DOI: https://doi.org/10.14256/JCE.2125.

2017 Građevinar 9/2018

Primljen / Received: 16.9.2017. Calculation procedure for determining wind action from vortex-induced
Ispravljen / Corrected: 4.4.2018.
Prihvaćen / Accepted: 11.4.2018.
vibration with verification of fatigue strength of steel structures
Dostupno online / Available online: 10.9.2018.

Authors:
Subject review
Marina Rakočević, Svetislav Popović
Calculation procedure for determining wind action from vortex-induced
vibration with verification of fatigue strength of steel structures

An applicable calculation procedure, implemented with an object-oriented software that

calculates amplitudes of oscillation due to vortex shedding, is presented in this paper
Prof. Marina Rakočević, PhD. CE using a stadium lighting pole as an example. The fatigue strength of material is proven,
University of Montenegro for the defined S-N curves, according to the nominal stress method. Available Eurocode-
Faculty of Civil Engineering based calculation methods are also analysed and their advantages and disadvantages
[email protected] are presented.

Key words:
vortex shedding, fatigue of material, lock-in, spectral and resonant method, cantilever

Pregledni rad
Marina Rakočević, Svetislav Popović
Proračunski postupak za određivanje djelovanja vjetra pri vrtložnom
osciliranju uz provjeru otpornosti na zamor čeličnih konstrukcija
Svetislav Popović, MSc. CE
Europoles GmbH & Co. KG U radu se na primjeru konzolnog nosača za osvjetljenje stadiona prikazuje primjenljiv
Neumarkt in der Oberpfalz, Deutschland proračunski postupak za koji je napisan objektno-orijentirani program koji izračunava
[email protected], amplitude osciliranja pri vrtložnom osciliranju. Za definirane S-N krivulje, dokaz
[email protected] pouzdanosti materijala na zamor se provodi prema metodi nominalnih naprezanja.
Analiziraju se i dostupne metode proračuna iz Eurokoda i upućuje se na njihove
prednosti i nedostatke.

Ključne riječi:
odvajanje vrtloga, zamor materijala, lock-in, spektralna i rezonantna metoda, konzola

Übersichtsarbeit
Marina Rakočević, Svetislav Popović
Berechnungsverfahren zur Bestimmung der Windeinwirkung bei
wirbelerregten Querschwingungen mit Überprüfung der Betriebsfestigkeit
von Stahlkonstruktionen

In der Abhandlung wird am Beispiel eines Kragarmträgers für die Stadionbeleuchtung

das anwendbare Berechnungsverfahren dargestellt, für das ein objektorientiertes
Programm geschrieben wurde, das die Amplitude der Schwingungen bei wirbelerregten
Querschwingungen berechnet. Für die definierten S-N-Kurven wird der Nachweise der
Betriebsfestigkeit des Stahls gemäß der Methode der Nennspannungen durchgeführt.
Analysiert werden auch die zugänglichen Berechnungsmethoden aus dem Eurocode
und es wird auf deren Vorteile und Nachteile hingewiesen.

Schlüsselwörter:
Wirbelerregung, Materialermüdung, Lock-in, Spektral- und Resonanzmethode, Kragarm

GRAĐEVINAR 70 (2018) 9, 793-809 793

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

1. Introduction conclusions. However, reliability must strictly be proven for

structures of special significances and this by conducting detailed
The influence of the phenomenon of oscillation in transverse theoretical, numerical, and experimental studies.
direction, in the case of free-standing cantilever structures The best know models of vortex shedding applied in standards
subjected to vortex shedding, is a highly significant and topical can be found in: Eurocode [1] Method 1 – harmonic/vortex-
issue for practicing engineers and scientific community. Slender resonance method and Method 2 – spectral method, CICIND
structural girders such as chimneys, vertical lighting poles, standards for steel chimneys – spectral method [2], Canadian
overhead transmission lines, towers, and pylons, are particularly standard NBC [3], Danish standard [4], and Australian/N.
sensitive to this phenomenon, and must therefore be analysed Zealand standard [5]. An unofficial "Hybrid Method" – ESDU
with special care. (Item Nos. 96030 and 96031) [6, 7] is also often used in practice.
The starting assumption that the researchers of this Practical results obtained by this method have proven to be in
phenomenon have agreed on is that the resonance problem in good harmony with measured amplitudes of steel and concrete
the lock-in state causes amplitudes which, although in most chimneys, cf. [8].
cases incapable of rapidly endangering bearing capacity of the Regarding the last mentioned method, research has revealed
structure on the global or local level, do constitute a hazard that it estimates in a satisfactory way the oscillation amplitudes
due to material fatigue effect, which is a direct consequence of in the case of three out of four tested structures. That is why
the cyclic nature of the oscillation of structures in the resonant such calculation approach can be proposed as an alternative in
state. A usual graphical representation of von Karman’s vortices case of large-size structures with high critical velocities when,
for the mentioned oscillations is shown in Figure 1. Although at a low value of Scruton number, the vortex-resonance method
there are many literature data on the limits of critical velocities can underestimate, and the spectral method can overestimate,
causing resonant problems, an all-embracing model that the oscillation amplitude due to vortex shedding.
could estimate with high level of certainty the effects of this As already mentioned, two methods are used in Eurocode. The
phenomenon on the structures, at arbitrary flow regimens advantages and disadvantages of both methods are presented
described by Reynolds number, has not as yet been proposed. in available literature [8-21]. This literature enables us to
That is why even today, decades after basic principles and values derive conclusions about various assumptions and adequate
describing this phenomenon have been agreed upon, the theory consequences that may be expected if these calculation models
is still being analysed and harmonized with the amplitudes and are accepted as relevant. Theoretical bases for both calculation
damage patterns observed in practical situations. methods, with appropriate observations and suggestions, are
presented in this paper.
The scope of this paper is limited to vortex shedding i.e.
oscillation of the structure in transverse direction. It should be
noted that oscillation parallel to the wind can also contribute
to the total damage and can thus also be a subject of future
research. In some specific structures, such as new generation of
compact overhead transmission lines, the conductor galloping
can significantly contribute to the total accumulation of damage,
and so it must be taken into account in an appropriate way.

