts assembled, both for 7 and 28 days after the mixture and injection. The results
obtained were all satisfactory according to limits imposed by the standard (27 MPa and 30 MPa
respectively for 7 and 28 days), apart the results of layout 6 (22.3 MPa and 24.1 MPa for 28 days)
and in one mixture of layout 7 (22 MPa for 28 days). In these two cases it was considered the full
cross-section of the mold (1600 mm2) but the exudation was excessive on those specimens which
conducted to a considerable soft layer on the top of the specimen which reduced the effective
resistant area.
After the tests were concluded on each layout, the sand was removed, the micropiles were
exhumed and the grout distribution was measured and recorded. Similarly to what occurs to
production micropiles, the distribution of the grout was observed to be very different from test to
test. It was impossible to reproduce the same grout geometries in two different specimens and
consequently to reproduce the same resistant properties. Figure 2.136 and Figure 2.137 show an
illustration of the obtained geometries for all the considered specimens.
123
Layout 2
Layout 3 Layout 4
Figure 2.136: Single tests grout distribution
Layout 5 (G1) Layout 6 (G2)
Layout 7 (G3)
Figure 2.137: Group tests grout distribution
124
It was observed that the grout did not come out of the tube in a uniform way. Because the sand is
so loose the grout spread more horizontally than vertically, as preferable, along the tube walls.
That is the reason why the grout is only connected to the tube walls close to the grout exit holes
region and not along the entire micropile. In Figure 2.136 and for layout 2, it is also possible to
observe a horizontal grout plate that was formed between the two single micropiles.
For the sake of results comparison, it was considered that the failure displacement was 10% of the
micropile diameter (10.16mm) for tension tests and 20% of the pile diameter (20.32 mm) for
compression tests. The difference of the failure displacements is related to the shape of the force-
displacement curves where it can be checked that for 10 mm of displacement, the tensile capacity
is basically fully mobilized while for the compression cases the full mobilization occurs for higher
displacements, in the neighborhood of 20 mm, due to the influence of the end bearing.
Figure 2.138 present the global force displacement curves obtained for the grouted and ungrouted
compression tests. Similarly, Figure 2.139 show the global force displacement curves of the tensile
tests. It may be clearly observed that the load amplitude during the cyclic phase of the loading is
larger for the grouted micropiles which shows the benefit and the improvement due to the grout
injection.
20 20
0 0
Displacement (mm)
Displacement (mm)
‐60 ‐40 ‐20 0 20 40 ‐60 ‐40 ‐20 0 20 40
‐20 ‐20
‐40 ‐40
‐60 ‐60 M3 Cyc+Comp Gr

M3 Comp Ung M4 Comp Gr
M4 Comp Ung
‐80 ‐80 M5 Cyc+Comp Gr
M5 Cyc+Comp Ung
M6 Cyc+Comp Gr
M6 Cyc+Comp Ung
M7 Cyc+Comp Ung M7 Cyc+Comp Gr
‐100 M8 Cyc+Comp Ung ‐100 M8 Cyc+Comp Gr
Axial Force (kN) Limit Displacement
Axial Force (kN) Limit Displacement
Ungrouted tests Grouted tests
Figure 2.138: Force-displacement curves (single tests – compression)
M3 Ten Ung 25 M3 Cyc+Ten Gr 25
M4 Ten Ung M4 Ten Gr
M5 Cyc+Ten Ung M5 Cyc+Ten Gr
M6 Cyc+Ten Ung
20 20
M6 Cyc+Ten Gr
Displacement (mm)
Displacement (mm)
M7 Cyc+Ten Ung M7 Cyc+Ten Gr
M8 Cyc+Ten Ung 15 M8 Cyc+Ten Gr 15
Limit Displacement Limit Displacement
10 10
5 5
0 0
‐20 ‐15 ‐10 ‐5 0 5 10 15 ‐20 ‐15 ‐10 ‐5 0 5 10 15
‐5 ‐5
Axial Force (kN) Axial Force (kN)
Figure 2.139: Force-displacement curves (single tests – tension)
The global force-displacement curves of the compression tests on group specimens are presented
on Figure 2.140 while the global force-displacement curve of the tensile groups tests are presented
on Figure 2.141. The beneficial effects of the grouting may be observed, as the load amplitude
during the cycles is, once again, higher for the grouted micropiles.
125
20 20
0 0
Displacement (mm)
Displacement (mm)
‐200 ‐150 ‐100 ‐50 0 50 ‐200 ‐150 ‐100 ‐50 0 50
‐20 ‐20
‐40 ‐40
G1 Comp Ung G1 Comp Gr
G2 Comp Ung G2 Comp Gr
G3 Comp Ung ‐60 G3 Comp Gr ‐60
‐80 ‐80
Figure 2.140: Force-displacement curves (group tests – compression)
G1 Ten Ung 25 G1 Ten Gr 25
G2 Ten Ung G2 Ten Gr
G3 Ten Ung 20 G3 Ten Gr 20
Displacement (mm)
Displacement (mm)
15 15
10 10
5 5
0 0
‐40 ‐20 0 20 40 60 ‐40 ‐20 0 20 40 60
‐5 ‐5
Figure 2.141: Force-displacement curves (group tests – tension)
The beneficial effects of the grouting may be observed, as the load amplitude during the cycles is,
once again, higher for the grouted micropiles.
The results obtained, in terms of resistance and initial (or post-cyclic when applicable) stiffness are
presented in Table 2.44 and Table 2.45, respectively, for single and group tests.
126
Table 2.44: Single tests resistance and static/post-cyclic stiffness
Resistance Stiffness
Layout Micropile Loading Grout
(kN) (kN/m)
Compression No -4.04 662
Tension No 2.28 1053
M3 Cyclic +
Yes -17.37 8333
Compression
Cyclic + Tension Yes 2.66 7692
2
Compression No -2.92 114
Tension No 2.53 1220
M4
Compression Yes -22.07 885
Tension Yes 7.34 1176
Cyclic +
No -5.66 1724
Compression
Cyclic + Tension No 3.58 2273
M5
Cyclic +
Yes -42.91 14286
Compression
3
Cyclic +
No -5.82 806
Compression
M6
Cyclic +
Yes -21.87 6250
Compression
Cyclic +
No -4.73 1136
Compression
M7
Cyclic +
Yes -34.63 6667
Compression
4
Cyclic +
No -6.68 1667
Compression
M8
Cyclic +
Yes -31.81 9091
Compression
127
Table 2.45: Group tests resistance and static/post-cyclic stiffness
Group resistance Stiffness

Layout Group Loading Grout
(kN) (kN/m)
Cyclic +
No -26.26 11111
Compression
5 G1
Cyclic +
Yes -145.96 17544
Compression
Cyclic +
No -23.91 20000
Compression
6 G2
Cyclic +
Yes -145.47 50000
Compression
Cyclic +
No -26.14 10000
Compression
7 G3
Cyclic + (higher than) -
Yes 50000
Compression 177.35
Table 2.46 presents the values obtained for the unit skin friction on the tension tests. It is
presented, for each test, the percentage of grouted length in comparison with the embedded
length, the correspondent medium diameter of the grout and the mean unit skin friction (qs) for
each case. The value of mean unit skin friction presented for the grouted tests refers to the
grouted length of the micropile while in the remaining portion it was assumed a value of unit skin
friction similar to the value presented for the ungrouted tests.
Table 2.46: Unit skin friction
Medium Diameter Grouted Length qs

