Page 1
8th
International Congress on Computational Mechanics
Volos, 12 July – 15 July 2015
NUMERICAL SIMULATION OF FLANGE-BOLT INTERACTION
IN WIND TUBRINE TOWER CONNECTIONS
Aikaterini I. Ntaifoti1, Konstantina Koulatsou
2 and Charis J. Gantes
3
1Institute of Steel Structures
National Technical University of Athens
Athens, GR-15780, Greece
e-mail: [email protected]
2Institute of Steel Structures
National Technical University of Athens
Athens, GR-15780, Greece
e-mail: [email protected]
3Institute of Steel Structures
National Technical University of Athens
Athens, GR-15780, Greece
e-mail: [email protected] , web page: http://users.ntua.gr/chgantes
Keywords: wind turbines, tower connections, bolted ring flanges, numerical analyses, contact nonlinearity,
ultimate load
Abstract. Wind turbines constitute the most cost-effective way for the exploitation of the available wind
potential, thus investigation of the behavior of such structures is of particular interest. Since fatigue is one of the
most common types of structural failure of wind turbines, due to the dynamic nature of wind loads acting on
their blades, consideration of connections between adjacent parts of wind turbine towers is very significant and
demanding. In the present paper, a wide range of numerical analyses are carried out, using appropriate finite
element software, in order to simulate one such typical connection. A detailed numerical model is presented,
including only a part of the L-shaped connection i.e. one bolt, as well as the connected ring flanges, using solid
elements. Contact elements are appropriately taken into account to introduce the connection’s nonlinear
behavior caused by the interaction between flanges and bolts. Objective of this work is to understand in depth
the connection’s static behavior and to determine its ultimate load under static loading. For that purpose,
parametric analyses are performed in order to evaluate the influence of different parameters on the connection’s
load transfer mechanism and on its strength.
1 INTRODUCTION
Nowadays, the need for better exploitation of the available wind resources constantly grows and wind energy
is gaining increased attention, leading to an unprecedented expansion of the utilization of wind turbines. This
triggers an effort for deeper investigation of the design, construction and operation of the mechanical and
structural parts of wind turbines. The dimensions of modern wind turbines are growing, as well as the loads
acting on them in order to take better advantage of the available wind potential. Due to the dynamic nature of
wind loads [1], which are the prevailing loads on wind turbines, fatigue of connections between adjacent tower
parts is one of the most common types of structural failure. Thus, further investigation of the structural behavior
of the tower connections, as well as of the way in which the applied actions are transferred through the tower, is
of particular interest.
The most common type of wind turbine in use today is that with free-standing steel tubular towers, with the
tower having a conical shape. Due to the large length of the tower of modern wind turbines, it is divided into
different shorter sections. At both ends of these tower sections ring flanges are pre-welded, as shown in Figure 1.
During the erection of the wind turbine these ring flanges are bolted together with closely spaced fully preloaded
bolts.
Such connections have been investigated in the past by means of approximate analytical models [2], as well
as numerical analysis and experimental tests [3, 4]. In the present research a detailed numerical model is
presented as part of the investigation of the nonlinear response of such connections up to collapse. For that
purpose a typical 3-bladed wind turbine with 1.5mW rated power is considered, having a 82.39m long conical
tubular tower. The examined connection is located at the upper 85% of the tower height and in the presented
Page 2
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
numerical model only one of the bolts and its effective part of the flange are simulated, resulting in an L-shaped
connection. The connection is simulated using 3D solid elements for the ring flange, the tower shell and the bolt,
as well as appropriate contact elements to account for the interaction between flanges and bolts and between nuts
and flanges.
Figure 1. Wind turbine tower segment ring flange
Purpose of the present paper is to determine the ultimate load of this part of the connection under static
loading and to investigate how it is affected by different parameters. Both material and geometric nonlinearity is
taken into consideration in the analyses. Geometrical nonlinearity exists due to the fact that the imposed load on
the connection is resisted through the redistribution of the contact forces over the contact surfaces. In a parallel
investigation, the entire ring flange connection has been simulated and investigated via a less refined numerical
model [5]. The two models, the present one focusing on local bolt behavior, and the other one examining global
connection behavior, will be employed for fatigue investigations in future research work.