2. Theoretical bases
Figure 1. Von Karman’s vortices
2.1. Spectral method
The experience has shown that in most cases the analysis of the
above mentioned structures is directly dictated by proofs of fatigue The spectral method, better known as Hansen method for
of materials at individual critical details in which the concentration calculation of vortex shedding, presented in EN 1991-1-4 [1],
of stress occurs. This assertion gains a special importance in derives its roots from papers published by Vickery and Clark [22],
application of a simplified analysis that is based on the use of the which were additionally extended by Vickery and Basu [23-25].
nominal stress method. Thus it can often occur that, although the According to Vickery-Basu Model, a general expression of the
structure is not endangered by static action of external forces, a ratio of amplitude oscillation ymax on top of a uniform cantilever
completely contrary conclusion can be made for the relevant beam to critical section width b, in the first mode of oscillation,
design criterion – fatigue of material due to vortex shedding. is written as follows:
The proof of fatigue by means of more complicated local
approaches that take into account the hot-spot or notch stress is
an alternative that, in most cases from the everyday engineering (1)
practice, does not justify the time and resources invested,
although such methods do lead to somewhat more favourable where:

794 GRAĐEVINAR 70 (2018) 9, 793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

g -maximum amplitude factor depending on resonance inadequate by the authors of this paper. In addition, assumptions
frequency (the value of this parameter varies between 3, from calculation models lead to the conclusion that this method
5 and 4) has been excluded from the German national annex [26] for
n1 - oscillation frequency in first mode the following reasons: inconsistency of predicted results, use
Sci(n) - spectral density of the coefficient of generalized is limited to the basic mode of oscillation, use is limited to
transverse force structures with regular distribution of dimensions along the
r - air density at vortex shedding main axis of the structure (which is in civil engineering practice
me - effective mass per unit length, given in F.4 [1] typical for chimneys only), it cannot be used for elements in
d - critical damping ratio containing structure damping and group or linear disposition, it cannot be applied for structures
aerodynamic damping the response of which must be calculated using modal analysis
St - Strouhal number, E.1.6.2 [1] and taking into account geometric nonlinearity of the system
f(f) - oscillation mode function. (i.e. for overhead transmission lines during their service life), and
the model is highly dependent on local effects that influence the
It is presented in Eurocode [1] in the closed form for the first aerodynamic damping parameter. The authors consider that the
mode of oscillation as a spectral method for calculation of spectral method – in its original form – without an appropriate
vortex shedding: extension of parameters of aerodynamic damping as a function
of intensity of turbulence, as defined for instance in the French
national annex [27], is in fact conservative and uneconomical.
; ; This type of spectral method would prevent construction of
(2) structures that are normally built in Central Europe, based
on the current state-of-the-art. This assertion has also been
; confirmed by the author of the method [12].
As to laminar winds in case of temperature inversion, the authors’
experience - based on construction of stadium lighting facilities in
where, in addition to the symbols already defined in Eq. (1): Norway - actually gives advantage to the spectral method, but not
kp - peak factor in the sense of material fatigue, but rather in the sense of taking into
sy - standard deviation of displacement account large amplitudes in static sense, as they can be comparable
Sc - Scruton number to deflections of the structure caused by wind with the return period
Cc - aerodynamic constant, dependent on the cross-sectional of 50 years. The spectral method will not be used in the calculation
shape, and for the circular cylinder also dependent on the procedure given in the example. A more detailed analysis of this
Reynolds number Re, as defined in E.1.3.4, [1]; given in method, and the proposal for a practically applicable calculation
Table E.6, [1] procedure that would avoid at least some of the mentioned
Ka - aerodynamic damping parameter, given in E.1.5.3 (4), [1] deficiencies, will be the subject of some future research.
aL - normalised limiting amplitude giving the deflection of
structures with very low damping, given in Table E.6 [1] 2.2. Vortex - resonance method
h, b - the height and width of structure; For structures with
varying width, the width at the point with largest The vortex-resonance method is based on the papers presented
displacements is used. by Scruton et al. (summarized in [28]), Rumman [29], and
Ruscheweyh [13-18, 30]. It is based on the assumption that
By comparing measured amplitudes with the results of this the vortex shedding phenomenon creates sinusoidal forces of
model, Clobes, Willecke and Peil [10] observed that the model harmonic nature that are perpendicular to wind direction. Unlike
provides very rare and extreme values of oscillation amplitudes spectral method, it is deterministic in nature and is based on a
for low values of Scruton number, and that it is not suitable for clearly defined mechanics.
application in the fatigue analysis of material in case of structures The starting equation of the Ruscheweyh model is based on the
where a considerable safety is expected during service life modal force Q(t) which is, for the normalised oscillation mode
(mostly 50 years). It is not suitable to combine the influence due fi,y(z), described with:
to resonant oscillation according to this model with the number
of loading cycles estimated according to expression (E.10) (3)
from [1], and this regardless of the bandwidth value adopted.
The calibration of the total number of oscillations for use in Symbols given in expression (3) are:
combination with rare amplitudes calculated according to the z - height coordinate of the girder of total height h;
spectral method has not as yet been presented in literature, and F(z,t) - cross-sectional wind load due to vortex shedding in time
it greatly restricts the use of this method. Arbitrary setting this t, by unit length, where the following is valid
value to N > 200 · T leads to minimum number of oscillations
amounting to 104, which is considered oversimplified and (4)

GRAĐEVINAR 70 (2018) 9, 793-809 795

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

where: A greater change in wind speed that disturbs the attained

q(z) - dynamic velocity pressure "synchronous state" is not included in this empirical parameter,
b(z) -  dimension (diameter or width) of cross section as a and must be separately taken into account during estimation
function of height coordinate of the number of oscillations a structure is subjected to in the
cF(z) - non-dimensional shape factor describing load amplitude design-defined time period of construction and service life.
nshe - frequency of structure oscillation caused by vortices As it is already indicated that this is a resonance problem, the
g(z) - sign factor pointing to force orientation in the direction of maximum response of the structure of height h is obtained in
the sign of mode shape. the case of "overlapping" or "matching" of the natural frequency
ni,y of the ith oscillation mode with the frequency due to vortex
It is assumed that these forces act at a particular length that action nshe, where the amplitude yF,max is realized for reference
is defined as "correlation length" L. The term itself comes from values of cross section b and the corresponding dynamic wind
the field of stochastics, and its objective is to take aeroelastic pressure:
effects into account. The space correlation coefficient R12 for
fluctuating forces in two critical points along cross section is
defined as:
(7)