Layout Micropile/Group Grout
(mm) (%) (kPa)
No 101.6 0 2.6
M3 a b c
2 Yes 156.0 0 3.0
No 101.6 0 2.9
M4 a a
Yes 156.0 35 11.7 c
No 101.6 0 4.0
M5
Yes 163.0 37 7.0 c
3
No 101.6 0 -
M6 a a
Yes 156.0 35 -
No 101.6 0 6.0
M7
Yes 158.0 38 15.9 c
4
No 101.6 0 6.3
M8
Yes 146.0 31 12.9 c
No 101.6 0 1.9
5 G1
Yes 140.0 27 7.1 c
No 101.6 0 1.8
6 G2
Yes 147.0 42 8.5 c
No 101.6 0 1.9
7 G3
Yes 142.0 36 19.3 c
a
value not measured – mean value from M5, M7 and M8 specimens
b
assumed value due to the low grout distribution along the pile outside wall
128
c
mean unit skin friction on the grouted length. The mean unit skin friction on the ungrouted length of the
micropile is similar to the value of the ungrouted specimens
The mean unit skin friction ranged from 2.6 to 6.3 kPa for the single ungrouted micropiles and from
3.0 to 15.9 kPa for the grouted single specimens. For these soil properties, considering a medium
density of 1.58 g/cm3, a friction angle of 33.8º and a friction angle between the soil and the
ungrouted pile of 20º, the resulting mean unit skin friction of is 3.8 kPa. For the grouted micropiles
it was obtained a value of 7.0 kPa considering the same soil properties except a friction angle
between the soil and the ungrouted pile of 33.8º. Both of the unit skin frictions estimated for
grouted and ungrouted tests are located in the interval of values obtained in the experimental
tests.
In the same manner, the unit skin friction varies between 1.8 and 1.9 kPa for ungrouted group
specimens and between 7.1 and 19.3 kPa on the grouted group cases.
The observation of Table 2.46 shows that the unit skin friction of the piles in the groups is lower
than the single micropiles and that is the reason why the group efficiency in tension is lower than
100%. The explanation to this fact lies on the pile installation procedure. As the groups were
placed before the sand, the access to pluviate the sand on the central area between the piles was
limited, and the sand was likely on a looser state, conducting to lower friction angles and
consequently to lower values of the unit skin friction.
The values obtained for each test where the cyclic loading was considered are presented on Figure
2.142 for the single tests and on Figure 2.143 for the group tests.
Cycle Stiffness ‐ M3 Cycle Stiffness ‐ M5
12 12
x 104
x 104
10 10
8 8
Kc (kN/m)
Kc (kN/m)
6 6
4 4
2 2
0 0
1 2 3 4 5 1 2 3 4 5
Cycle Cycle
Cycle Stiffness ‐ M6 Cycle Stiffness ‐ M7
12 12
x 104
x 104
10 10
8 8
Kc (kN/m)
Kc (kN/m)
6 6
4 4
2 2
0 0
1 2 3 4 5 1 2 3 4 5
Cycle Cycle
Cycle Stiffness ‐ M8
12
x 104
10
8
Kc (kN/m)
6
4
2
0
1 2 3 4 5 6 7 8 9 10
Cycle
Figure 2.142: Single tests cyclic stiffness
129
Cycle Stiffness ‐ G1 Cycle Stiffness ‐ G2
350 350
x 104
x 104
300 300
250 250
Kc (kN/m)
Kc (kN/m)
200 200
150 150
100 100
50 50
0 0
1 2 3 4 5 1 2 3 4 5
Cycle Cycle
Cycle Stiffness ‐ G3
350
x 104
300
250
Kc (kN/m)
200
150
100
50
0
1 2 3 4 5
Cycle
Figure 2.143: Group tests cyclic stiffness
From the analysis of Figure 2.142 and Figure 2.143 it is possible to observe, as expected and in the
same manner as in the initial/post-cyclic stiffness, that the grouted specimens presented a higher
cyclic stiffness than the correspondent ungrouted specimen, both for single and group tests. For
this analysis, the mean value for the stiffness of the 5 cycles (or 10 cycles in the case of M8
micropile) was considered.
The grouted specimens show higher resistances than the corresponding ungrouted cases. This
improvement is more emphasized in compression than in tension tests.
The gain in the resistance is expected and it is due to the increase of grout-to-ground bond
strength, the micropile diameter and the improvement of the soil characteristics as was observed in
the DMT tests.
In both the single and the group tests, it was observed that the grout causes improvement in the
specimens’ resistances, as the larger cycles (higher force amplitudes) were obtained. This was
expected because of the force required to mobilize an imposed displacement was higher in grouted
specimens.
In the large majority of the tests, with exception of M6 and M7, the compression resistances are
higher than the correspondent tension resistance due to the influence of the tip effect. Both for
single and group tests, the differences are higher for grouted than for ungrouted specimens. For
higher displacements of M6 and M7, higher compression than tension resistances was obtained,
which was in agreement with the rest of the tests.
The spacing effect is not significant for the ungrouted tests because the increase in the micropiles
spacing lead to very reduced differences between the ungrouted sets. The differences observed for
grouted specimens are higher but in this case it is not possible to state that the tests conditions are
similar from each layout because, as it was presented, there are differences between the grout
distributions from test to test.
A comparison between the mean values of resistance of the single specimens and the
correspondent group specimens (group effect) showed that the efficiency coefficient is higher than
100% for the compression tests, in the case of the ungrouted tests. If the results from the
specimens without cyclic loading, prior the monotonic were neglected, the efficiency coefficient is
close to 100%.
The values obtained for the mean unit skin friction on single micropiles, both grouted and
ungrouted, fall quite well within the analytical estimations performed ahead of the tests. The values
for the unit skin friction for the single micropiles when compared with the group specimens were, in
130
average, higher 132% for ungrouted tests and 29% for grouted tests. The looser sand in the
middle of the group specimens may have conducted to lower friction angles.
The obtained static monotonic stiffness (or post-cyclic when applicable) is, in average, 490%
higher for the grouted than for the ungrouted specimens. The improvement of the grout is more
relevant on the single micropiles where an average improvement of 630% was obtained against an
improvement of 200% for the group cases.
The monotonic stiffness of the specimens with cyclic loading was in average 476% higher than the
obtained for specimens without cyclic loading, unlike expected.
In the case of the cyclic stiffness it was concluded that, for every case studied, the grouted
specimens provided higher stiffness than the correspondent ungrouted, both for compression and
tension loadings.
The comparison between the compressive and the tensile cyclic stiffness shows that for the
ungrouted single tests, the tensile cyclic stiffness is higher than the respective compressive, while
for the grouted tests an improvement was observed between the compressive and tensile tests.
This is due probably to an increase in the sand density after the compression ungrouted tests
resulting in higher tensile ungrouted cyclic stiffness, while in the grouted cases it may occur a
detachment of the some grout after the compression grouted tests conducting to lower tension
grouted cyclic stiffness values.
On the group tests, taking as the exception the G1 group, the compressive stiffness is higher, both
for ungrouted and grouted tests.
As the relevance of the grouting in both strength and stiffness of the micropiles was shown, it is
therefore very important to achieve the best grouting procedure in order to optimize use of the
micropile and promote the highest ground-to-grout bond strength.
2.3.8.6 LCA analysis (shallow vs. hybrid foundations)

From the results presented on the previous chapters related with the foundation dimensions both
for shallow as for hybrid foundations, a LCA analysis was carried out in order to access about the
benefits of the introduction of micropiles as a feasible reinforcement for the current wind tower
foundations.
The geometries and the materials considered for each foundation studied are according to Table
2.40 and Table 2.41, respectively for shallow and hybrid foundations.
The analysis was performed by the software GaBi [57]. The Life-Cycle Assessment includes the
production of materials for the foundations, their transportation to the construction site, the
excavation of soil for the foundations and its deposition into a landfill. For the hybrid foundations,
the use of the drilling equipment and respective fuel consumption were also taken into account.
According to the information provided from a contractor, a drilling rate of 6 m/hour and a fuel
consumption of 8 liters/hour were considered. It was also assumed no maintenance for the
foundations during the service life of the towers.
After a sensitive analysis and due to the negligible importance of the transportation process, the
distances from the production plant to the construction site were considered to be 10 km for all
materials.
The life cycle analysis was carried out according to ISO standards 14040 [58] and 14044 [59]. The
CML methodology [60] was used for the quantification of the following environmental categories:
acidification potential, eutrophication potential, global warming potential, ozone depletion potential
and photochemical ozone creation potential. In addition, an indicator expressing the total primary
energy demand was considered. Hence, the environmental categories selected for the analysis are
summarized in Table 2.47.
Table 2.47: Environmental indicators considered for LCA
Indicator Unit
131
Acidification Potential (AP) kg SO2-Equiv.
Eutrophication Potential (EP) kg Phosphate-Equiv.
Global Warming Potential (GWP) kg CO2-Equiv.
Ozone Layer Depletion Potential (ODP) kg R11-Equiv.
Photochem. Ozone Creation Potential (POCP) kg Ethene-Equiv.
Primary energy demand MJ
The results of the life cycle analysis, taking into account the environmental categories, are
indicated in Table 2.48 and Table 2.49 for the shallow and hybrid foundations respectively. The
lower values are highlighted in bold, for each environmental category.
Table 2.48: Results of the environmental analysis for the shallow foundations
POCP [kg
AP [kg EP [kg GWP [kg ODP [kg PED
Ethene-
SO2-Eq.] PO4-Eq.] CO2-Eq.] R11-Eq.] (MJ)
Eq.]
2,58E+0 3,36E+0 1,37E+0
Steel tower 3,83E-04 3,37E+01 1,11E+06
2 1 5
Tower
80 m Concrete tower 3,91E+02 4,77E+01 1,96E+05 6,79E-04 5,25E+01 1,71E+06
Hybrid tower 3,18E+02 3,89E+01 1,60E+05 5,54E-04 4,28E+01 1,39E+06
4,89E+0 6,54E+0 2,65E+0
Steel tower 6,67E-04 6,30E+01 2,08E+06
2 1 5
Tower
6,35E+0 8,62E+0 3,49E+0
Steel tower 8,25E-04 8,12E+01 2,68E+06
2 1 5
Tower
Table 2.49: Results of the environmental analysis for the hybrid foundations
AP [kg EP [kg GWP [kg ODP [kg POCP [kg PED