2 ULTIMATE STRENGTH OF THE L-SHAPED CONNECTION
2.1 Description of numerical model
The dimensions of the aforementioned simulated connection are presented in Figure 2. The inner and outer
diameters of the wind turbine tower, at the position of the connection, are equal to 3.04m and 3.3m, respectively.
The ring flanges are bolted together with fully preloaded bolts M36-10.9 and their thickness is 60mm. The
tower’s shell thickness adjacent to this connection is 11mm.
Figure 2. Vertical section of the connection with dimensions in mm
Due to symmetry, the numerical model includes only the upper flange and the half bolt, in order to reduce the
model’s size and computational cost. The top surface of the lower flange is also simulated as fully fixed, in order
to employ the appropriate contact boundary conditions to the bottom surface of the upper ring flange. Tolerance
gaps between the two ring flanges, as well as the upper flange and the bolt head are taken equal to 0.2mm. The
ring flanges’ holes are considered 3mm larger than the bolt’s diameter, according to pertinent EC3
recommendations for normal bolts [6]. The bolt head is designed according to DIN6914 [7] and the bolt’s
pretension force is taken equal to the maximum allowable value (70% of the bolt tensile strength) [6]. Successive
Page 3
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
mesh refinement is employed, using FEMAP software [8], to achieve a proper balance between accuracy of the
results and computational cost. The nonlinear analyses are performed using finite element program ADINA [9].
Wind loads acting on the rotating blades of the wind turbine are transferred to the tower as well as to the
connection as a bending moment with direction perpendicular to the wind, which constitutes the prevailing
action, as well as an accompanying shear force. Moreover, an axial force due to weight of electromechanical
equipment and weight of the upper part of the tower is also present. In Figure 3 the ADINA numerical model is
presented, where the combined effect of bending moment and axial force is transferred as a distributed upwards
vertical load acting on the tower shell, while the shear force is neglected. For both the ring flanges and the tower,
as well as for the bolt, a bilinear elastoplastic material with hardening up to rupture is applied, assuming steel
qualities for the ring flanges and the pylon and 10.9 for the bolt.
Figure 3. Numerical model created in ADINA
2.2 Results of numerical analyses
The results extracted from the aforementioned numerical analyses are given in the present section. Namely,
in Figure 4 the equilibrium path of the L-shaped connection is shown for a maximum applied load at the tower
shell equal to 22kN per node. In the horizontal axis the y-displacement of point M (shown in Figure 3) at the top
of the modeled part of the tower shell is illustrated. It is observed that the connection’s equilibrium path is
curved even for small values of the imposed load with a gradual stiffness decrease. This is due to the nonlinear
nature of the contact.
Figure 4. Connection’s equilibrium path
In Figure 5 the stress distribution on the upper flange and the bolt at the two characteristic points 1 and 2 of
the equilibrium path, illustrated in Figure 4, is shown. In general terms, stress concentration focuses around the
bolt’s hole and the bolt’s shaft, while the major part of the ring flange remains elastic. It is noted that the bolt
shaft enters plasticity first at its bottom, due to combined tension and bending. Material yielding occurs also at
M
Page 4
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
the bottom edge of the upper flange, where the connection’s two ring flanges are in contact. For the maximum
value of imposed load material failure of the bolt’s shaft as well as of the ring flange is at advanced state.
Point 1
Point 2
Figure 5. Von Mises contours of the upper flange and the bolt at the two characteristic points
In order to better appreciate the qualitative features of the connection’s behavior the axial force of the bolt, as
well as the vertical contact forces between the two ring flanges and the maximum equivalent stress at the
welding between flange and tower shell are also presented as functions of the imposed load. The graph of the
bolt’s axial tensile force is illustrated in Figure 6 and the resultant of contact stresses along z axis developed at
the interface between the two ring flanges in Figure 7. At the beginning of load application the tolerance gaps are
totally closed and gradually the two ring flanges are moving apart. From these figures the redistribution of
contact forces occurring during load transfer through the connection is evident, following three successive
phases. Firstly, the imposed force is received through the redistribution of the vertical stresses on the ring flange,
while the bolt force remains constant, equal to the initial pretension force. Then, the bolt force increases and
stress distribution continues. In the last phase the redistribution ends and the remaining load is received
exclusively from the bolt, with its axial force continuously increasing.