(5)

where are forces fluctuating along the unit length of a where the following symbols are used besides the already
cylindrical body. defined ones:
It is assumed that the mean quadratic fluctuating force is fi,y(z) - maximum normalised amplitude of oscillation
constant along the unit length, and so . qref,b - dynamic wind pressure at critical height with diameter b,
.
For the cylinder cross section b, in case when the distance where the vortex separation occurs.
between two critical points is reduced, where
approaches 1. In the mentioned expression, is the distance Significance of Scruton number Sc and Strouhal number St
between two measuring points as a function of diameter b. was already recognized in initial models that described this
When the defined critical points are spreading apart, phenomenon. The Scruton number describes the sensitivity of
approaches zero. The expression (6): oscillations as a function of structural damping and the ratio of
structural mass to fluid mass (air), by means of the following
expression:
(6)
(8)

where , defines the spatial limit of statistical events, where dsis the structural damping expressed by the logarithmic
such as small fluctuations of critical velocity, brief turbulent decrement, and mi,e is the equivalent mass per unit length for
wind action, or synchronous vortex separation. In this way, the mode i, according to [1].
action of vortex forces (3) is limited to "correlation lengths" L The damping of this model consists only of structural damping
that represent an empirical value based on measurements, and that is expressed in form of logarithmic decrement while,
where is mostly adopted in the form of an exponential unlike spectral method, the aerodynamic damping is neglected.
function. A graphical presentation of spatial correlation of a The reasons for this are manifold, and it is already indicated
cylindrical body is shown in Figure 2. in theoretical bases of spectral method that the intensity
of turbulence greatly influences the value of aerodynamic
damping. This parameter is difficult to determine, which is why
its estimation, in case of absence of appropriate guidance from
standards, constitutes a highly challenging task for everyday
engineering practice. The fundamental problem of detailed
calculation in higher oscillation modes in the modal analysis
of, for instance, an overhead transmission line at its operating
stage, is related to the fact that the logarithmic decrement of
damping defined in [1], Table F.2, provides guidelines for the
fundamental oscillation mode only. In addition, the proposed
values are not comprehensive, i.e. they provide guidelines
Figure 2. Spatial correlation of lateral forces for cylindrical body for a particular type of structure only, so that in many cases

796 GRAĐEVINAR 70 (2018) 9, 793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

some additional assumptions have to be made. The analysis of The meaning of expression (10) is not obvious and requires
measured amplitudes presented in [10] shows an exponential introduction of a new parameter called "effective correlation
growth of oscillation amplitude for the values of Sc < 5. If the length" Lj, and of the standard deviation of load in form of the
geometry of the structure is selected in such an unfavourable value clat (round-mean-square value) that is defined in Eurocode
way, it is recommended to have special consultations with as the "lateral force coefficient". In this way, the newly defined
experts from the field of dynamic influences at wind load, so as replacement effective correlation length contributes to the model
to enable more appropriate interpretation of results. Strouhal by increasing the action of cross acting forces (4) by the factor of
number St describes the relationship between the critical wind amplitude g which, in practice, varies from √2, that corresponds to
velocity Vcrit,i and the natural frequency of the structure in the harmonic excitation of the system described by vortex resonance
i-th oscillation mode ni,y, with the following expression: model, and max. ≈ 4 that characterises the spectral model.
And, finally, by replacing equation (10) in (7) we obtain a
(9) recognisable form of equation for the amplitude of oscillation
that is present in EN 1991-1-4, (E.7) [1]:
This expression provides an answer to the dependency of the critical
wind velocity to the scalar value St and frequency characteristics (11)
of the structure. Many detailed data have been published about
the St value in relevant literature. The proposal for adopting this
value is presented in Figure 3. In most cases, this value is defined In the equation (11), K stands for "mode shape factor". It is indicated
for circular cylinders as amounting to St ≈ 0.20. Experience shows in literature [30, 31] that the said parameter does not have a
that this conclusion is very often inadequate as the nature of vortex "physical significance". On the other hand, the effective correlation
shedding is dependent on the flow regimen, which is defined by length factor Kw includes aeroelastic nature of forces through the
the Reynolds number. Thus the simplified value of 0.20 is mostly phenomena described in equations (5), (6) and (10) in a simplified
adopted for the subcritical area of flow regimen (Re < 2 x 105), while way, so that it is proportional to the set of dynamic pressure
the St value mostly varies from 0.20 to 0.30 in the supercritical coefficients, when orthogonal solutions of the eigenvalue-
area (Re > 5 x 106). After introduction of EN 1991-1-4 this value problem of a cantilever beam are defined in the form of analogous
was set to 0.18 for all values of the Reynolds number. Individual functions. This assertion was used by Dickel who included all these
measurements have shown that this value may even amount to observations in a developed calculation procedure [31, 34-36].
0.16 for a low value of Reynolds number [10, 32]. That is why it
is recommended in the calculation method to adopt the Strouhal
number of 0.18 for all values of the Reynolds number, as defined
in [1], while for structures of special significance it is recommended
to conduct additional experimental and numerical investigations in
order to make a more realistic estimation of this value.
In the sense of the so called maximum amplitude factor g that is
Lj / b yF,max / b
also present in the spectral model (1), it can be stated that this
6 ≤ 0.10
factor is included in the Ruscheweyh resonance model through
4,8 + 12 · yF,max / b 0.10 - 0.60
the following expressions:
12 ≥ 0.60

(10)
Figure 4. Lj/b as a function of dimensionless amplitude yF,max/b,
according to [1, 13, 37]

Re St

105 ≤ Re ≤ 4 · 105 0.19

4 · 105 ≤ Re ≤ 106 -0.7674 + 0.1709 log10(Re)

106 ≤ Re ≤ 1.6 · 106 1.3752 - 0.1862 log10(Re)

Re > 1.6 · 106 0.22

Figure 3. Strouhal number as a function of Reynolds number – left according to [8]; and right according to [33]

GRAĐEVINAR 70 (2018) 9, 793-809 797

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

As the relationship between the dimensionless amplitude This deficiency can very easily be avoided using the procedure
yF,max/b and the assumed effective correlation length Lj, is the presented in literature [31] that is based on the finite element
final criterion on convergence of the solution (Figure 4 according method. For the bar finite element presented in Figure 5, starting
to [1, 13, 37]), it is necessary to use in the design an iterative from the normalised form of oscillation fr which satisfies the
procedure which, in case of high oscillation amplitudes, will not known form of eigenvalue problem:
underestimate contribution of aeroelastic forces. The following
final condition is proposed as the convergence criterion: (15)