SO2-Eq.] PO4-Eq.] CO2-Eq.] R11-Eq.] Ethene-Eq.] (MJ)
2,10E+0 2,46E+0 9,57E+0 8,53E+
Steel tower 1,46E-03 2,78E+01
2 1 4 05
Tower Concrete 1,49E+0
3,72E+02 4,43E+01 1,71E+05 2,81E-03 4,88E+01
80 m tower 6
9,66E+0
Hybrid tower 2,40E+02 3,00E+01 1,17E+05 1,46E-03 3,11E+01
5
4,01E+0 4,88E+0 1,89E+0 1,60E+
2 1 5 06
6,76E+02 8,33E+01 3,30E+05 3,28E-03 8,87E+01
100 m tower 6
1,81E+0
6
5,55E+0 7,18E+0 2,81E+0 2,23E+
2 1 5 06
1,08E+03 1,43E+02 5,61E+05 4,70E-03 1,37E+02
150 m tower 6
2,84E+0
6
The global warming potential (GWP) aims to quantify the emission of greenhouse gases, such as
CO2 and CH4, to the atmosphere. Due to its major influence in climate change, the results for the
life cycle analysis, focusing on global warming, are illustrated in Figure 2.144 for the shallow
foundations and for the hybrid foundations, considering the three types of towers (steel (S),
concrete (C) and hybrid tower (H)) and for the three heights (80 m, 100 m and 150 m).
132
700
600
500
GWP (Ton CO2 eq.)
400
300
200
100
0
S80 C80 H80 S100 C100 H100 S150 C150 H150
Shallow foundations Hybrid foundations
Figure 2.144: Comparison of global warming potential (GWP) for shallow and hybrid
foundations
In addition, the comparative analysis between shallow and hybrid foundations, in terms of the total
primary energy demand, is indicated in Figure 2.145. Likewise, it is observed from Figure 2.145
that hybrid foundations achieve a lower value for the primary energy demand, independently of the
type and height of the tower.
6000
5000
Primary energy (GJ)
4000
3000
2000
1000
0
S80 C80 H80 S100 C100 H100 S150 C150 H150
Shallow foundations Hybrid foundations
Figure 2.145: Comparison of primary enegy demand (PED) for shallow and hybrid
foundations
The results provided by the LCA pointed for a lower environmental impact in the hybrid foundations
in comparison with the correspondent shallow, for every structure geometries and typologies
considered as well as for every environmental indicators considered in the analysis. The main
reason for the better environmental performance is the reduction of the mass achieved by hybrid
foundations, in spite of the use of additional equipment.
It was found a reduction of 10% to 30% of the global warming potential by using hybrid instead
shallow foundations, while for the primary energy demand a reduction of 9% to 25% was achieved.
2.3.9 Design recommendations for hybrid foundations (shallow reinforced with

micropiles)
Micropiles are a deep foundation system used both of new foundations as in retrofitting of older
foundations. The classification of the micropiles is based on the type of grout injection considered.
Table 2.50 and Figure 2.146 represent the types of micropiles considered by the Federal Highway
Administration design and construction guide [30] where the American construction and design
recommendations for micropiles are presented. In the European standards, EN 14199:2005
provides the specifications for micropile execution and EN 1997-1:2004 has the applicable design
specifications.
133
Table 2.50: Classification of micropiles [30]
Type A The grout is placed under gravity only.
Type B The cement grout is placed into the hole under pressure as the temporary steel drill
casing is withdrawn
Type C Primary grout placed under gravity head, then one phase of secondary “global”
pressure grouting
Type D Primary grout placed under gravity head or under pressure, then one or more phases
of secondary “global” pressure grouting
Type E Self-boring micropiles
Figure 2.146: Classification of micropiles [30]
The design of hybrid foundations is very similar to the design of shallow foundations after removing
the portion of the loads resisted by the micropiles.
The main verifications to consider for the design hybrid foundations are the following:
 Soil bearing capacity according to EC7-1 [45], EC8-5 [46] and DNV [47] - In this case, the
bearing capacity was estimated by determining the percentage of load transferred for the
shallow portion of the foundation and for the micropiles.
In order to evaluate the percentage of the load resisted by the micropiles, a numerical model may
be used to simulate the foundation supported by springs. These are used to simulate both the soil
reaction and the effect of the micropiles, through a soil-foundation interaction analyis. The
experimental tests carried out provide valuable information for the determination of the spring
parameters as it will be possible to extrapolate the obtained results for a full scale element and so
reproduce accurately the micropile behaviour.
The tests performed show that the skin friction on ungrouted micropiles in cohesionlles soil may be
determined using the following expression:
q s  K   'v tg ( ) 20
where K is the horizontal stress coefficient, based on the at-rest coeffient K0 and will typicaly vary
from 0.8 to 1.5 K0 for bored and driven ungrounted micropiles, ’v is the effective vertical stress
134
and  is the micropile/soil friction angle, wich depends on the material. On the tests performed
K=0.44 and =20º.
For the grouted micropiles, the tests showed an increase in the side friction and it is possible to
conclude that in a full size prototype the ultimate grout-to-ground resistance depends on the type
of micropile and ground conditions. The proposals of Bustamante and Doix [32] and FHWA [30] are
adequate but should be complemented with control load tests.
Regarding the micropile stiffness, the tests performed in Histwin2 showed that the controlling
parameters are the soil’s deformability modulus E (or the shear modulus, G) and the grouting
process. For the numerial analysis, hyperbolic models such as the proposed by Randolph and Wroth
and modified by McVay et al. [34] are applicable and may be used to define a simpler bi-linear
curve:
 0 r0   rm     rm  r0   r0 0 R f
z  ln    and  21
Gi   r0    rm   r0     f
where z is the vertical displacement, Gi is the initial shear modulus, 0 is the mobilized skin-
friction, f is the ultimate skin-friction, r0 is the micropile radius, Rf is the failure ratio from the
hyperbolic model and rm is the integration limit.
For the number of cycles performed, there was not a substancial loss of stiffness, but further
research is required in order to extrapolate for full scale prototypes.
The axial force obtained in each micropile must be lower than the minimum value between the
internal (structural) and the external (geotechnical) resistances of the elements. For EC7 Design
Approach 1, the external resistance must determined both for EQU/type 1 and type 2
combinations. If the seismic loading is taken into account the external resistance must also be
determined for this load case.
The remaining portion of the loads must be supported only by the shallow foundation and the
bearing capacity under the foundation must be verified for this situation.
 Limit state of equilibrium according to EC7-1 [45]
This verification is fulfilled if the axial resistance in the micropiles never surpass the correspondent
resistance (internal and external).
 Micropile spacing
Related to the micropile spacing, it is better to guarantee a distance of 2 to 3 times of the

micropiles diameter in order to avoid group effect phenomena and so obtain higher individual
resistances.
 Base sliding
In this case it may be assumed that the base sliding under the foundation is supported only by the
friction between the shallow portion and the soil. If this requirement is not fulfilled, the mobilization
of the shear resistance of the micropiles is also possible.
2.3.10 Accessories in steel towers

Following accessories can be found in current tubular steel wind towers:
• Door opening and External Door
• Baskets for power cables
• Ladders
• Resting platforms
• Internal Platforms
• Protection barriers
135
• Welded studs
• Wiring
• Power cables and bus
• Lift
• Earth connections between segments
• Anchor Points
• Oil retaining vales and welded platforms
The solutions used for the accessories and for the assembling and connection inside the tower
depend on the wind turbine producer and are usually proprietary solutions. An overview of the
most common solutions is given in the deliverable D.9, together with some specific proposals to
improve efficiency in terms of their assemblage in factory.
2.4 Conclusions
Advantages of steel towers with bolted friction connection (FrC) compared to the welded flange
connection (WFC) have been shown. The material saving by the use of FrC enables to reduce
environmental impact up to 16 %. The reduction of impact increases with the height of the tower.
Moreover, the use of FrC enables for the complete dismantling of the tower and its re-use. The re-
use of towers (partially or completely) is an important advantage of steel towers in comparison to
other types of towers, particularly in relation to the environmental category of global warming
which becomes increasingly important as the wind power sector becomes mature and change of
the first generation of towers become an economical issue.
During the project, a new FEA analysis approach has been implemented, a model beyond the state
of art is developed. This model is used to completely explain the behaviour of the friction
connection with opened slotted holes and provides theoretical confidence for design of the
connection.
A feasibility study has confirmed the possibility of assembling the towers, which is an important
step towards implemantion of the modularized concept in practice.
2.5 Exploitation and impact of the research results

