See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/321935270

Influence of execution tolerances for friction

connections in circular and polygonal towers for
wind converters

Article in Advanced Steel Construction · December 2017

DOI: 10.18057/IJASC.2017.13.4.2

CITATIONS READS

0 165

7 authors, including:

Marko Pavlović Milan Veljkovic

Delft University of Technology Delft University of Technology
30 PUBLICATIONS 98 CITATIONS 169 PUBLICATIONS 558 CITATIONS

SEE PROFILE SEE PROFILE

Daniel Pak Carlos Rebelo

Universität Siegen University of Coimbra
42 PUBLICATIONS 127 CITATIONS 172 PUBLICATIONS 549 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

ULTIMATEPANEL (2016-2019): Curved thin panels for structural application. Portuguese Foundation for
Science and Technology. Referência: PTDC/ ECM-EST/1494/2014 View project

OUTBURST (2016-2019): Optimal and aesthetic design of curved steel bridges. Fund for Coal and Steel
Grant Agreement No. RFCS-2015-709782 View project

All content following this page was uploaded by Luís Simões da Silva on 25 December 2017.

The user has requested enhancement of the downloaded file.

 Advanced Steel Construction Vol. 13, No. 4, pp. 343-360 (2017) 343

INFLUENCE OF EXECUTION TOLERANCES FOR

FRICTION CONNECTIONS IN CIRCULAR AND
POLYGONAL TOWERS FOR WIND CONVERTERS
Christine Heistermann 1, Marko Pavlović 1,4,*, Milan Veljković 1,4,
Daniel Pak 2,5, Markus Feldmann 2,Carlos Rebelo 3 and Luis Simões da Silva 3
1
Steel Structures, Luleå University of Technology, 97187 Luleå, Sweden
2
Institute for Steel Construction, RWTH Aachen University, Mies-van-der-Rohe-Straße 1, 52074 Aachen, Germany
3
ISISE; Department of Civil Engineering, University of Coimbra, Rua Luís Reis Santos, Pólo 2,
3030-788 Coimbra, Portugal
4
Delft University of Technology, Faculty of Civil Engineering and Geosciences, Stevinweg 1,
2600 GA Delft, The Netherlands
5
Universität Siegen, Paul-Bonatz-Str. 9-11, 57076 Siegen, Germany

*(Corresponding author: E-mail: m.pavlovic@tudelft.nl)

Received: 9 June 2016; Revised: 15 August 2016; Accepted: 29 October 2016

ABSTRACT: Friction connections with long open slotted holes have been proven to be a competitive alternative to
the conventional flange connections in steel tubular towers for wind energy converters. As full-scale tests are not
available, results of Finite Element Analysis (FEA) of the real-scale tower geometry are used in this paper to
investigate the influence of tower cross section shape, execution tolerance (gap between the shells) and length of the
connection on the bending resistance. Buckling behaviour of the shell in the vicinity of the friction connection in
circular and polygonal towers is compared. The friction connection is thoroughly examined and recommendations for
execution tolerances are given. The influence of two types of the execution tolerances on the connection strength is
considered: inward bended “fingers”, leading to inclined gaps, and a parallel gap created by different tower
diameters.

Keywords: steel tubular tower, wind energy converter, friction connection, long open slotted hole, circular cross
section, polygonal cross section, execution tolerance, FEA

DOI: 10.18057/IJASC.2017.13.4.2

1. INTRODUCTION

Steel tubular towers are the most common support for wind energy converts. They reach heights up
to 100 m, which is limited by the size of the shell diameter of up to 4.5 m. Transportation issues
impose the maximum value for shell diameter. In recently finished RFCS (Research Fund for Coal
Steel) projects, “High-strength tower in steel for wind turbines (HISTWIN)” Veljkovic et al. [1] and
“High steel tubular towers for wind turbines (HISTWIN2)” Veljkovic et al. [2], the use of technical
innovations to increase competiveness of steel tubular towers was one of the main topics. One
important innovation is the use of so-called “friction connections with long open slotted holes” for
in-situ execution, see Figure 1. This connection has been shown as a competitive alternative to the
conventional ring flange connection of the steel tubular tower for a hub-height up to 100 m. The
material cost for the connection, which takes into account shell material and bolts, is about 80 %
lower for the adequate resistance, according to Veljkovic et al. [1].