(12) the following solution Xr can be written for the oscillation mode
r with circular frequency wr:
The expression (12) emphasizes that the iterative procedure
is conducted in all points of the girder (s) and, at that, the (16)
necessary number of iterative steps is marked with k. The
procedure proceeds until fulfilment of the condition defined by The assumption of the defined force vector, as a function of the
expression (12), i.e. until attainment of the difference of less stiffness matrix K, mass matrix M and bar displacement vector
than 1‰ between two iterative steps. n, is adopted as follows:
By calculating the amplitude, the oscillation gains its quantitative
meaning. Thus the amplitude of oscillation yF,max,i,s of an arbitrary (17)
point s in the oscillation mode i can be expressed as follows:

yF,max,i,s = yF,max · fi,y(s) (13) where: is the force vector, and

is the displacement vector, while
stiffness matrix and mass matrix are derived for interpolation
where fi,y(s) is the normalised value of the oscillation mode for functions in form of Hermite polynomials for the bar exhibiting
the oscillation shape i in the point s. the length l, stiffness EI and weight per m’ m:
As the final objective of the nominal stress calculation is to
verify material fatigue, it is necessary to determine internal
forces that occur along the girder length due to vortex shedding. (18)
The following expression is presented in Eurocode [1]:

(14) By applying expression (16), it can be concluded that

is valid which, once (17) is applied, leads toward local nodal forces
where m(s) is the mass of the structure per unit length, while Fw(s) of the element Qr in the r-th form of oscillation mode, cf. [31]:
are inertia forces per unit length of the girder. It can be noticed
that this expression is only valid in case of a cylindrical structure (19)
where there are no discontinuities in mass and stiffness.
For tapered elements, i.e. for element with an inclined sides, it is
possible to obtain a mean value of the moment of area at ends,
with discretisation to small elements, which has in fact been
applied in the calculation procedure defined in this paper, in
such a way that , while the replacement
value for ml in the mass matrix is adopted as the difference of
normal force at bar ends, according to the theory of first order,
taking into account contribution of concentrated masses.

3. Calculation procedure

3.1. Calculation of actions due to vortex shedding

All theoretical observations presented in Section 2.2 have been

implemented in the authors’ own object oriented software
package that offers solution to quantitative oscillation due to
Figure 5. Generalized forces and displacements of the transversely vortex shedding, as well as appropriate internal forces that
loaded bar, according to [31] can be used to calculate nominal stresses. The software can

798 GRAĐEVINAR 70 (2018) 9, 793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

also be used for calculation of hybrid structures created by into relative damping z. The fixed value z amounting to 0.05
combining several materials such as concrete, steel, or glass is defined for the oscillation mode m that is associated with
fibre reinforced polymers (GFRP). the frequency at which a considerable percentage of mass
The authors’ own program for determining influences due participation, greater than 90 %, is realised. Therefore we have
to vortex shedding was written in the object-oriented code (23):
because of complexity of the calculation procedure. Specialized
classes were derived using the inheritance method, which is (23)
particularly favourable in repeated use, namely in the iterative
calculation procedure for calculating amplitude as a function of where wd is the circular frequency of the damped system, and
an assumed effective correlation length. w is the circular frequency of the undamped system. However,
The evolution of expression (19) can be made using modal due to small damping value, expression (20) can be considered
analysis, which was implemented in the calculation procedure in the same form for both damped and undamped system.
via the link with an external program (COM technology). The Damping values in modes that are situated in the range between
COM (Component Object Model) is an industrial binary-interface the base mode and mode m are calculated by means of linear
that enables inter-process communication of objects in a wide interpolation (24):
variety of programming languages. In the scope of program
development, it was used to link its own program based on (24)
object-oriented code written in C++ programming language,
with the external program "RFEM-/RSTAB-Zusatzmodul As simple linear interpolation is often insufficient, and as
RF-COM/RS-COM" [38]. This program extension offers two it potentially does not take into account possible nonlinear
possibilities and both of them are based on modal analysis. reduction in damping value at lower frequencies due to small
To begin with, it is possible to make an additional evaluation of contribution of stiffness matrix to the total damping, it is
the solution calculated using expression (13). The objective of the necessary to extend the set of data by extrapolation of results
numerical procedure is to define the system’s response in the until 2.5 m modes of oscillation. For the frequency range above
sense of displacement of points according to expressions (11) and the mode m, the extrapolation can be made using the following
(13) and this for the assumed harmonic forces that are proportional expression (25):
to the product of nodal masses m(s) and the corresponding
displacements of each point of the selected oscillation mode:
(25)
(20)
Finally, by solving the expression (26):
These forces act on synthetic correlation lengths LS, that are
equal to the distances between zero points of the normalised
oscillation mode. This model requires adoption of the
corresponding damping model. The authors of this paper (26)
suggest application of the Rayleighs’ proportional damping.
The differential equation of motion of the system under external
load F(t) that is variable in time can be expressed as follows:
the damping of the system becomes fully defined for each data
(21) set. Indices 1, i.e. 2.5 m in expressions for a and b point to the
set of data from which coefficients are calculated. A new data
where y(x) are displacements of points, is the velocity, and set is obtained by averaging the results obtained for the sets
is the acceleration. and . Graphical representation of
This traditional equation defines the system in which external relative damping as a function of circular frequency for various
forces are given as a function of the matrix of mass [M], oscillation modes enables observation of the set that has the
damping [C] and stiffness [K] of the system. Viscous damping is greatest reduction in damping value at lower modes. Coefficients
expressed as the combination of mass and stiffness, as follows: a and b are defined for such set using expression (26).
The calculation using time integration in frequency domain is
(22) iterative in character and is repeated until a desired harmonic
response is attained. The time step Dt and the final value of
Coefficients a and b will be determined according to the applied damping have the greatest influence on the shape of
following calculation procedure. Structural damping is adopted response of the system. Decrease of time step often positively
for the frequency corresponding to the basic mode in the amount influences attainment of a stable, harmonic response of the
as defined in [1] by converting the logarithmic decrement ds system.