The project results have improved technological readiness level of the innovative modularized
bolted towers on micropiles from TRL3, starting with laboratory tests of the down-scalled
components to TRL 4 by proving that integrated concept has potential of working in an operational
environment. Results were generalized into design guidance for connections and foundations.
Corresponing increal in the manufatring readiness level of MRL 5 is proveen by the end user of the
project results, Martiefer. Their manufacturing capabilitites for the modularized bolted tubular
towers for hub heights more 100 m is demonstrated by the feasibility test.
2.5.1 Actual applictations

Based on the feasibility tests the possibility of using modularized towers is proven. Plans to build
up the pilot tower using HISTWIN2 concept are discussed with our industrial partners. However,
due to the economical situation in the primary market area and the restructiunhg of the steel
producer development strategies the concept has not been implemented in commercial application.
2.5.2 Technical and economic potential for the use of the results
Based on the research results it may be concluded that the technical, economic and manufacturing
potential for the use of the results is rather big. Analysis of the market potential savings is not
explicitly addressed in the report because it is out of the project scope. However, an estimation of
possible savings of about 100 mil € per year on European market, if all towers are built in the way
proposed in the report, is stated. This estimation is based only on comparison of prices of the ring
flanges and costs of the bolted connection for the friction connection [1] which were considered in
136
the HISTWIN project and it is based on Scandinavian best practice. Cost estimates are prone to
rapid changes. All fabrication technologies, including Submerged Arch Welding (SAW) and Flux
Cored Arch Welding (FCAW) are constantly improving market competiveness and contribute to
decrease of the LCOE (Levelized cost of electricity) generated from wind power. These efforts
accompanied with innovation in the tower execution reduce needs for subsidies. The project has
been focused to increase competiveness of bolted connections by introducing the innovative
detailing for execution of the towers higher than 100 m without considering of advanced of other
technologies.
Very good response of the engineering community is received. A new production facility of the
polygonal steel towers with longitudinal bolted connection is built in Denmark, http://andresen-
towers.com/. A similar concept exists in USA, http://northstarwind.com/ which is marketed by
http://alphatec.ind.br/site/ in South America.
2.5.3 Any possible patent filing

No patent application has been undertaken.
2.5.4 Publications / conference presentations resulting from the project

2011
Baniotopoulos, C.C., Lavassas, I., Nikolaidis, G. & Zervas, P. (2011), On the Structural Design of
Large Wind Turbine Towers in Earthquake Areas, Proc. Of the seventh international conference on
the behavour of steel structures in seismic areas, Santiago de Chile 9-11.1.11.
Fontoura, B., Jesus, A., Correia, J., Silva, A., Matos, R., Rebelo, C. and Simões da Silva, L.:
Comparação da resistência à fadiga entre o aço S355 e o aço de alta resistência S690, 2011, Proc.
2º Congresso Nacional sobre Segurança e Conservação de Pontes, Coimbra, June 29 – July 1 (in
portuguese)
Fontoura, B., Matos, R., Jesus, A.M.P., Rebelo, C., Simões da Silva, L. and Veljkovic, M.: Avaliação
da Resistência à Fadiga de Ligações por Atrito em Aço Estrutural de Alta Resistência S690, 2011,
Proc. VIII Conferência de Construção Metálica e Mista, Centro Cultural Vila Flor, Guimarães,
Portugal, 24-25 November (in portuguese)
Heistermann, C.: Behaviour of pretensioned bolts in friction connections: towards the use of higher
strength steels in wind towers, May 2011, Luleå: Luleå Tekniska Universitet, 123 p. (Licentiate
thesis/Luleå University of Technology)
Heistermann, C. & Veljkovic, M.: Loss of pretension in bolted connections: influence of loading,
Sept. 2011, Proceedings of the 6th European Conference on Steel and Composite Structures:
Eurosteel 2011, August 31 – September 2, 2011, Budapest, Hungary. Budapest: European
Convention for Constructional Steelwork, ECCS, p. 399 – 404, 6 p.
Lavassas, I., Nikolaidis, G., Zervas, P. & Baniotopoulos, C.C. (2011), Design of Large Scale Wind
Turbine Towers in Seismic Areas, Procedings of the XIIth Hellenic Metal Structures Conference,
Volos 29.9.-1.10.11, 272-279.
Limam, M., Veljkovic, M., Bernspång, L., Rebelo, C. and Simões da Silva, L.: Modelling of friction
connection for wind towers using finite element methods, 2011, Eurosteel 2011, August 31 -
September 2, Budapest, Hungary
Matos, R., Fontoura, B., Rebelo, C., Jesus, A., Veljkovic, M., and Simões da Silva, L.: Fatigue
behavior of steel friction connections: Experimental and Numerical Results, Eurosteel 2011, August
31 - September 2, Budapest, Hungary
2012
Heistermann, C., Heistermann, T., Limam, M. & Veljkovic, M.: Finite element analysis of a single
lap joint, 2012, Nordic Steel Construction Conference 2012, September 5 – 7, 2012, Oslo, Norway:
Proceedings. Oslo: Norwegian Steel Association, 673 – 682, 10 p.
137
Heistermann, C. & Veljkovic, M.: Bolts for slip resistant joints in towers for wind turbines, Dec.
2012, Proceedings of the METNET Seminar 2012 in Izmir, Metnet Annual Seminar in Izmir, Turkey,
10 – 12 October 2012, Virdi, K. & Tenhunen, L. (eds.), Hämeenlinna, Finland: HAMK University of
Applied Science, CH. New Steels and Bolted Connections, p. 31 – 39, 9 p.
Jesus, A., Matos, R., Fontoura, B., Rebelo, C., Simões da Silva, L. and Veljkovic, M.: A comparison
of the fatigue behavior between S355 and S690 steel grades, 2012, Journal of Constructional Steel
Research, 79
Limam, M., Heistermann, C. & Veljkovic, M.; Finite element analysis of a single shear lap joint
connection, 2012, Nordic Steel Construction Conference 2012, September 5 – 7, 2012, Oslo,
Norway: Proceedings. Oslo: Norwegian Steel Association, 8 p.
Rebelo, C., Simões, R., Matos, R., Simões da Silva, L., Veljkovic, M. and Pircher, M.: Structural
monitoring of a wind turbine steel tower, 2012, Proc. 15th International Conference on
Experimental Mechanics, Porto, 22-27 July 2012
Rebelo, C., Veljkovic, M., Simões da Silva, L., Simões, R. and Henriques, J.: Structural monitoring
of a wind turbine steel tower – Part I: system description and calibration, 2012, Wind and
Structures, Vol.15, No.4
Rebelo, C., Veljkovic, M., Matos, R.. and Simões da Silva, L.: Structural monitoring of a wind
turbine steel tower – Part II: monitoring results, 2012, Wind and Structures, Vol.15, No.4
2013
Alves, C.: Fatigue behavior of wind towers: comparative analysis of flange connections of hybrid
steel-concrete wind towers, March 2013, MSc thesis, University of Coimbra
Figueiredo, G.: Structural behavior of hybrid lattice – tubular steel wind tower, July 2013, MSc
thesis, University of Coimbra
Heistermann, C., Veljkovic, M. Simões, R., Rebelo, C. & Simões da Silva, L.: Design of slip resistant
lap joints with long open slotted holes, 2013, Journal of Constructional Steel Research, 82, p 223 –
233, 11 p.
Makarios, T. & Baniotopoulos, C.C. (2013), Torsional-Translational Behaviour of Irregular Wind