In the friction connection, the outer shell (upper segment) is equipped with fitted bolts, close
tolerance holes, on the side of the bolt head, while the inner shell (lower segment) has open slotted
holes with the width of the hole equal to the normal clearance hole diameter, see Figure 1.
Feasibility tests on tower segments, 2 m in diameter, carried out within HISTWIN and
HISTWIN2 project, Veljkovic et al. [1] and [2], proved that the bolts can be pre-installed in the
344 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

upper segment and easily slid on top of the lower segment with an execution tolerance providing a
10 mm initial gap between the shells, see Figure 2. Diameters of towers with common height of 80
– 100 m are large e.g. up to 4 m and thicknesses of shells can go up to 60 mm, depending on site
condition, wind turbine class, and steel grade used. As a rule of thumb, the initial shell thickness
which can be assumed for the pre-designed considerations is about 1% of the shell diameter.
Therefore, relatively larger execution tolerances are needed for the execution of real-scale tower.

Figure 1. The Friction Connection in Tower for Wind Converters, Veljkovic et al. [1]

a) HISTWIN Project [1] b) HISTWIN2 Project [2]

Figure 2. Feasibility Tests Performed in Previous Projects
Down-scale experiments on tower segments with friction connection have been conducted within
the scope of HISTWIN project, Veljkovic et al. [1], but neither tests or field measurements are
available with real dimensions of a tower. It is worth noting that photos shown in Figure 2 show
feasibility testing which is performed so the upper segment of the tower is placed at the ground.
This choice is made by workers in the workshop, just for their convenience, this decision has no
consequences on any conclusion relevant for the research project.

It is unrealistic to expect that experiments up to failure of such large specimens will ever be realised,
especially having in mind the possibilities of advanced FEA as demonstrated in this paper.
Thorough examination of slip and buckling behaviour of the friction connection in a tower with
circular cross section is given by Pavlović et al., [3] and [4], respectively. The time dependent loss
of the bolt preloading force in the friction connection has been considered in Heistermann et al. [5].
Polygonal tower shapes have been studied by Garzon [6] and Reinke [7], providing numerical and
 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 345

experimental evidences on the buckling behaviour of tower segments with polygonal cross section,
respectively. The friction connection in polygonal towers has been reported in the HISTWIN2
project, Veljkovic et al. [2].

The aim of the research presented here, is to comprehensively examine the behaviour of the friction
connection in circular and polygonal towers including the influence of local buckling of the shell in
the vicinity of the connection and the influence of execution tolerances. Quasi-static analysis is
made with ABAQUS FE software package, see [8], using the explicit dynamic solver coupled with
damage material models. The assumed tower segment has a diameter of the shell of D = 3374 mm,
thickness t = 24 mm, steel grade S460 and high-strength bolts M48, grade 10.9, are used for the
connection. Details of the connection are shown in Figure 3. Dimensions and values of design loads
(MEd = 45.8 MNm) at the analysed connection cross section are based on the design of a “real”
tower with the common ring flange connection. Local buckling of the shell in the vicinity of the
friction connection is analysed considering design shell imperfections in the circular and the
polygonal tower.

Figure 3. Layout of the Friction Connection in a Real-scale Tower Considered in the Study

A possible solution to ensure the execution tolerance is the provision of a gap between the outer
and the inner shell obtained by different diameters. The second alternative is to pre-bend “fingers”
of the inner shell which has the same outer diameter as the inner diameter of the outer shell. These
alternatives and the size of the execution tolerance are examined in this study.

In order to validate the FE modelling and the computational procedure, the experimental results of
the down-scale 4-point bending tests conducted by Pak and Naumes [9] on 8 mm thick shells with a
diameter of 1 m and the total span of about 7 m are used.

2. FINITE ELEMENT ANALYSIS

Two FE models of the friction connection shown in Figure 3 are created: a connection in a tower
with circular cross section (FCC) and a connection in a tower with polygonal cross section (FCP),
as shown in Figure 4. The friction connection is designed with three rows of 56 bolts each, and
cover plates to have sufficient slip resistance so that the buckling resistance of the connection
becomes the dominant failure mode.

The polygonal tower shape has 14 edges (folds). The governing criterion for such division is to be
at the boundary to local buckling of a flat part of the cross section, a segment between the folds, i.e.
keeping the slenderness of the flat part close to the lower limit for the cross section of class 4
according to EN 1993-1-1 [10], see Eq. 1 and Eq. 2.
346 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

cflat / t  42  cflat  42  0.71 24  716 mm (1)

D
nedge   14.8 . (2)
cflat

a) Circular Tower Shape (FCC)

b) Polygonal Tower Shape (FCP)

Figure 4. Different Tower Shapes Considered in the Case of Friction Connection

2.1 Geometry, Boundary Conditions and Mesh

The FEA model geometry and boundary conditions of one half of the tower segment including the
most possible refinement of the connection details shown in Figure 5 and Figure 6. Symmetry
boundary conditions for bending are used to reduce the computation time. The tower segment is
3 m long on each side of the connection. This size of the model is chosen for proper modelling of
the connection, local buckling of the shell and to allow redistribution of meridional stresses.
Preliminary FEA was used to justify dimensions and costs of the feasibility tests which has been
confirmed by the detailed analysis shown here. Cross section surfaces of the shells are fully
kinematically constrained to the reference points.