Figure 6. Axial tensile bolt force as function of imposed load
Page 5
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
Figure 7. Contact force z at the interface between the two ring flanges
Ιn Figure 8 images of the distribution of the contact reaction at the lower surface of the upper flange are
illustrated for different values of the imposed force. Blue color denotes loss of contact, while red and purple
show areas of high reactions. Initially, before load application, the distribution is extended through the whole
surface, and as the imposed load increases it is gradually transferred towards the edge. The maximum value of
stress developing at the position of welding between ring flange and tower shell remains low and in any case
below the material’s yield limit, even for the connection’s ultimate load. Moreover, the pertinent diagram is
constantly linear, as shown in Figure 9.
Figure 8. Illustration of the redistribution of the contact reaction at the interface of the two ring flanges for
different values of the imposed force
Figure 9. Maximum equivalent stress at the position of welding
Page 6
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
3 PARAMETRIC INVESTIGATION OF THE L-SHAPED CONNECTION
In the present section a wide range of parametric analyses are presented in order to evaluate the influence of
different parameters on the connection’s stiffness and ultimate strength, as well as on the way the load is
transferred through the connection. The parameters investigated are the bolt’s pretension force and quality, as
well as the thickness of the ring flanges and the bolt’s diameter. These parametric analyses are carried out
retaining all other factors of the numerical model same as described in section 2 and modifying only the specific
parameter examined.
Regarding the bolt’s pretension force, the initial value, assumed as equal to 70% of the bolt’s tensile strength,
is reduced to values between 60% and 10%. The bolt quality is decreased from 10.9 in the initial numerical
model to 8.8, while lower bolt quality is not considered, taking into account that these are preloaded bolts.
Concerning the thickness of the connection’s ring flanges, it is reduced from 60mm to 40mm and the bolt’s
diameter is changed from M36 to M20, M24, M27 and M30.
3.1 Influence of different parameters on the connection’s ultimate strength
In Figures 10 to 13 the equilibrium paths of the examined numerical models for the four investigated
parameters are presented. It is observed that the connection’s ultimate strength is affected by all these
parameters. The bolt’s pretension force and the thickness of the ring flanges affect not only the ultimate load of
the connection, but also its stiffness. As shown in Figures 10 and 12, the initial slope of the equilibrium path is
reduced for smaller values of pretension and thinner ring flanges. On the other hand, the quality and the diameter
of the bolt do not affect the connection’s stiffness, according to Figures 11 and 13.
Figure 10. Connection’s equilibrium paths for different values of the bolt’s pretension force
Figure 11. Connection’s equilibrium path for different bolt qualities
Page 7
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
Figure 12. Connection’s equilibrium paths for different values of the flange thickness
Figure 13. Connection’s equilibrium paths for different bolt diameters
3.2 Influence of different parameters on load transfer mechanism
Next, the influence of the load transfer mechanism by the aforementioned investigated parameters is
presented. Regarding the bolt pretention force, it is noted that the load resisted by the bolt gradually increases for
smaller values of pretension, thus the stress range acting on it becomes significant (Figure 14). As a result, the
connection’s resistance against fatigue is expected to be significantly reduced.
Figure 14. Equilibrium paths of the bolt’s axial force and of the vertical contact forces for different pretension
forces
Ιn Figure 15 the respective results regarding the bolt quality are presented. It is observed that this parameter
does not affect load transfer, but only the value of the imposed force at which material failure occurs. The part of
the load for which the bolt force remains constant is almost the same for both bolt qualities.
Page 8
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
Figure 15. Equilibrium paths of the bolt’s axial force and of the vertical contact forces for different bolt qualities
Similarly, it is noted that the influence of the thickness of the ring flanges on the connection’s behavior is
almost negligible. As illustrated in Figure 16, the redistribution taking place is only slightly different for each
value of thickness.