Energy Structures, Proceedings of the 2013 World Congress on Advances in Structural Engineering
anc Mechanics (ASEM 2013), Jeju, Korea, 8.12.09.13
Matos, R., Cruz, J., Rebelo, C. and Veljkovic, M.: Feasibility Tests on Single Shear Lap Friction
Connections for Wind Towers, 2013, Proc. IX Congresso de Construção Metálica e Mista & I
Congresso Luso-Brasileiro de Construção Metálica Sustentável, Matosinhos, 24-25 October, pp:
665-673, paper 74
Matos, R., Pinto, P., Rebelo, C. and Veljkovic, M.: Laboratory Testing of Single Micropiles. 2013,
Proc. IX Congresso de Construção Metálica e Mista & I Congresso Luso-Brasileiro de Construção
Metálica Sustentável, Matosinhos, 24-25 October, pp: 685-694, paper 64
Matos, R. and Rebelo, C.: Dynamic Measurements of a Steel Wind Tower, 2013, Proc. ICOVP 2013
– 11th International Conference on Vibration Problems, Lisbon, 9-12 September, pp: 308 (digital
paper nr. 470)
Moura, A., Rebelo, C. and Gervásio, H.: Análise comparativa de torres eó1icas em aço e híbridas
aço-betão para alturas e potências variáveis, 2013, II Congresso Luso-Africano de Construção
Metálica Sustentável, Maputo, Mozambique, July 19 (in portuguese)
Pires, P.: Design of concrete-steel transitions in a hybrid wind turbine tower, July 2013, MSc thesis,
University of Coimbra
Secer M., Efthymiou, E. & Baniotopoulos, C. C.: Wind Loading Evaluation on Steel Turbine Towers:
Codified Provisions and Standards, Proceedings of the International Conference of Wind Energy
Science and Technology, RUZGEM 2013, Ankara, 3.-4.10.13
138
Tran, A. T., Veljkovic, M., Rebelo, C. & Simões da Silva, L.: Resistance of door opening in towers
for wind turbines, Jun. 2013, Proceedings of SEECCM III: 3rd South-East European Conference on
Computational Mechanics - an ECCOMAS and IACM Special Interest Conference, Kos Island,
Greece 12 – 14 June 2013, Papadrakakis, M., Kojic, M., Tuncer, I. & Papadopoulos, V. (eds.)
Zygomalas, I., Baniotopoulos, C. C.: Life Cycle Assessment of Traditional Concrete and Steel
Micropile Foundations for Wind Energy Converters, Proceedings of the 3rd International Exergy,
LCA and Sustainability Workshop & Symposium (ELCAS3), Nisyros 7-9.7.2013, 67-75
2014
Gervásio, H., Rebelo, C., Moura, A., Veljkovic, M. and Simões da Silva, L.: Comparative life-cycle
assessment of tubular wind towers and foundations. Part 2 – Life-cycle analysis, Engineering
Structures, (in print)
Heistermann, C., Tran, A. T., Veljkovic, M. & Rebelo, C.: Flangeless Connections in steel tubular
wind towers, Jan. 2014, Proceedings of the METNET Seminar 2013 in Luleå, Sweden, on 22 -23
October 2013, p 157 – 168, 12 p.
Lavasas, I., Nikolaidis, G., Zervas. P. & Baniotopoulos, C. C.: A Comparative Study of Door
Strengthening Systems of a 2 MW Wind Turbine Steel Tower, Proceedings of the National
Conference of Metal Structures, Tripoli, 2.-4.10.14 (in print).
Makarios, T. & Baniotopoulos, C. C.: Wind Energy Structures: Modal Analysis by the Continuous
Model Approach, Journal of Vibration and Control 20 (3), 395-405.
Makarios T. & Baniotopoulos, C. C.: Wind Turbine Towers with Partially Fixed Foundation: Modal
Analysis by the Continuous Model Approach, Earthquakes & Structures (submitted).
Matos, R., Pinto, P.L., Rebelo, C., Veljkovic, M. and Simões da Silva, L.: Laboratory testing of
micropiles in loose sand (submitted for publication in the Canadian Geotechnical Journal, Dec
2013)
Rebelo, C., Moura, A., Gervásio, H., Veljkovic, M. and Simões da Silva, L.: Comparative life-cycle
assessment of tubular wind towers and foundations. Part 1 – Structural design, Engineering
Structures, (in print)
Zygomalas, I. & Baniotopoulos, C. C.: Wind Turbine Steel Towers Environmental Impact
Assessment, Proceedings of Eurosteel 2014, Napoli, 10.-12.09.2014 (in print).
2.5.5 Any other aspects concerning the dissemination of results

The project results were presented by RWTH, UC and LTU, at the most prestigious event in this
area: Advances in Wind Turbine Towers, Bremen, the last week of august. The consortium was
present in 2012, 2013 and will take part in 2014, http://www.windturbine-towers.com/.
139
List of figures
Figure 2.1: Type of connections used in steel tubular towers
Figure 2.2: Coordinate system used for the design of wind turbines; a) Hybrid tower; b) Steel
or concrete tower
Figure 2.3: Life cycle analysis of a wind tower.
Figure 2.4: Life cycle analysis of wind towers (1st scenario).
Figure 2.5: Life cycle analysis of wind towers (2nd scenario).
Figure 2.6: Life cycle analysis of wind towers (3rd scenario).
Figure 2.7: a) Lighthouse; b) Mobile crane
Figure 2.8: Cross section imperfection: out-of-roundness
Figure 2.9: First Eigenmode of a polygonal (dodecagonal) cross section (left hand side) and circular
cross-section (right hand side) [14]
Figure 2.10: First eigenmode of tubes under bending with different cross sections [15]
Figure 2.11: MRd,eff(number of edges): Comparison between polygonal and circular cross section
(reduced stress method)
Figure 2.12 MRd,eff/MRd,eff,circle(Apolygon/Acircle): Comparison between polygonal and circular cross

section (Reduced Stress Method)
Figure 2.13 Imperfection types used in non-linear analysis
Figure 2.14 Flow chart of the design stress method
Figure 2.15 a) Overview of the finite element model and b) detail of the overlapping area with node
coupling
Figure 2..16 deformed shape of the finite element model
Figure 2.17 Load displacement curve for cross section with (SQS) and without (VQS) longitudinal
joint
Figure 2.18 Load displacement curve for cross section with different bolt distances
Figure 2.19 Load displacement curve for cross sections with different bolt distances
Figure 2.20: Polygonal cross-section section with number of edges ranged from 8 to 22 with
increment by 2
Figure 2.21: Tower segment, 4m in diameter, considered in FEA
Figure 2.22: a) Buckling mode and b) failure shape of a polygonal cross-section made of three
segments with 8, 18 and 20 edges and a circular cross-section under axial compressive. All
columns have a 4m diameter.
Figure 2.23: a) Buckling mode and b) failure shape of a polygonal cross-section made of three
segments with 8, 18 and 20 edges and a circular cross-section under bending. (All tubular sections
have 4 m diameter).
Figure 2.24: Ultimate axial resistance N.cR for polygonal cross-sections ranged from 8 to 22 edges
and one circular cross-section. Columns have a diameter of 4 m and 5 m length.
Figure 2.25: Ultimate stresses for polygonal cross-sections ranged from 8 to 22 edges and one
circular cross-section. Columns have a diameter of 4 m and 5 m length.
Figure 2.26: Comparison of the Moment between analytical and numerical analysis for tubes
ranged with 8 adjacent edges to 22 compared with a circular tube.
Figure 2.27: General view and a cross-section of an analysed 5-meter long hexadecagonal element
(dimensions are given in millimetres)
140
Figure 2.28: First two out of 5 analysed positions of axis of row of bolts
Figure 2.29: Imperfections in simplified models
Figure 2.30: Elasto-plastic material model definition
Figure 2.31: Elasto-plastic material model definition Force-displacment curves for models under
compression with constant bolt spacing and different distance of axis of row of bolts from the
longitudinal edge of the model (material model: elastic)
compression with different bolt spacings and with constant 125mm-distance of axis of row of bolts
from the longitudinal edge of the model (material model: elastic)
compression with different elastic and elasto-plastic material models.
Figure 2.34: Elasto-plastic material model definition Comparison of force-displacement curves for
models with different bolt spacings and 2 types of imperfections.
Figure 2.35: Elasto-plastic material model definition Stress distributions (effective stress range 0-
500MPa) for peak loads: a) bolt spacing 416.67mm and 625mm; b) bolt spacing 1000mm
500MPa) for the post-buckling range of loads: a) bolt spacing 416.67mm and 625mm; b) bolt
spacing 1000mm
Figure 2.37: Elasto-plastic material model definition Comparison of force-displacement curves for
models with different bolt spaceings and 2 types of imperfections
500MPa) for peak loads: a) bolt spacing 416.67mm and 625mm; b) bolt spacing 1000mm
500MPa) for the post-buckling range of loads: a) bolt spacing 416.67mm and 625mm; b) bolt
spacing 1000mm
Figure 2.40: Geometry of Models.
Figure 2.41: Boundary conditions.
Figure 2.42: First buckling mode of the models: (a) Model without door opening, (b) Model of a
door opening with varying thickness, (c) Model of a door opening with stiffener.
Figure 2.43: Relation between moment and rotation of models - door opening under compression.
Figure 2.44: Relation between moment and rotation of models - door opening under tension.
Figure 2.45: Distribution of normal stress at ultimate load: (a) Model without door opening, (b)
Model of a door opening with varying thickness, (c) Model of a door opening with stiffener.
Figure 2.48: Comparison of using higher grade steels for stiffener.
Figure 2.49: Comparison of using higher grade steels for door opening.
Figure 2.50: Prestressed friction connection with fingers and slotted holes
Figure 2.51: Loading of the fingers in wind turbines (left) and static system of the fingers (right)
Figure 2.52: General Test set up
Figure 2.53: Test set up of the small scale buckling tests
Figure 2.54: Closing of the gap (left); moment distribution in the finger (right)
Figure 2.55: Load displacement curves for tests with elastic and plastic deformations
141
Figure 2.56: Ultimate load over length of the finger for the test series 1 and 8 and corresponding
Abaqus simulations
Figure 2.57: Load displacement curves for tests with different gaps and fixed length of 350 mm
Figure 2.58: Load displacement curves for tests with different gaps and fixed length of 500 mm
Figure 2.59: Load displacement curves for tests with a thickness of 10, 15 and 20 mm with a fixed
length
Figure 2.60: Load displacement curves for tests with a thickness of 10, 15 and 20 mm with a fixed
length of 500 mm
Figure 2.61: Utilization ratio of the tests and Abaqus calculations over slenderness with only elastic
deformation during closing the gap
Figure 2.62: Utilization ratio of the tests and Abaqus calculations over slenderness for fingers with
plastic deformation during closing the gap
Figure 2.63: Utilization ratio of the tests and Abaqus calculations over slenderness for fingers with
plastic deformation during closing the gap
Figure 2.64: Buckling curve and tested slenderness’s
Figure 2.65: Layout of the tower segment and connection considered in the study.
Figure 2.66: Different tower shapes considered for the friction connection.
Figure 2.67: Different approaches for provision of assembling tolerances.
Figure 2.68: Different lengths of the fingers in the case of circular tower shape with gap.
Figure 2.69: Indentions at the corners of the polygonal tower shape with gap between the shells.
Figure 2.70: Geometry and boundary conditions.
Figure 2.71: Preloading of the bolts.
Figure 2.72: Stress-strain curves used in FE models.
Figure 2.73: Nominal and numerical tensile tests results
Figure 2.74: Plasticity and ductile damage parameters for bolt materials.
Figure 2.75: Four point bending tests of down-scaled wind turbine tower connections [18].
Figure 2.76: FEA models of down-scaled wind tower connection tests.
Figure 2.77: Comparison of the results of verification FEA of ring flange connection tests.
Figure 2.78: Comparison of the results of verification FEA of friction connection tests.
Figure 2.79: Moment-rotation curves for ring flange connection and friction connection.
Figure 2.80: Buckling of the tower shell at the ultimate bending moment.
Figure 2.81: Evolution of membrane and bending stresses in the cylindrical tower shell with friction
and ring flange connection.
Figure 2.82: Comparison of behaviour of ring flange and friction connection.
Figure 2.83: Moment-rotation curves for different length of fingers in friction connection.
Figure 2.84: Post-buckling shape for different lengths of fingers.
Figure 2.85: Stress concentration at the indention root of the polygonal tower shape.
Figure 2.86: Moment-rotation curves for different methods to provide assembling tolerances.
Figure 2.87: a) Conjunction of longitudinal and transversal joint; b) cut of an inner plate
Figure 2.88: Possible overlapping possibilities, left asymmetric; right; symmetric
142
Figure 2.89: Flow chart of the program of the transversal friction connection, including the design
resistance checks according to EN 1993
Figure 2.90: Overview of the Excel-Tool 1B: ‘Pre-Design of longitudinal friction connection’
Figure 2.91 - Different solutions for friction connections
Figure 2.92: HISTWIN prototype assembly
Figure 2.93: Cutting techniques (HISTWIN prototype)
Figure 2.94: Long open slotted holes
Figure 2.95: Indication of the cutting zone
Figure 2.96: Robot cutting trajectory and robot
Figure 2.97: Rolling process
Figure 2.98: Global view of the model
Figure 2.99: Superposition of the plates
Figure 2.100: Overlapping joint of shear and friction connection
Figure 2.101: Cover plates
Figure 2.108: Lifting equipment
Figure 2.109: Thermal cutting process
Figure 2.110: Print screens - obtained of the CNC program
Figure 2.111: Width of the open holes
Figure 2.112: Height of the conn. region
Figure 2.113: Dirt around the holes (after thermal cutting)
Figure 2.114: Method for the estimation of the rolling radius
Figure 2.115: Curvature check pieces
Figure 2.116: 15 mm thick connection assembly
Figure 2.117: Assembly of the lower segment (20 mm thick connection)
Figure 2.118: Final assembly of both segments (20 mm thick connection)
Figure 2.119: Measurement points on the prototype
Figure 2.120: Gap measurement with analogic calliper
Figure 2.121: Gap measurement points
Figure 2.122: Geometry and geometric parameters of the octagonal foundation
Figure 2.123: Patrick and Handerson deep foundations solutions [42]
Figure 2.124: Micropile application on wind turbine foundations [44]
Figure 2.125: Bilinear micropile springs
Figure 2.126: Micropile locations