upper reference point

Ux=Uz=Rx=Ry=0
Z symmetry BC

Z symmetry BC

lower reference point

Ux=Uy=Uz=Rx=Ry=0

Figure 5. Geometry of the Model and Boundary Conditions

 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 347

Bolts and nuts are modelled with the real thread geometry, see Figure 6 and preloading of the bolts
is applied by the turn-of-nut method, see Figure 7. Tetrahedron solid elements (C3D4) are used to
form the bolt mesh. Global element size of the bolt is 11 mm, while in the threaded zone the
element size is reduced to 5 mm. Eight node hexahedron solid elements with reduced integration
(C3D8R) are used for the shell with element size of 20 mm. The shells and the cover plates are
meshed with four elements through the thickness to properly take into account the bending
stiffness.

30 mm

p = 5 mm
l = 130 mm

36.4 mm

Figure 6. Geometry and Mesh of the Bolt, Washer and Nut

General contact interaction procedure is used in ABAQUS/Explicit with “hard” formulation of

normal behaviour and “penalty” friction formulation of tangential behaviour, see [8]. The friction
coefficient of 0.14 is set for bolts and nuts threads surface pairs according to recommendation given
in ECCS Publication No. 38 [11]. For all other surface pairs in the model a nominal Class A friction
surface is assumed with the slip factor μ = 0.5 according to EN 1993-1-8 [12] in the parametric
study performed here. The sophisticated modelling of bolts is necessary to have appropriate
stiffness of the bolt and connection, however the full benefits of its complex analysis are
emphasised at the ultimate limit state of the connection, see references [2], [3], [4], and Pavlovic et
al. [13].

2.2 Loading and Computational Procedure

Two computation steps are analysed for each model: preloading of the bolts and bending to the
failure. Preloading of the bolts is applied by the turn-of-nut method. Hexagon edges of nuts are
kinematically coupled to the reference points in the centre line of each nut, as it is shown in Figure
7a. Those reference points are turned by applying changes in boundary conditions, i.e. by rotating
around the axis parallel to the shank of each bolt. The targeted preloading force of 960 kN is
calculated according to EN 1993-1-8 [12] for M48 bolts, grade 10.9. Rotations of the nuts are
calibrated in each case to achieve this targeted preloading force. Values of the bolt preloading
forces are obtained by summation of node forces in the cross section of the bolt, see. In Figure 7b,
the stress in the bolt, shells and cover plate in the friction connection are shown for the final stage
of preloading.
348 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

Table 1. Rotations of the Nuts and Resulting Preloading Forces

Case Grip Gap Nut Obtained
length (mm) rotation preloading
(mm) (rad) force (kN)
FCC 60 0 3.83 960
60 10 16.40 1024
60 20 28.95 892
60 30 41.52 893
FCP 60 0 1.00 905
60 10 13.60 964

The bending moment is applied “displacement controlled” by imposing rotations of upper and
lower reference point as shown in Figure 5.

a) The Turn-of-nut Method b) Stresses in the Bolt, Shells and Cover Plate
Figure 7. Preloading of the Bolts

Explicit dynamic solver of ABAQUS is known to be robust for this kind of analysis, where
complex contact interactions coupled with large deformations and nonlinear behaviour are present,
see [8]. It does not have usual convergence problems such as the implicit static solver. However,
the computation may be quite long, if the real dynamic response of the model is required. Loading
is supposed to be quasi-static. In this model the dynamic solver is used to efficiently solve the
quasi-static problem. Calculation time is shortened by using variable non-uniform mass scaling
technique with the target time increment set to t = 5.0 x 10-5 s. Artificial durations are adopted for
two computation steps: 7 s and 10 s, for preloading of the bolts and loading up to failure,
respectively. Trial computations are made in order to select the appropriate target time increment
and to avoid unwanted inertia forces in quasi-static FEA.

2.3 Material Models

Measured data of material properties are used for the verification FEA, while nominal material
values are used for the bolts and the shells in the real-scale FEA, as given in Table 2. Isotropic
hardening plasticity with modulus of elasticity of E0 = 210 GPa, and Poisson’s ratio of v = 0.3 is
used for all materials. Bolts, washers and nuts are set to nominal values of stress at the yield point
fy = 900 MPa and ultimate strength fu = 1000 MPa, with the ultimate elongation A = 10 %
according to ISO 898-1 [14] for bolt material grade 10.9, see Figure 8. A parabolic shape of the
nominal stress-strain curve is assumed to enable the definition of the damaged plasticity material
model of bolts.
 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 349