GRAĐEVINAR 70 (2018) 9, 793-809 799

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

An unstable and stochastic response can be noted for very low in an obvious way by assumptions of the vortex-resonance
damped structures, especially for a girder composed of mixed model, according to algorithm shown in Figure 6. Thus, for a
materials. In this case, the adequacy of this model can become tapered structure, it can not be stated with certainty that the
questionable. In the engineering practice, the procedure in which position of "antinodes" with the maximum value of normalised
damping is increased by appropriate coefficients is in most cases oscillation mode represents the position of critical cross-
sufficiently accurate for preliminary calculations. Calculated section for the vortex separation, considering the change in the
internal forces are also multiplied by the same coefficient. As diameter or the width in points close to the critical one.
simple multiplication of coefficients a and b can lead to results Vickery and Clark [22] showed that, in case of these structures,
that are not on the side of safety, the authors suggest the use there is a great probability of occurrence of critical cross-section
of the Lehrsch-relative damping for preliminary assessments b at the height z in oscillation mode j, with the validity of a
in such non-standard cases. The proposed coefficient of following expression (27):
multiplication of damping and the calculated internal forces
amounts to 10. This value should be taken as an approximation
only, and it is up to the practising engineer performing the (27)
calculation to carefully select the factor by which the damping
and internal forces should be multiplied. The use of the proposed For the situation of initial vortex separation in the zone close to the
coefficient has been proven appropriate on a large number of top of the structure, it can be expected that the increase in wind
preliminary estimations and feasibility studies in Germany, in the velocity could potentially "descend" the vortices towards lower
case of concrete telecommunication poles. Vertical extension of cross-sections, which is a direct consequence of the variability of
these poles was planned by adding steel or polymer reinforced structural geometry. It is clear that in the case of tapered structures
glass fibres or antenna on top of the structure. The use of this the spatial coefficient of correlation increases, and the aerodynamic
principle enables reduction of inconsistencies to a reasonable contribution of forces decreases, but, at the first glance, it remains
level. This approach is not adequate for detailed analyses or final unclear in what way this would affect the entire response of the
calculations as it may lead to disproportionality in response of the system, considering disproportionality of the change of these
system and adopted damping. values. Due to these unknowns that require further research of
The advantage of this calculation approach lies in the this phenomenon, the authors of this paper recommend, based
information about the time in which the structure will attain on practical experience, that special consultations with experts
its resonance maximum. If this time is too short, i.e. if it is from this field be made in case of structures with taper greater
established by integration of the differential equation of motion than 25 mm/m. It is not unusual that these unknowns be taken
(21) that the amplitude is attained several seconds after force into account in such cases by increasing the effective correlation
application, especially within the first two oscillation modes, length. This can be done using modal analysis in the frequency area
then this points to the low damping of the system and to great according to the right branch of the algorithm presented in Figure
susceptibility to the vortex shedding phenomenon. 6. Harmonic forces acting on the structure in j effective correlation
The use of this advantage, in combination with appropriate lengths Lj, with critical cross-sections bj, can be written as follows:
interpretation of Scruton number, which should not amount
to less than 6, has proven to be a good engineering practice as
evidenced by examples of several hundreds of steel, concrete and (28)
hybrid girders that have been designed and built by users of the
program that is based on the mentioned calculation procedure.
Another advantage offered by modal analysis is related to where the index i denotes the oscillation mode under
limiting and geometrical conditions that have not been covered consideration.

Table 1. Comparison with reference structure (b = 914 mm)

ni.y [Hz] δ clat St vcrit [m/s] M [t/m] K Sc Kw yF.max [mm] Source

1.721 0.015 0.241 0.20 7.87 0.090 0.129 2.59 0.465 128 This investigation
1.720 0.012 0.250 0.21 7.50 0.090 0.133 2.10 0.560 183 [16]
1.700 0.012 0.200 0.20 7.80 - 0.130 2.10 0.525 149 [39]
2.017 0.015 0.200 0.20 9.22 0.075 0.128 2.16 0.488 132 [35]
1.721 0.012 0.287 0.21 7.49 0.090 0.128 2.07 0.521 192 [35]
1.721 0.012 0.250 0.21 7.49 0.090 0.128 2.07 0.496 159 [35]
1.721 0.015 0.242 0.20 7.86 0.090 0.128 2.59 0.474 130 [35]
1.721 0.015 0.200 0.20 7.86 0.090 0.128 2.59 0.452 103 [35]

800 GRAĐEVINAR 70 (2018) 9, 793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

mode. This leads toward various

oscillation amplitudes, i.e. various states
of stress during the defined service
life of the structure. Thus, during their
construction and use, structures are
subjected to a spectrum of harmonic
forces, for which structural resistance
to material fatigue has to be proven.
Eurocode [40] offers two possibilities in
this respect.
The first calculation procedure is based
on the known Palmgren-Miner rule. The
same rule is applied in EN 1993-1-9 [40],
and can be written as follows:

(29)

where ni(Ds) is the number of cycles of

stress load for a specified stress range
Ds = smax - smin, for which Ni(Ds) cycles
of load is expected before the structure
suffers damage. D stands for total
damage.
For the selected connection detail, the
resistance to fatigue is defined by the
material fatigue curve (S-N curve), Figure
7. Using this method, the stress spectrum
Figure 6. Calculation algorithm for modal analysis in resonant area using Component Object is treated, in a simplified way, as a set of
Model (COM) technology n load blocks with the constant stress
range Ds = smax - smin.
In order to evaluate the calculation model presented in this The resistance to fatigue is mathematically estimated by means
paper, a comparison was made with the reference structure of accumulation of individual damage caused by individual stress
that has often been analysed in literature: steel chimney in blocks of each stress set. The number of stress sets is defined
Aachen, Germany. The comparison of results is presented in by the number of analysed frequency modes and the number of
tabular form, as shown in Table 1. A good correspondence critical positions in which the vortex separation occurs.
of results can be noted for oscillations in the natural mode,
with a slight deviation from the measured amplitude. The
oscillation amplitude of 140 mm was measured at the
frequently observed frequency of 1.72 Hz [16, 31]. The results
are related to the assumption of fixed support at the bottom
of the structure, which is why they deviate from individual
results that are based on the assumption of elastic support,
cD = 5.85·105 [kNm/m]. The calculated deviation of 8.6% from
the measured amplitude can be explained by simplifications
in the calculation, and by imperfections of the calculation
model.