143
Figure 2.127: Grout exit holes protection (tube à manchette)
Figure 2.128: Micropile positions in layouts – front and top view
Figure 2.129: Grout pressure system
Figure 2.130: Sieve analysis curves
Figure 2.131: Friction angle vs. density index (triaxial tests)
Figure 2.132: Single specimens geometry
Figure 2.133: Group specimens geometry
Figure 2.134: DMT results
Figure 2.135: Grout control tests results
Figure 2.136: Single tests grout distribution
Figure 2.137: Group tests grout distribution
Figure 2.138: Force-displacement curves (single tests – compression)
Figure 2.139: Force-displacement curves (single tests – tension)
Figure 2.140: Force-displacement curves (group tests – compression)
Figure 2.141: Force-displacement curves (group tests – tension)
Figure 2.142: Single tests cyclic stiffness
Figure 2.143: Group tests cyclic stiffness
Figure 2.144: Comparison of global warming potential (GWP) for shallow and hybrid foundations
Figure 2.145: Comparison of primary enegy demand (PED) for shallow and hybrid foundations
Figure 2.146: Classification of micropiles [30]
List of tables
Table 2-1– Scenarios for tower design
Table 2-2: Wind section forces.
Table 2-3: Earthquake load, not combined.
Table 2-4: Damage Equivalent Loads, m = 4, NRef = 2x108 cycles.
Table 2-5: Turbine mass, rotor speed and admissible frequency range.
Table 2-6: Soil characteristics.
Table 2-7: Forces applied at the center of the foundation base
Table 2-8: Governing load cases and design checks.
Table 2-9: Design results for concrete towers.
Table 2-10: Design results for steel towers.
Table 2-11: Design results for hybrid towers.
Table 2-12: Environmental indicators for LCA
Table 2-16: Environmental indicators for the three towers (3rd scenario)
Table 2-17 Ultimate load of 4 m diameter cylindrical cross section.
144
Table 2-18 Ultimate load of 4 m diameter polygonal cross section.
Table 2-19: Geometry and properties of polygonal tubes with 5000 mm length and 8 mm
thickness.
Table 2-20: Analytical vs. numerical analysis results for critical stresses and critical loads under
compressive loading.
Table 2-21: Numerical and analytical results for the compressive resistance load NcR and bending
resistance moment for polygonal cross-sections ranged from 8 to 20 edges and comparison with a
circular cross-section of 4 m diameter.
Table 2-22: Bending resistance moment for analytical and numerical analysis with three
imperfections measurements acc. to EN 1993 part a-6.
Table 2-23: Comparison of eigenvalues between models.
Table 2-24: Relation between thickness, moment and rotation at ultimate load of the models.
Table 2-25: Performed tests.
Table 2-26: Compared tests.
Table 2-27: Elastic and plastic length for the different gap sizes
Table 2-28: Compared tests
Table 2-29: Elastic and plastic length for the different thicknesses
Table 2-30: Tests to compare influence of finger width
Table 2-31: Cases considered in the study.
Table 2-32: Rotations of the nuts and achieved preloading forces.
Table 2-33: Comparison of results for ring flange connection and friction connection.
Table 2-34: Fatigue performance of the bolts in the ring flange and friction connection.
Table 2-35: Influence of the length of the fingers on ultimate bending resistance.
Table 2.36 – Shape measurements
Table 2.37 – Gap measurements
Table 2.38: Load table (base of foundation)
Table 2.39: Soil characteristics
Table 2.40: Octagonal shallow foundations design results
Table 2.41: Octagonal hybrid foundations design results
Table 2.42: Experimental tests description
Table 2.43: Soil physical properties
Table 2.44: Single tests resistance and static/post-cyclic stiffness
Table 2.45: Group tests resistance and static/post-cyclic stiffness
Table 2.46: Unit skin friction
Table 2.47: Environmental indicators considered for LCA
Table 2.48: Results of the environmental analysis for the shallow foundations
Table 2.49: Results of the environmental analysis for the hybrid foundations
Table 2.50: Classification of micropiles [30]
List of acronyms and abbreviations
145
ASTM American Society of Testing and Materials
AUTH Aristotle University of Thessaloniki
CML Institute of Environmental Sciences – University of Leiden
CNC Computer Numerical Control
CO Coordinator
CT Concrete tower
DMT Flat Dilatometer Test
DNV Det Norske Veritas
EC Eurocode
FCTUC Faculdade de Ciências e Tecnologia da Universidade de Coimbra
FEM Finite Element Method
FHWA Federal Highway Administration
FrC Friction connection
HISTWIN High-Strength Steel Towers for Wind Turbines (RFCS project)
HSS High-strength Steel
ISO International Organization for Standardization
LCA Life Cycle Analysis
LTU Lulea University of Technology
NP Norma Portuguesa (Portuguese Standard)
QSA Qualified Security Assessor
RFC Ring Flange Connection
RWTH Rheinsch-Westfälische Technische Hochschule Aachen
WFC Welded flange connection (the same as RFC)
146
List of References
[1] M. Veljkovic, C. Heistermann, W. Husson, M. Limam, M. Feldmann, J. Naumes, D. Pak, T.
Faber, M. Klose, K.-U. Fruhner, L. Krutschinna, C. C. Baniotopoulos, I. Lavassas, A. Pontes,
E. Ribeiro, M. Hadden, R. Sousa, L. Simões da Silva, C. Rebelo, R. Simões, J. Henriques, R.
Matos, J. Nuutinen, and H. Kinnunen, “High-strength tower in steel for wind turbines
(Histwin),” Brussels, 2012
[2] Moura, A., “Comparative analysis of steel and hybrid steel-concrete wind towers for
variable height and power ”, Master Thesis, University of Coimbra, Coimbra 2012
[3] LaNier, M, Berger/ABAM Engineers Inc., “LWST Phase I Project Conceptual Design Study:
Evaluation of Design and Construction Approaches for Economical Hybrid Steel/Concrete
Wind Turbine Towers”, subcontractor report NREL/SR-500-36777, Colorado, 2005.
[4] ISO 14040 (2006). Environmental management - Life cycle assessment - Principles and
framework. International Organization for Standardization. Geneva, Switzerland.
[5] ISO 14044 (2006). Environmental management - Life cycle assessment - Requirements
and guidelines. International Organization for Standardization. Geneva, Switzerland.
[6] Guinée, J.B.; Gorrée, M.; Heijungs, R.; Huppes, G.; Kleijn, R.; Koning, A. de; Oers, L. van;
Wegener Sleeswijk, A.; Suh, S.; Udo de Haes, H.A.; Bruijn, H. de; Duin, R. van;
Huijbregts, M.A.J. Handbook on life cycle assessment. Operational guide to the ISO
standards. I: LCA in perspective. IIa: Guide. IIb: Operational annex. III: Scientific
background. Kluwer Academic Publishers, ISBN 1-4020-0228-9, Dordrecht, 2002, 692 pp.
(www.cml.leiden.edu/research/industrialecology/researchprojects/finished/new-dutch-lca-
guide.html)
[7] Ardente, F., Beccali, M., Cellura, M. and Lo Brano, V. (2008). “Energy performances and life
cycle assessment of an Italian wind farm”. Renewable and Sustainable Energy Reviews 12,
pp. 200–217.
[8] Worldsteel association (2011) “Methodology report – Life cycle inventory study for steel
products”. World Steel Association (worldsteel), Brussels.
[9] GaBi 6 (2012). Software-System and Databases for Life Cycle Engineering. Version 5.56.
PE International AG, Leinfelden-Echterdingen, Germany.
[10] Eurocode 3: Design of steel structures, CEN, European Committee for standardization,
Brussels, Belgium
[11] Eurocode 2: Design of concrete structures, CEN, European Committee for standardization,
Brussels, Belgium
[12] Eurocode 7: Geotechnical Design, CEN, European Committee for standardization, Brussels,
Belgium
[13] Autodesk Robot Structural Analysis 2011
[14] Matzner, I.: “Stabilitäts- und Wirtschaftlichkeitsuntersuchungen kaltgeformter und
kreisförmiger Hohlprofile für den Windenegieanlagenbau” Lehrstuhl für Stahlbau RWTH
Aachen, Study 2007
[15] MSC Marc/Mentat 2005r3 (64bit) Finite Element Software
[16] Abaqus, Abaqus documentation v6.12. Simulia Dassault Systmes, 2012
[17] Limam M. FE Modelling of Friction Connections in Tubular Tower for Wind Turbines.
Licentiate thesis. Lulea, Sweden: Lulea University of Technology: 2011.
[18] Pak D, Naumes J. High-strength tower in steel for wind turbines (HISTWIN): WP2.5 – Large
Scale 4-Point-Bending Tests. Background Document - RFSR-CT-2006-00031. Brussels,
Belgium: European Commission (RFCS); 2010.
[19] Garzon O. Reistance of polygonal cross-sections - Application on steel towers for wind
turbines. Licentiate thesis. Lulea, Sweden: Lulea University of Technology: 2013.
[20] EN1993-1-6: Eurocode 3: Design of steel structures. Part 1-6: Strength and stability of
shell structures: European Committee for Standardization (CEN); 2007.
[21] EN1993-1-1: Eurocode 3: Design of steel structures. Part 1-1: General rules and rules for
buildings. Brussels, Belgium: European Committee for Standardization (CEN); 2005.