Table 2. Material Properties used in FEA

Material Steel grade Thickness Yield Ultimate Ultimate
strength strength elongation
t (mm) fy (MPa) fu (MPa) A (%)
Down-scale S355J2+N 8 415 534 -
verification FEA
High-strength bolts 10.9 - 900 1000 10
Shells in real-scale S460 24 460 550 -
FEA

A ductile damage model is used to model the failure of the bolts. The hardening part of the material
behaviour is defined by the plasticity curve while the softening part and the failure are governed by
the damage initiation criterion and the damage evolution law. Parameters of ductile damage
initiation criterion and damage evolution law are derived analysing undamaged and damaged
material response in a standard (round bar) tensile test, as described by Pavlović et al. [145]. The
standard (round bar) tensile test model is created and material parameters are calibrated by
comparing the nominal stress-strain curve to the FE results, as shown in Figure 8. Details of the
calibration procedure for the material models of the bolt are given by Pavlovic et al. 3].
1200

1000
Nominal stress (MPa)

800
HV 10.9 - Nominal
600 HV 10.9 - Coupon FEA

400

200

0
0 2 4 6 8 10 12
Nominal strain (%)
Figure 8. Comparison of Nominal and FEA Tensile Tests Results for Bolts

The same element type (C3D4) and corresponding size (5.0 mm) used in the validation of the
material model is used for bolts in the FE models of real-scale connections.

2.4 Verification of FEA by Down-scale Experiments

In order to prove the validity of the FEA study presented here, verification is made with respect to
the experimental data of down-scaled tests of tower segments. The layout of the 4-point bending
experiments conducted at RWTH Aachen University, see Veljkovic et al. [1] and Pak and Naumes
[9], is shown in Figure 9a. The ring flange connection and the friction connection using 32 M20
bolts and 24x3 M20 bolts, respectively, were tested with shell diameter 1 m and thickness 8 mm
over a 7 m span. The same computational procedure, which is used for the verification of FEA, is
used for the real dimensions of the tower segment in this study.
350 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

P/2

P/2

24 slotted holes:
l = 340 mm
w = 23 mm

a) test set-up b) detail of the connection c) FE model

Figure 9. Experiments and FE Model of the Down-scaled Tower Segment

Comparisons of the experimental and FE results for the friction connection are shown in Figure 10.
More details about verification of FEA are given by Pavlovic et al. ]. Very good agreement is
achieved for both failure modes buckling of the shell in case of the ring flange connection and slip
failure for the specimen where the friction connection is used. This has justified use of the same FE
model for analysis of the full scale connections.
7
2000 Pu,FEA = 1920 kN

1800 5 Compression side

1600
3
1400
FJ1
Slip (mm)

1200 1
Load (kN)

FJ2
1000
-1
800
FJ1 - Exp.
600
FJ1 - Exp.
-3
FJ2 - Exp.
400 FJ2 - Exp. FJ - FEA
FJ - FEA -5 Tension side
200

0 -7
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Deflection (mm) Deflection (mm)
a) load-deflection curves b) slip at the end of finger
Figure 10. Results of the Verification FEA vs. Experiments on the Friction Connection

2.5 Shell imperfections

Friction connections in circular and polygonal tower shapes, FCC and FCP, respectively are firstly
analysed with and without shell imperfections in the compression zone. For the circular tower
shape (FCC) a dimple imperfection is applied, as shown in
Figure 11a, with a maximum amplitude of 13 mm calculated according to EN 1993-1-6 [15]
assuming the fabrication tolerance quality class B. For the polygonal tower shape (FCP) a harmonic
shape of the local imperfection is applied, as shown in
Figure 11b. The maximum amplitude of the harmonic imperfection of 4 mm is assumed as
approximately b/200, where b is the width of the flat part of the polygonal cross section between
two folds. The position of the imperfection is obtained from non-linear analysis of the initially ideal
geometry, see Figure 15. This initial deformation is introduced in the subsequent model under a
stress-free conditions.
 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 351

a) circular cross section (FCC) b) polygonal cross section (FCP)

Figure 11. The Imperfections Applied to the Shell, Adjacent to the Connection

2.6 Execution Tolerances

A parametric study of the friction connection’s bending resistance regarding the execution
tolerances is made analysing several parameters: tower shape, length of the connection (length of
the “finger”), and type and size of execution tolerance. Two different types of execution tolerances
are considered for the friction connection in circular tower: a gap between the shells due to
different diameters of the segments and an inclined gap due to bending of fingers, as it is shown in
Figure 12. Both alternatives are investigated for the maximum gap in the range of 10 – 30 mm.
gap gap

a) parallel gap between the shells b) inclined gap due to bending of the fingers
Figure 12. Different Types of the Execution Tolerances.