3.2. Fatigue analysis

From the above presented theory, it is clear that a structure

can oscillate at various frequencies and, additionally, the vortex
separation can occur at various heights in the same frequency Figure 7. Calculation principle using cumulative damage method

GRAĐEVINAR 70 (2018) 9, 793-809 801

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

Proving fatigue resistance according to the theory of cumulative The method of equivalent damage factors with the constant
damage is generally a highly complex task for the engineering equivalent stress range is widely used in the case of steel
practice. The set of stress ranges is required for successful structures of cylindrical type, such as steel chimneys. These
fatigue check, i.e. for defining the sum of damage. However, the structures are characterized by oscillation in fundamental
number of stress-ranges due to vortex shedding is limited , and mode, and so this mode is normally taken into account when
computation according to this method is not overly demanding determining fatigue of material. Here it is possible to determine
in the case of vortex shedding. A possible alternative to this the equivalent damage factor, i.e. the equivalent stress range
approach is the use of the concept of equivalent damage factors. without much complications by means of (9.3) from [41] or (32).
For this computation approach, defined in EN 1993-1-9 [40], On the other hand, the oscillation in second mode is typical for
the equivalent stress range DsE,2 can be defined according to: tapered structures, and so this method would prove to be overly
complex and, in the authors’ opinion, inadequate for everyday
(30) use, as it requires adoption of computation principles that are
not defined by Eurocode. This method will therefore not be used
where : in the computation example presented in this paper.
gFf - partial factor for actions, i.e for the equivalent constant It should be noted that Eurocode defines the value of partial
stress ranges coefficient for verifying resistance to fatigue gMf as ranging from
DsE,2 - equivalent constant amplitude stress range for 2 million 1.0 to 1.35 depending on damage-tolerant and safe-life criteria,
cycles and consequences of possible failure. It is not unusual to use
Qk - characteristic value of a single variable action. the value of 1.0 for this parameter in the case of structures
for which regular maintenance and checking of connection
Proof of resistance to fatigue can be expressed as follows (31): details is obligatory. However, as structures subjected to vortex
shedding are typically situated in urban areas, it may reasonably
gFf · DsE,2 ≤ Dsc / gMf (31) be stated that their failure would represent an imminent danger
for the surrounding area. In such cases, according to the authors’
where: opinion, and based on the damage-tolerant and safe-life
Dsc - reference value of fatigue strength at 2 million cycles concept, it is necessary to adopt at least the partial coefficient
gMf - partial factor for fatigue strength. gMf = 1.15 for significant consequences in case of failure. Based
on the aforementioned theory, the nominal stress calculated in
The equivalent damage factor is specified for individual types closed form for cross-section requiring calculation of resistance
of structures in appropriate parts of Eurocode EN 1993. In case to fatigue can be presented in the following form (33):
of vertical load-bearing elements presented in this paper, it is
appropriate to use the equivalent damage factor according
to EN 1993-3-1 [41] and EN 1993-3-2 [42]. For an arbitrary (33)
spectrum, an equivalent stress range corresponding to 2 million
oscillation cycles can be determined according to Eurocode, but
with some restrictions. In fact, material fatigue curves from EN where Wij is the moment of resistance of critical cross-section,
1993-1-9 define double slope, m = 3 or m = 5. As only one slope and Mij(t) can be calculated using the mentioned calculation
of the S-N curve is defined in [41] by expression (9.3), there is a procedure based on the finite element method according
limitation that cannot be avoided without adopting assumptions to expression (17), or based on time integration in resonant
that are not covered by the standard. In a general case, for area, i.e. using modal analysis (Figure 6). Calculated stresses
one-sided slope of the S-N curve, according to expression (9.3) constitute nominal stresses, based on net area of cross-section
from [41], the reference stress range for the selected reference of the basic material. Although classification of calculation
number of cycles can be written as follows (32): approaches for estimating fatigue of material is not the subject
of this paper, it should be noted that the method of nominal
stresses is a conservative global approach that requires the least
(32) amount of time. Figure 8 shows calculation approaches that are
most often used in current practice. It can be noted that the
accuracy of results obtained using the nominal stress method
where: decreases with an increase in complexity of the detail that is
DsE,ref - 
reference stress range for the selected reference being analysed. In addition, the S-N curve for a particular detail
number of cycles Nref is often not available in practical situations, which additionally
ni - number of stress cycles in spectrum block i complicates the calculation. In such cases, the use of the hot-
Dsi - stress range in spectrum block i spot-stress method is obligatory. This is only very superficially
m - slope of the fatigue curve, i.e. stress exponent of the treated in [40] and so it is often necessary to use additional
fatigue curve. literature sources, primarily IIW recommendations [43] and

802 GRAĐEVINAR 70 (2018) 9,793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

DNV recommended practice [44], especially with regard to the It remains unclear what limits of the critical wind velocity
rules considering adoption of the size and type of final elements should be adopted in the analysis. It has already been stated
and the stress extrapolation rules. The ENS-Method (Effective that the deviation of wind velocity that exceeds the one causing
Notch Stress Method) is an even more comprehensive method, lock-in is not included in empirical parameter of the effective
and the most complex procedure involves the use of LEFM correlation length, and this fact has to be taken into account
(Linear Elastic Fracture Mechanics) approach. during estimation of the number of oscillations the structure
is subjected to during excitation. The range of critical velocity
limits has been extensively studied by professional community.
Two limit cases can be observed. Ruscheweyh [13, 30] states
that non-symmetry of the frequency range can be noted, which
was used by Clobes et al. [49] when defining eccentricity of 2/5
as related to 3/5 of the total resonance area. An alternative
approach was proposed by Simiu and Scanian (summarized in
[50]), where the reference critical velocity is defined as the lower
limit of critical area, ucrit,low = vcrit, with maximum bandwidth of
≈ 0.30 (ucrit,high ≈ 1.30ucrit,low ).
Figure 8. Comparison of the most frequently used fatigue analysis The mean value of bandwidth of e0 = 0.20 is adopted in this
methods paper according to [1], with definition of the already mentioned
eccentricity according to [49], (Figure 9). Results are calculated
3.3. Assessment of number of stress cycles and lock- for the statistical model based on the Weibull distribution
in phenomenon with the shape and scale parameters adopted for the location
at which the structure is situated. The comparison is made
The theory presented by Repetto and Solari can also be used with the proof of material fatigue that is calculated based on
for the calculated nominal stresses. These authors offered a the number of cycles according to EN 1991-1-4 (E.10) [1], for
closed solution for fatigue analysis due to oscillation of slender the standard function of density distribution with the shape
cantilever beams parallel and perpendicular to the direction of and scale factors of k = 2 and u0 = 1/5 um,Lj, where um,Lj is the
wind [45-47]. The approach presented in [45, 47] can in principle reference mean velocity of wind at the height of critical cross-
be adopted, with additional generalisation based on assumptions section at which the vortex is formed.
of the vortex-resonance model and according to [48]: Future research can be oriented towards application of numerical
-- the resonance method assumes harmonic response of the methods based on CFD (computational fluid dynamics), as in [51],
system in lock-in state; and towards experimental research, as in [52]. These approaches
-- for all reference critical velocities entering in the range of enable study of application of the system for reducing resonant
resonant frequency, the calculated resonance response is oscillation, such as TMD (Tuned Mass Damper) systems or spiral
considered to be the maximum value in this critical range, Ds aerodynamic dampers (helical strakes). It should be noted that
= smax - smin = 2smax. transition zone at the occurrence of lock-in still constitutes
an unexplored area. That is why the authors plan to use CFD
As a consequence of these assumptions, the defined stress programs to undertake new research of the influence of lock-
amplitude probability can be treated as unique, P(Dsn), which is in on the occurrence of additional harmonic components that
why the total number of structural oscillations in the i-th mode amplify the oscillation amplitude.
of oscillation ni(Dsir) (34) during the time interval T expressed in
years, amounts to:

(34)

where PU(Ui*) defines the probability of occurrence of critical

velocity that can cause resonant state in critical cross-section
at the height z = z*, in mode i. When wind direction is taken
into account, i.e. when there is a significant variation in local
roughness of terrain in a particular direction, the number of
structural oscillations ni(Dsir) (34) becomes:

(35)

where Pq(qir) defines the probability of occurrence of the mean

critical velocity of wind in the direction r. Figure 9. Lock-in model

GRAĐEVINAR 70 (2018) 9, 793-809 803

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

4. Numerical example oscillation modes for Hcrit = 36.36 m and Hcrit = 21.73 m are given
in Table 3, while the bending moment values for cross-sections
A steel pole for stadium lighting, of variable cross-section, 37.9 in which fatigue resistance is checked are given in Table 4.
m in height, is considered. Figure 10. Weibull distribution according to [53] – for measuring
The objective is to calculate, using the previously defined instrument placed at the height of 10 m, Hannover
calculation procedure, the pole structure oscillation amplitude
due to vortex shedding, and to provide the check of material
fatigue for the critical details using the Palmgren-Miner theory
of cumulative damage.
As to the mean wind velocity profile, design parameters
correspond to the wind zone 2, and terrain category II-III,
according to [1, 26]. It is assumed that the structure is located in
Hannover (52°27‘00’’N 09°42’00’’E), with the mean coefficients
of Weibull distribution amounting to A = 4.50, k = 1.79, according
to [53], Figure 10. At the top, the structure measures 500 mm
in diameter, the slope of the structure is 25 mm/m, and the
wall length and thickness of 16-angle cross-sections amount
to: 9700 mm/5 mm; 1200 mm/6 mm, 1200 mm/8 mm, 9300 Connection details for which the fatigue check was made are
mm/12,5 mm – measuring from top to bottom. shown in Figure 12. The category 80 according to Table 8.4
It is assumed that the critical position of vortex shedding is [40] was adopted for the cross-section at the bottom of the
at antinode points, i.e. at the positions of local amplitudes of pole (Figure 12, bottom). In this case, the critical detail under
oscillation modes, although it can be seen in Figure 11 that the consideration is the fillet weld at the connection between the
vortex separation at these cross-sections is not likely due to the annular horizontal compression ring and the steel pole. The
effect of aerodynamic disturbance of fluid flow by the platform detail at the top part of Figure 12 is the longitudinal automatic
and the supporting structure of floodlights for stadium lighting. weld that has to be checked for fatigue at the point of slip-joint
Frequency characteristics of the structure were tested for the case used for the extension of the structure, where local stiffness
of 1D and 3D structure, and less favourable calculation results were change occurs due to the overlap of steel shell elements with
finally adopted as relevant, cf. Figure 11. Nodal masses shown in the variable thickness. The category 140 was adopted for this
Figure 11 are presented in tabular form for clarity reasons. extension, Figure 12. Control checks of other details will not
Calculation parameters adopted in the analysis are given in be shown in order to simplify the numerical example. It can
Table 2. Calculated values for vortex shedding in the first two be noted that bending moment at the bottom of the pole has

Table 2. Computation parameters

ni.y [Hz] me [kg/m] St [-] δs [-] ρ [kg/m3] ν [m2/s]

1.095 170.213 0.18 0.015 1.250 1.5 10-5

3.821 230.468 0.18 0.015 1.250 1.5 10-5

Table 3. Calculated values for vortex shedding in first two oscillation modes

ni.y [Hz] Hcrit [m] vcrit [m/s] clat [-] Sc [-] Re [-] K [-] Kw [-] yF.max [mm]

1.095 36.36 3.29 0.700 13.98 118477 0.131 0.245 26.8

3.821 36.36 11.47 0.386 18.93 413491 0.184 0.463 28.9

3.821 21.73 19.03 0.200 6.89 1136788 0.184 0.463 68.3

Table 4. Bending moments [kNm]

ni.y [Hz] z = 0 [m] z = 9.3 [m] z = 19.2 [m] z = 29.5 [m]

1.095 78.0 50.8 23.8 5.6

3.821 272.6 75.5 73.1 54.5

3.821 643.9 178.3 172.8 128.8

804 GRAĐEVINAR 70 (2018) 9, 793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

Nr [-] z [m] m [kg] Nr [-] z [m] m [kg]

41 37.90 46.66 20 21.70 807.49
40 37.30 123.98 19 21.65 109.80
39 37.25 82.85 18 20.92 265.08
38 35.22 26.43 17 19.20 247.89
37 35.90 114.20 16 19.20 410.50
36 35.85 111.68 15 17.22 430.09
35 34.54 67.92 14 15.24 449.68
34 34.45 143.20 13 13.26 469.27
33 32.86 159.93 12 11.28 488.85
32 31.18 168.74 11 9.30 508.38
31 29.50 117.12 10 9.30 717.82
30 29.50 209.86 9 7.44 744.83
29 27.78 220.90 8 5.58 134.97
28 26.07 308.87 7 5.25 220.73
27 26.00 223.08 6 5.20 289.18
26 24.35 13.85 5 4.50 241.02
25 24.25 287.01 4 4.45 305.94
24 24.20 222.13 3 3.72 798.85
23 22.63 19.33 2 1.86 422.88
22 22.50 267.34 1 0.90 389.48
21 22.45 110.07 0 0.00 264.54