Figure 16. Equilibrium paths of the bolt’s axial force and of the vertical contact forces for different values of the
flange’s thickness
Finally, it is observed that the bolt diameter has similar effect on the load transfer mechanism as the
pretension force. More specifically, for smaller diameters the stress redistribution of the contact surfaces taking
place is constantly reduced.
Figure 17. Equilibrium paths of the bolt’s axial force and of the vertical contact forces for different bolt
diameters
4 CONCLUSIONS
The response of a typical connection between adjacent parts of shell-type wind turbine towers has been
investigated by means of a detailed numerical model and nonlinear analyses up to collapse. Specifically, a
connection of an 80m tall tubular conical steel tower of a wind turbine with rated power 1.5mW is considered.
Such connections consist of ring flanges that are pre-welded on the shell parts and are the bolted together with
preloaded bolts. Only a typical L-shaped part of such connection has been simulated using 3D solid elements for
the ring flange, the tower shell and the bolt, as well as appropriate contact elements to account for the interaction
between flanges and bolts and between nuts and flanges, in order to evaluate the local behavior. In
accompanying work, a simpler model of the entire connection is analyzed. Numerical results lead to useful
Page 9
Aikaterini I. Ntaifoti, Konstantina Koulatsou and Charis J. Gantes
conclusions about the connection’s ultimate strength and its stiffness. Due to flange-bolt interaction, the
connection’s equilibrium path is softening already at low load levels, in spite of the elastic stress distribution.
Furthermore, parametric analyses offered a wide range of information about the influence of different parameters
either on the ultimate load, or other significant characteristics of the connection. The ultimate strength of the
examined model was decreased due to reduction of the connection’s dimensions and pretension. Parameters,
such as the thickness of the ring flanges and the bolt’s pretension force influenced significantly not only the
ultimate load but the initial stiffness, as well. Generally, when it comes to the way the load is resisted, all
parameters except for the bolt’s quality, influence the stress range acting on the bolt and the redistribution of
stresses taking place between the two ring flanges. As a result, fatigue strength of these connections is expected
to decrease abruptly when dimensions of its components are reduced. Detailed fatigue investigation will take
place in the next phase of this research.
ACKNOWLEDGEMENT
This research has been co-financed by the European Union (European Social Fund - ESF) and Hellenic
national funds through the Operational Program "Competitiveness and Entrepreneurship" of the National
Strategic Reference Framework (NSRF 2007-2013) - Research Funding Program: Bilateral R&D Cooperation
between Greece and China 2012-2014, under project SEAWIND with code 12CHN184.
REFERENCES
[1] Schmidt, H., Jakubowski, A. (2001), Ermüdungssicherheit imperfekter vorgespannter Ringflanschstösse in
windbeanspruchten turmartigen Stahlbauten, Universität Essen.
[2] Petersen, C. (1993), “ Stahlbau: Grundlagen der Berechnung und baulichen Ausbildung von Stahlbauten”,
Vieweg.
[3] Wang, Y.Q., Zong, L., Shi, Y.J. (2013), “Bending behavior and design model of bolted flange-plate
connection”, Journal of Constructional Steel Research, Vol. 84, pp. 1-16.
[4] Hoang, V.-L., Jaspart, J.-P., Demonceau, J.-F. (2013), “Behaviour of bolted flange joints in tubular
structures under monotonic, repeated and fatigue loadings I: Experimental tests”, Journal of Constructional
Steel Research, Vol. 85, pp. 1-11.
[5] Ntaifoti, A.I. (2014), Investigation of the behavior of wind turbine tower connections by numerical
simulations, Diploma Thesis, Department of Civil Engineering, National Technical University of Athens
(in Greek).
[6] EN 1993‐1‐8: 2005, Eurocode 3: Design of steel structures – Part 1‐8: Design of joints, European
Committee for Standardization, 2005.
[7] DIN 6914, High Strength hexagon head bolts, Deutsche Norm, 1989.
[8] FEMAP 101, Introduction to Femap with NX Nastran Student Guide, Version 10.1.1, Siemens PLM
Software, 2010.
[9] ADINA System 8.3, Release Notes. ADINA R & D Inc., 71 Elton Avenue, Watertown, MA 02472; USA.
2005.