[22] EN1993-1-8: Eurocode 3: Design of steel structures. Part 1-8: Design of joints. Brussels,
Belgium: European Committee for Standardization (CEN); 2005.
147
[23] ISO 898-1: Mechanical properties of fasteners made of carbon steel and alloy steel. Part 1:
Bolts, screws and studs. Fourth edition. Brussels, Belgium: European Committee for
Standardization (CEN); 2009.
[24] Trattnig G, Antretter T, Pippan R. Fracture of austenitic steel subject to a wide range of
stress triaxiality ratios and crack deformation modes. Engineering Fracture Mechanics
2008;75(2):223–235.
[25] Rice JR, Tracey DM. On the Ductile Enlargement of Voids in Triaxial Stress Fields. Journal of
Mechanics Physics of Solids. 1969;17:201–217.
[26] Lemaitre J. A Continuous Damage Mechanics Model for Ductile Fracture. Journal of
Engineering Materials and Technology. 1985;107(1):83–90.
[27] Bonora N, Ruggiero A, Esposito L, Gentile D. CDM modelling of ductile failure in ferritic
steels: Assessment of the geometry transferability of model parameters. International
Journal of Plasticity. 2006 November;22(11):2015–2047.
[28] EN1993-1-9: Eurocode 3: Design of steel structures. Part 1-9: Fatigue: European
Committee for Standardization (CEN); 2007.
[29] REPOWER System drawing: R050405-SZ (B); Internal HISTWIN document
[30] Federal Highway Administration (FHWA), Micropile Design and Construction Guidelines, US
Department of Transportation, December 2005.
[31] Shenbaga R Kaniraj, Design aids in Soil Mechanics and Foundation Engineering, Tata
McGraw Hill, 2008, New Delhi.
[32] Bustamante & Doix, "Une méthode pour le calcul des tirants et des micropieux injectés".
Bulletin de liason des laboratoires de Ponts et Chaussés, 140, Novembre-Décembre, pp.
75-92, 1985.
[33] Schmertmann, J. H., "Guidelines for Cone Penetration Test Performance and Design",
Report FHWA-TS-78-209, U.S. Department of Transportation, Washington, 1978.
[34] McVay, M. C., O'Brien, M., Townsend, F. C., Bloomquist, D. G., and Caliendo, J. A.
"Numerical Analysis of Vertically Loaded Pile Groups," ASCE, Foundation Engineering
Congress, Northwestern University, Illinois, July, 1989, pp. 675-690.
[35] FB-MultiPier v4, Florida Bridge Software Institute, University of Florida, USA
[36] Coelho, N., “Caracterização Laboratorial do Comportamento de um Solo Arenoso para a
Previsão de Resistência Axial de Microestacas”, M.Sc. thesis, Universidade de Coimbra,
2011
[37] EN445: Grout for prestressing tendons. Test methods, Portuguese version 2000
[38] EN447: Grout for prestressing tendons. Basic requirements, Portuguese version 2000
[39] EN 196-1: Methods of testing cement – Part 1: Determination of strength, Portuguese
version 2006
[40] Autodesk Robot Structural Analysis 2011
[41] Eurocode 1: Actions on structures, CEN, European Committee for standardization, Brussels,
Belgium
[42] Earth Systems Global, Inc., P&H Foundations for Wind Turbine Support. Available from
http://earthsys.com/Library/P%20and%20H%20Presentations/Patrick_Henderson_Foundat
ions_Pamphlet.pdf, 2009
[43] H. Svensson, “Design of Foundations for Wind Turbines”, M.Sc. thesis, Department of
Construction Sciences – Structural Mechanics, Lund University, Lund, Sweden, 2010
[44] H. Aschenbroich, “Root Pile Support of Vancouver’s Hollow Tree – Parallels to Wind Turbine
Foundations”, In Proceedings of the 10th ISM Workshop, Washington, DC., 2010
[45] CEN, “Eurocode 7 – Geotechnical design – EN 1997 Part 1: General rules”, European
Committee for Standardization, Brussels, 2004
[46] CEN, “Eurocode 8 – Design of structures for earthquake resistance – EN 1998 Part 5:
Foundations, retaining structures and geotechnical aspects”, European Committee for
Standardization, Brussels, 2004
[47] DNV and Risø, “Guidelines for Design of Wind Turbines”, 2nd Edition, Norway, 2002
[48] International Electrotechnical Commission, “IEC 61400 - Wind Turbines – Part:1 Design
requirements”, 3rd edition, 2005-08, Switzerland, 2005
148
[49] CEN, “Eurocode 8 – Design of structures for earthquake resistance – EN 1998 Part 1:
General rules, seismic actions and rules for buildings” European Committee for
Standardization. Brussels, 2004
[50] Robot Structural Analysis, User’s Manual, Version 2011. Autodesk Inc., San Rafael,
California, USA, 2011
[51] ASTM, “Standard Test Method for Maximum Index Density and Unit Weight of Soils Using a
Vibratory Table”, D4253-00 Standard, American Society for Testing and Materials, USA,
2000
[52] ASTM, “Standard Test Methods for Specific Gravity of Soil Solids by Water Pycnometer”,
D854-05 Standard, American Society for Testing and Materials, USA, 2005
[53] LNEC – Laboratório Nacional de Engenharia Civil, “Solos, Determinação da Densidade das
Partículas”, LNEC NP-83 Standard, Laboratório Nacional de Engenharia Civil, Portugal.
Portuguese Version, 1965
[54] CEN – Comité Européen de Normalisation, “EN445: Grout for prestressing tendons - Test
methods”, Portuguese version, 2000
[55] CEN – Comité Européen de Normalisation, “EN447: Grout for prestressing tendons - Basic
requirements”. Portuguese version, 2000
[56] CEN – Comité Européen de Normalisation, “EN 196-1: Methods of testing cement – Part 1:
Determination of strength. Portuguese version, 2006”
[57] GaBi 6, Software-System and Databases for Life Cycle Engineering, Version 5.56, PE
International AG, Leinfelden-Echterdingen, Germany, 2012
[58] ISO 14040, “Environmental management - Life cycle assessment - Principles and
framework”, International Organization for Standardization. Geneva, Switzerland, 2006
[59] ISO 14044, “Environmental management - Life cycle assessment - Requirements and
guidelines”, International Organization for Standardization, Geneva, Switzerland, 2006
[60] J.B. Guinée, M. Gorrée, R. Heijungs, G. Huppes, R. Kleijn, A. de Koning, L. van Oers, A.
Wegener Sleeswijk, S. Suh, H.A. Udo de Haes, H. de Bruijn, R. van Duin, M.A.J. Huijbregts,
“Handbook on life cycle assessment. Operational guide to the ISO standards. I: LCA in
perspective. IIa: Guide. IIb: Operational annex. III: Scientific background”, Kluwer
Academic Publishers, ISBN 1-4020-0228-9, Dordrecht, 692 pp
(www.cml.leiden.edu/research/industrialecology/researchprojects/finished/new-dutch-lca-
guide.html), 2002
149
HOW TO OBTAIN EU PUBLICATIONS
Free publications:
• one copy:
via EU Bookshop (http://bookshop.europa.eu);
• more than one copy or posters/maps:
from the European Union’s representations (http://ec.europa.eu/represent_en.htm);
from the delegations in non-EU countries (http://eeas.europa.eu/delegations/index_en.htm);
by contacting the Europe Direct service (http://europa.eu/europedirect/index_en.htm) or
calling 00 800 6 7 8 9 10 11 (freephone number from anywhere in the EU) (*).
(*) The information given is free, as are most calls (though some operators, phone boxes or hotels may charge you).
Priced publications:
• via EU Bookshop (http://bookshop.europa.eu).
KI-NA-27-226-EN-N
The main achievements of the research project are further development of
existing and new solutions for steel tubular wind towers and the development
and validation of alternative steel-intensive piled foundations. Experimental,
numerical and analytical study performed allow better understanding of the
concept of slotted friction-grip bolted connections proposed in previous
research project, HISTWIN, and its extension to the longitudinal joining of steel
segments that form the proposed ‘modularized’ steel tower for wind turbines.
Polygonal shape of the cross section has been considered numerically and in
down-scaled experiments. The comparison is made with the circular cross-
section. Comparative life-cycle evaluation of steel, hybrid (steel-concrete)
and concrete towers and respective foundations using micropiles has been
performed. Optimization of door opening stiffening has been addressed and
advantages of using higher strength steel grades have been investigated.
Feasibility tests consisting of fabrication and assembling of two downscale
segments of the modularized tower has been performed. Above the state of
art numerical models are used to investigate the structural behaviour of the
tower shell including local and global imperfections, effect of connection,
behaviour of the connection components, and contribution of increasing steel
grade. In addition numerical analysis of contribution of micropiles to foundation
stability is validated by experimental down-scale experiments. The project has
achieved TRL 4 of the innovative modularized bolted towers because the basic
components are integrated and proven to have potential of working together in
an operational environment. Results were generalized into design guidance for
connections and foundations.
Studies and reports
ISBN 978-92-79-48076-8
doi:10.2777/899646