Denomination of the cases considered in the parametric study is given in Table 3. The first letter
indicates the tower shape (C – circular, P – polygonal), the second is the digit designating the size
of execution tolerance (0 means no gap, 1 means 10 mm gap, 2 means 20 mm gap, 3 means 30 mm
gap). The third character denotes the steel grade of the shell (B means S460 in this study) and the
remaining three digits represent the length of the fingers (LF = 450 mm, LF = 550 mm and
LF = 650 mm). The last lower case character defines the type of the execution tolerance (g – gap
between the shells, b – inclined gap due to bending of the fingers). The overlapping length in the
connection is 50 mm longer than the fingers in all cases, as shown in Figure 3. Reference cases,
with ideal geometry (C0B450 and P0B450), have no gaps between the outer and the inner shell. No
dimple imperfections of the shell are taken into account in the parametric study to isolate the
problem of stress concentrations and buckling within the connection.
352 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

Table 3. Cases Considered in the Parametric Study

Case Tower Execution Length of the Type of
shape tolerance fingers (mm) exec. tol.
g (mm) LF (mm)
C0B450 circular 0 450 none
C1B450g 10 450 gap
C1B550g 10 550 gap
C1B650g 10 650 gap
C2B450g 20 450 gap
C3B450g 30 450 gap
C1B450b 10 450 bending
C2B450b 20 450 bending
C3B450b 30 450 bending
P0B450 polygonal 0 450 none
P1B450g 10 450 gap
P1B450b 10 450 bending

In the polygonal tower with assumed execution tolerances, slots are cut at the corners in the
connection zone, as shown in Figure 13, to allow the unrestrained deformation of the fingers during
preloading of the bolts. For the sake of simplicity no rounded root of the slot in the fold is modelled.
The width of those slots is 24 mm and the length is 50 mm longer then the fingers. No slots are
made in the initial case with perfect geometry where no gap between the shells exists, P0B450.

a) P0B450 b) P1B450g and P1B450b

Figure 13. Slots at the Corners in the Connection Zone of Polygonal Tower

3. RESULTS AND DISCUSSION

3.1 Number of Folds

According to the criteria for cross section classification given in EN 1993-1-1 [10], the number of
edges (folds) in the tower with polygonal cross section is supposed to be nedge ≤14.7 to comply with
the requirement for cross section class 4. The slenderness limit, Eq. 1, for the flat segments of the
tower assumes that the folds have enough small angle, Eq. 3. For the tower considered here the
angel between flat segments of 1540 corresponds to the nedge=14 according to eq.3. is too large to
keep folds fixed. This is obvious from results of experiments performed by Tran [16]. Therefore the
number is not supposed to be enough to to prevent its displacement at the ultimate limit state
governed by the stability.
 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 353

 2
 (1  ) for nedge  3 (3)
2 nedge
This expectations is confirmed by the analysis performed. It is clearly shown in Figure 14 that the
cross section undergoes global buckling characterized by the dominant radial movement of the
folds. Larger number of folds will lead to the weaker folds and smaller number of folds will lead to
slender segments, however the optimization of polygonal towers are left out of the scope of this
paper.  

Tension

Above conneciton

Undeformed

Below conneciton
Cross-section
above connection

Mx

Cross-section
below connection
1

1’

Compresion

Figure 14. Buckling of the Polygonal Cross Section

3.2 Influence of the Shell Imperfections

Post-failure deformed shapes for both cross section cases obtained in FEA without initial dimple
imperfections for finger length LF = 450 mm are shown in Figure 15. Failure mode in both cases is
local buckling of the shell, which includes rotation of the whole connection. This appears due to
eccentricity of the single lap joint.

a) FCC b) FCP
Figure 15. Post-buckling Deformed Shapes for Connections without Shell Imperfections
354 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

Moment-rotation curves for FCC and FCP, with and without imperfections are shown in Figure 16.
Initial stiffnesses are identical while the bending resistances are different. In both cases the ultimate
buckling resistance Mb,Ru calculated according to EN 1993-1-6 [15], assuming fabrication tolerance
class B, is surpassed.

100

EC3: Mb,Ru=82.4 MNm

Bending moment (MNm)

75

50
FCC - no imperfection
FCP - no imperfection
FCC - dimple imp. 13 mm
25 FCP - harmonic imp. 4mm

0
0 5 10 15 20 25
Rotation (rad x10-3)

Figure 16. Moment-rotation Curves of FCC and FCP for Finger Length LF = 450 mm

The bending resistance of the connection in the circular tower without imperfections is 5 % higher
compared to the polygonal tower. However, the same ultimate bending moment is obtained in the
friction connection in circular and polygonal tower: 96.8 MNm and 95.6 MNm, respectively, if
imperfections are considered. This is expected as the harmonic imperfections introduced in the
polygonal tower are smaller than the dimple imperfections introduced in the circular tower, 5 mm
vs. 13 mm, respectively. The folds in the polygonal tower stiffen the cross section.

Table 4. Bending Resistances of FCC and FCP With and Without Design Shell Imperfections
Case Imperfections Bending resistance (MNm)
Shape
Size Without With Influence of
(mm) imperfections imperfection imperfection
FCC Dimple 13 101.8 96.8 -5 %
FCP Harmonic 4 95.7 95.6 0%

Deformed shapes and meridional stresses in the compression zone of the shell at the load step prior
to buckling are shown in Figure 17 for FCC and FCP with design imperfections. Buckling shape in
FCC is different compared to the case without imperfection. If imperfections are not considered,
the shell local buckling is affected by eccentricity of the single lap joint resulting in an inclination
of the whole connection, see Figure 15a. If the design imperfections are applied, buckling occurs in
the zone of imperfection without inclination of the whole connection, see Figure 17a. In FCP shell
imperfections are rather small, and similar buckling shape including the inclination of the
connection can be seen in Figure 15b and Figure 17b. From the influence of design imperfections
on the resistance it is clear that the circular tower is more sensitive to buckling of the shell, in the
vicinity of the friction connection, compared to the polygonal towers.
 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 355

a) FCC (M = 90.7 MNm) b) FCP (M = 89.5 MNm)

Figure 17. Deformed Shapes and Meridional Stresses in the Shell Prior to Buckling

3.3 Influence of the Connection Length

Meridional stresses due to bending of the fingers by preloading of the bolts for different lengths of
the fingers are shown in Figure 18. It can be noticed that the case with shortest fingers (C1B450)
has the lowest meridional stresses since the gap is closed by the simultaneous deformation of upper
and lower shell. In the other two cases (C1B550 and C1B650) closure of the gap between the shells
is obtained by bending of the fingers, resulting in larger meridional stresses. Therefore, there is no
use of providing long fingers in order to reduce normal stresses in fingers. Certainly, a bit longer
slotted holes may be beneficial for alignment of the tower segments during the execution, but this
aspect is not taken into account here but there are no structural reasons to use longer fingers,
especially not if the inclined gap is obtained in inward bending of the fingers

a) C1B450g b) C1B550g c) C1B650g

Figure 18. Meridional Stresses in the Fingers due to Closing of the Gap by Tightening of the Bolts

3.4 Influence of the tower shape

Moment-rotation curves for the friction connection with and without the gap of 10 mm in both
circular and polygonal tower are shown in Figure 19. Reduction of the bending resistance due to
execution tolerance is 2 % and 7 %, in circular and polygonal tower, respectively, compared to the
cases without execution tolerances, see Figure 19 and
Table 5. The larger reduction in the polygonal tower is caused by the reduction of cross section area
of the shell by applying the corner slots, see Figure 13. The plastic bending moment resistance of
the cross section of the polygonal shell with corner slots Mp,Ru,red = 116 MNm is 4.2 % lower than
356 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

the plastic bending moment resistance of the gross cross section Mp,Ru = 121 MNm. This justifies
the difference in reduction of bending moment resistance obtained for the polygonal and circular
cross section.
100

Bending moment (MNm)

80

60

C0B450
40
C1B450g
P0B450
20
P1B450g

0
0 2 4 6 8 10
Rotation (rad x10-3)

Figure 19. Influence of the Corner Slots in Polygonal Tower on Reduction of Bending Resistance

Table 5. Influence of the Tower Shape

Case Tower Execution Ultimate Reduction
shape tolerance bending factor
moment
g (mm) Mult (MNm) Mult/ Mref
C0B450 circular 0 101.8 (Mref) 1.00
C1B450g 10 99.3 0.98
P0B450 polygonal 0 95.7 (Mref) 1.00
P1B450g 10 88.9 0.93

Stress concentrations at the roots of the corner slots in the polygonal tower are caused by
preloading of the bolts, see Figure 20b, which can reduce fatigue endurance of the shell. This needs
to be assessed in more detail to justify the application of polygonal tower shapes in towers for wind
converters (not scope of this study).

a) P0B450 b) P1B450g
Figure 20. Stress Concentration in the Corner Slot of the Polygonal Tower
 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 357

3.6 Influence of the Type and Size of the Execution Tolerances

Two different methods to provide execution tolerances in the circular tower:

- the gap between the shells, parallel gap established by different diameters of two tower segments
with and without fingers

- the same diameter of both tower segments but the inclined gap is established by inward plastic
bending of the fingers.,

Different sizes:10 mm, 20 mm and 30 mm of the maximum gap , are compared to the reference
(theoretical) cases of no gap, in Table 6. The reduction of bending resistance with regards to the
parameters considered above is shown in Figure 21.