High Steel Tubular Towers For Wind Turbines: (Histwin2)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

High Steel Tubular Towers For Wind Turbines: (Histwin2)

Uploaded by

Copyright:

Available Formats

High steel tubular towers

for wind turbines

Contact: RFCS Publications

Research Fund for Coal and Steel

M. Veljkovic, C. Heistermann, O. Garzon, M. Limam, A.T. Tran, Marko Pavlovic

M. Feldmann, F. Möller, C. Richter

C. Baniotopoulos, S. Gerasimidis, I. Zygomalas

L. Simões da Silva, C. Rebelo, P. Pinto, R. Matos, A. Moura, H. Gervásio

Grant Agreement RFSR-CT-2010-00031

Directorate-General for Research and Innovation

2015 EUR 27226 EN

Europe Direct is a service to help you find answers

Freephone number (*):

More information on the European Union is available on the Internet (http://europa.eu).

Cataloguing data can be found at the end of this publication.

Luxembourg: Publications Office of the European Union, 2015

© European Union, 2015

Printed on white chlorine-free paper

1 FINAL SUMMARY ............................................................................................ 5

2 SCIENTIFIC AND TECHNICAL DESCRIPTION OF THE RESULTS ............................ 13

2.3 Description of activities and discussion ................................................................. 14

2.3.1 Structural analysis of various tower types .......................................................... 14

2.3.1.1 Introduction ........................................................................................... 14

2.3.1.2 Tower loading and design methodology ...................................................... 15

2.3.1.3 Tower and foundation design and results.................................................... 17

2.3.1.4 Design results ........................................................................................ 18

2.3.2 LCA of various tower types .............................................................................. 21

2.3.2.1 Model for Life-Cycle Assessment ............................................................... 21

2.3.2.4 Comments and Conclusions ...................................................................... 28

2.3.3 Steel tubular towers: Alternatives for cross-sections ........................................... 28

2.3.3.2 Influence of longitudinal folds on the buckling stress .................................... 35

2.3.3.3 FEA of polygonal vs. circular tower segments .............................................. 38

2.3.3.4 Optimum spacing of bolts in single-lap connections of a hexagonal element

2.3.3.7 Segment Tests ....................................................................................... 51

2.3.4 Transversal friction connections ....................................................................... 58

2.3.4.1 Finger tests ........................................................................................... 58

2.3.4.2 Derivation of modified buckling curve ........................................................ 68

2.3.5 Longitudinal bolted connections ....................................................................... 89

2.3.6 Design Recommendations for modularized steel tubular towers............................. 90

2.3.7 Feasibility study ............................................................................................. 92

2.3.7.1 Preparation Phase ................................................................................... 93

2.3.7.2 Feasibility Tests ...................................................................................... 94

2.3.7.3 Manufacturing Process ............................................................................. 96

2.3.7.4 Conclusions ......................................................................................... 109

2.3.8.1 Introduction ......................................................................................... 110

2.3.8.2 Alternatives for tower foundations ........................................................... 111

2.3.8.3 Shallow foundations .............................................................................. 112

2.3.8.4 Foundations with steel micro-piles ........................................................... 114

2.3.8.5 Micropiles experimental tests.................................................................. 117

2.3.8.6 LCA analysis (shallow vs. hybrid foundations) ........................................... 131

2.3.10 Accessories in steel towers ........................................................................ 135

2.4 Conclusions .................................................................................................... 136

2.5 Exploitation and impact of the research results .................................................... 136

2.5.1 Actual applictations ...................................................................................... 136

2.5.3 Any possible patent filing .............................................................................. 137

LIST OF FIGURES ............................................................................................. 140

LIST OF TABLES ............................................................................................... 144

LIST OF REFERENCES ....................................................................................... 147

Tower Height and Rated Power of Wind turbine

Bending resistance of the friction connection is analysed in a parametric study, considering:

Main conclusions of this advanced FEA are summarized as follows:

Friction connections – ‘Finger’ tests and design guidance

Optimization of door opening reinforcements in tubular Wind Towers