Table 6. Influence of the Type and Size of Execution Tolerance on Bending Resistance
Case Type of the Size of the Ultimate Reduction
execution execution bending factor
tolerance tolerance moment
g (mm) Mult (MNm) Mult/ Mref
C0B450 none 0 101.8 (Mref) 1.000
C1B450g gap 10 99.3 0.975
C2B450g 20 96.7 0.950
C3B450g 30 94.4 0.927
C1B450b Bending 10 101.2 0.993
C2B450b (inclined) 20 100.2 0.984
C3B450b 30 100.3 0.985

The effect of execution tolerance provided by the parallel gap between the shells leads to higher
reduction of ultimate bending resistance compared to the inclined gap obtained by the inward
bending of the fingers. Almost linear dependency of the reduction factor with regards to the size of
the gap between the shells is obtained. With the largest execution tolerance of 30 mm gap between
the shells 7 % reduction of the bending resistance is obtained. Inward bending of the fingers causes
very small reduction, which is practically negligible. These results are expected as the less
eccentricity is introduced in the tower segment at the connection, if the gap is made by inward
bending of the fingers. This is illustrated in
Figure 22, where deformed shapes and meridional stresses after preloading of the bolts and after the
failure are shown for the reference case and the two types of execution tolerances.
1.00
0.99
Bending resistance reduction factor

0.98
0.97
0.96
0.95
0.94
0.93 Parallel gap
0.92
Inclined gap
0.91
0.90
0 10 20 30
Assembling tolerance (mm)

Figure 21. Influence of Execution Tolerances on Bending Resistance of the Connection

358 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

preloading after failure preloading after failure preloading after failure

a) no execution tolerance C0B450 b) gap between the shells - C3B450g c) bending of the fingers - C3B450b

Figure 22. Deformed Shapes and Meridional Stresses after Preloading of Bolts and after Failure

Meridional stresses along the inner and outer edge of the fingers in the lower shell are shown in
Figure 23 for different sizes and types of execution tolerances. Yielding criterion, fy = 460 MPa, is
reached below the fingers at z ≈ 650 mm for the 30 mm parallel gap , see Figure 23a. The inclined
gap is favourable since the meridional stresses are in elastic range, see Figure 23a. Moreover,
preloading of the bolts is simpler if the fingers are pre-bent as they can be fully straightened by
tightening of the lowest bolt row. Obviously the method of the inward finger bending for achieving
the execution tolerances is advantageous.
-500 -300 -100 100 300 500 -500 -300 -100 100 300 500
0 0
C1B450g inner C1B450b inner
outer outer
C2B450g 100 C2B450b 100
C3B450g C3B450b

200 200
Path along the finger (mm)

Path along the finger (mm)

300 300

400 400
fy = 460 MPa

fy = 460 MPa
fy = 460 MPa

fy = 460 MPa
500 500

600 600
inner
outer

700 700

800 800
Vertical (longitudinal) stress (MPa) Vertical (longitudinal) stress (MPa)
a) parallel gap b) inclined gap
Figure 23. Meridional Stresses along the Edges of Inner Shell after Preloading of the Bolts

4. CONCLUSIONS

Advanced FEA of the real-scale friction connection in a steel tubular tower for wind energy
converter have been conducted, relaying on a FEA verified by down-scaled experiments. The
friction connection is analysed in a parametric study considering circular and polygonal tower
shape, design shell imperfection, length of the fingers, and size and type of execution tolerance.
The following conclusions are drawn:
 Influence of Execution Tolerances for Friction Connections in Circular and Polygonal Towers for Wind Converters 359

1. Provision of the execution tolerance by a parallel gap between the shells higher than 20 mm
produces yielding in the root of the fingers. The bending resistance of the connection is 7 %
reduced for an execution tolerance of 30 mm gap. This method is not recommended for
larger execution tolerances. Execution tolerance provided by inward bending of the fingers
is recommended. In this case, the inclined gap, practically no reduction of bending
resistance is obtained up to a maximum tolerance of 30 mm.
2. The polygonal cross section of the tower is fully effective and no local buckling of the flat
part occurs if the number of edges is chosen such that the flat part of the cross section
satisfies the slenderness limit for the class 3 cross section according to EN 1993-1-1 [10].
3. Approximately 5 % higher bending resistance of the friction connection is obtained in the
circular tower compared to the polygonal tower, if shell imperfections are not considered.
However, design imperfections in the polygonal tower are lower compared to imperfections
of the circular tower. The same design bending resistance is obtained as for the circular
tower.
4. There is no need to use slotted holes that are significantly longer than the group of bolts in
the connection since the initial gap between the shells is closed by simultaneous
deformations of the upper and lower shell. A small extension of the slotted hole beyond the
last bolt in a row (1d0 is used in this study) should be provided to allow easy alignment of
the tower segments during execution.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the financial support by the European Research Fund of Coal &
Steel, Grant-Agreements No. RFSR-CT-2010-00031 “High steel tubular towers for wind turbines
(HISTWIN2)” as well as the Centrum for High-performance Steel (CHS) at Luleå University of
Technology, Sweden and Portuguese Regional Operational Programme CENTRO2020 within the
scope of the project CENTRO-01-0145-FEDER-000006.

REFERENCES

[1] Veljkovic, M., Heistermann, C., Husson, W., Limam, M., Feldmann, M., Naumes, J., Pak, D.,
Faber, T., Klose, M., Fruhner, K.U., Krutschinna, L., Baniotopoulos, C.C., Lavassas, I.,
Pontes, A., Ribeiro, E., Hadden, M., Sousa, R., Simões da Silva, L., Rebelo, C., Simões, R.,
Henriques, J., Matos, R., Nuutinen, J. and Kinnunen, H., “High-strength Tower in Steel for
Wind Turbines (HISTWIN)”, Final Report - RFSR-CT-2006-00031, 2012, European
Commission (RFCS).
[2] Veljkovic, M., Heistermann, C., Tran, A.T., Feldmann, M., Möller, F., Richter, C.,
Baniotopoulos, C.C., Gerasimidis, S., Zygomalas, I., Matos Silva, A., Simões da Silva L.,
Rebelo, C., Pinto, P., Matos, R., Moura, A., Gervasio, H., Siltanen, J., “High Steel Tubular
Towers for Wind RTurbines (HISTWIN2)”, Final Report - RFSR-CT-2010-00031, 2015,
European Commission (RFCS).
[3] Pavlovic, M., Heistermann, C., Veljkovic, M., Pak, D., Feldmann, M., Rebelo, C. and Simões
da Silva, L., “Connections in Towers for Wind Converters – Part 2: The Friction Connection
Behaviour”, Journal of Constructional Steel Research, 2015, Vol. 115, pp. 458-466.
[4] Pavlovic, M., Heistermann, C., Veljkovic, M., Pak, D., Feldmann, M., Rebelo, C., Simões da
Silva, L., “Friction Connection vs. Ring Flange Connection in Steel Towers for Wind
Converters”, Engineering Structures, 2015, Vol. 98, pp. 151-162.
 360 Christine Heistermann, Marko Pavlović, Milan Veljković, Daniel Pak, Markus Feldmann,Carlos Rebelo and Luis Simões da Silva

[5] Heistermann, C., Veljkovic M., Simões R., Rebelo, C. and Simões da Silva, L., “Design of
Slip Resistant Lap Joints with Long Open Slotted Holes”, J. Constr. Steel Res. 2013, Vol. 82,
pp. 223–233.
[6] Garzon, O., “Resistance of Polygonal Cross Sections - Application on Steel Towers for Wind
Turbines”, Licentiate Thesis, Luleå University of Technology, 2013.
[7] Reinke, T., “Tragverhalten von Stahlmasten mit polygonalem Querschnitt”, DASt
Kolloqium, Aachen, 2012.
[8] ABAQUS User Manual. Version 6.12. Providence, RI, USA: DS SIMULIA Corp; 2012.
[9] Pak, D. and 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.
[10] EN 1993-1-1, “Eurocode 3: Design of Steel Structures, Part 1-1: General Rules and Rules for
Buildings”, Brussels, Belgium: European Committee for Standardization (CEN), 2005.
[11] ECCS Publication No. 38, “European Recommendations for Bolted Connections in
Structural Steelwork”, Brussels, Belgium, ECCS, 1985.
[12] EN 1993-1-8: “Eurocode 3: Design of Steel Structures. Part 1-8: Design of Joints”, Brussels,
Belgium: European Committee for Standardization (CEN), 2005.
[133] Pavlovic, M., Heistermann, C., Veljkovic, M., Pak, D., Feldmann, M., Rebelo, C. and Simões
da Silva, L., “Connections in Towers for Wind Converters – Part 1: Evaluation of
Down-scaled Experiments”, Journal of Constructional Steel Research, 2015, Vol. 115, pp.
445-457.
[14] 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.
[145] Pavlović, M., Marković, Z., Veljković, M. and Buđevac, D., “Bolted Shear Connectors vs.
Headed Studs Behaviour in Push-out Tests”, J. Constr. Steel Res. 2013, Vol. 88, pp. 134-149.
[15] EN 1993-1-6: “Eurocode 3: Design of Steel Structures. Part 1-6: Strength and Stability of
Shell Structures”, European Committee for Standardization (CEN), 2007.
[16] Tran, A.T., Veljkovic, R., Rebelo, C. and Simões da Silva, L., “Resistance of Cold-formed
High Strength Steel Circular and Polygonal Sections - Part 1: Experimental Investigations”,
Journal of Constructional Steel Research, 2016, Vol. 125, pp. 227-238.

View publication stats