Figure 11. 3D-CAD Model of the structure, nodal mass distribution for 1D structure and oscillation modes for 3D girder

GRAĐEVINAR 70 (2018) 9, 793-809 805

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

the maximum oscillation value in the second mode for vortex Fatigue analysis results based on expression (29), for statistical
shedding at the height of 21.73 m – Table 4. data according to Weibull distribution as per [53], are presented
The maximum oscillation amplitude of 68.3 mm is also obtained in tables 5 and 6.
at oscillation in the second mode with vortex shedding at the
height of 21.73 m – Table 3. The dimensionless amplitude Table 6. 
Results of fatigue analysisfor cross-sections at various
yF,max/b amounts to 0.076 and so there is no need for iterative heights using cumulative damage concept, for Weibull
distribution according to [53] - total damage
increase of effective correlation lengths which, in the second
mode, amount to: L1 = 3.24 m and L2 = 5.38 m. Figure 12. Connection detail with selected categories and Wöhler-
lines (S-N fatigue curves)
Table 5. 
Result of fatigue analysis for cross-sections at various z [m] SD [-]
heights using cumulative damage concept, for Weibull wind 29.5 0.989
distribution according to [53]
19.22 0.008
9.3 0.000
z = 29.50 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 140 [N/mm2]
0 0.584
A = 4.50. k = 1.79. according to [53]
Second oscillation Second oscillation
First oscillation
mode, mode,
mode
Hcrit = 36.36 m Hcrit = 21.73 m
∆σ 5.69 55.90 132.07
n 1.98 × 108 5.19 × 107 6.82× 104
N Infinite 5.32× 107 5.32× 106
D 0.000 0.976 0.013
z = 19.20 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 140 [N/mm2]
A = 4.50. k = 1.79. according to [53]
Second oscillation Second oscillation
First oscillation
mode, mode,
mode
Hcrit = 36.36 m Hcrit = 21.73 m
∆σ 11.2 34.30 80.97
n 1.98 × 108 5.19 × 107 6.82× 104
N Infinite Infinite 8.34× 106
D 0.000 0.000 0.008
z = 9.30 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 140 [N/mm2]
A = 4.50. k = 1.79. according to [53]
The analysis of calculated values presented in Table 6 shows
Second oscillation Second oscillation
First oscillation that the structure is safe and that the fatigue resistance check
mode, mode,
mode based on the cumulative damage theory predicts no damage
Hcrit = 36.36 m Hcrit = 21.73 m
occurrence during the 50 years of lifetime for the mean wind
∆σ 11.50 17.10 40.41
profile according to Weibull distribution, adopted according to
n 1.98 × 10 8
5.19 × 10 7
6.82 × 104 [53]. For comparison, the fatigue resistance check was also
N Infinite Infinite Infinite provided for the number of load cycles defined according to
D 0.000 0.000 0.000
(E.10) [1], with the factors of shape and scale amounting to
k = 2 and u0 = 1/5 um,Lj, where um,Lj is the reference mean wind
z = 0.00 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 80 [N/mm2]
velocity at the height of critical cross-section in which vortex
A = 4.50. k = 1.79. according to [53] separation occurs. Results are presented in tables 7 and 8.
Second oscillation Second oscillation The analysis of values presented in Table 8 indicates that the
First oscillation
mode, mode, structure is not safe and that the fatigue resistance check
mode
Hcrit = 36.36 m Hcrit = 21.73 m based on the Palmgren-Miner theory predicts the damage
∆σ 8.20 28.50 67.37 occurrence during the 50-year design lifetime of the structure.
n 1.98 × 10 8
5.19 × 10 7
6.82 × 104 There is a big difference in calculated cumulative damage
when results from Table 6 are compared with those from Table
N Infinite 9.38 × 107 2.20 × 106
8 for the cross section at the height of 29.5m and for the cross
D 0.000 0.553 0.031 section at the bottom of the pole. The difference in calculated

806 GRAĐEVINAR 70 (2018) 9, 793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

damage can be described with factor 4.10 for the detail at the the number of cycles according to (E.10) [1], k = 2, v0 = vm,Lj /5
height of 29.5 m, and with factor 3.95 for the detail at the Displacements over time of the point at the top of the structure
bottom of the pole. at vortex shedding in critical cross-section at the heights of

z [m] SD [-]
Table 7. 
Result of fatigue analysis for cross-sections at various
heights using cumulative damage concept, for the assumed 29.5 4.057
number of cycles according to (E.10) [1], k = 2, v0 = vm,Lj /5
19.22 0.006
Table 8. 
Results of fatigue analysis for cross-sections at various 9.3 0.000
height using cumulative damage concept, for assessment of
0 2.307
z = 29.50 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 140 [N/mm2]
EN 1991-1-4 . k = 2. v0 = vm.Lj /5. eo = 0.20
Hcrit = 36.36 m and Hcrit = 21.73 m for the case of resonant
Second oscillation Second oscillation
First oscillation oscillation in the first and second modes, respectively, are
mode, mode,
mode
Hcrit = 36.36 m Hcrit = 21.73 m shown in Figure 13. Sensitivity to vortex-resonance of the
structure is greater for oscillation in the second mode where
∆σ 5.69 55.90 132.07
the oscillation amplitude attains approximately 7 seconds after
n 1.56 × 108 2.14 × 108 5.34 × 104 the application of harmonic forces (20). This information in
N Infinite 5.32 × 107 1.57 × 106 combination with the calculated Scruton number amounting to
6.89 (Table 3) leads to the conclusion that this is a border case,
D 0.000 4.023 0.034 and that further increase in pole slenderness or optimisation by
z = 19.20 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 140 [N/mm2] mass or stiffness reduction would potentially lead to significant
resonance problems.
EN 1991-1-4 . k = 2. v0 = vm.Lj /5. eo = 0.20
Figure 13. Displacements of yF point at the top of the structure over
Second oscillation Second oscillation time t, for a = 0.00376, b = 0.00123
First oscillation
mode, mode,
mode
Hcrit = 36.36 m Hcrit = 21.73 m
∆σ 11.2 34.30 80.97

n 1.56 × 108 2.14 × 108 5.34 × 104

N Infinite Infinite 8.34 × 106

D 0.000 0.000 0.006

z = 9.30 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 140 [N/mm2]

EN 1991-1-4 . k = 2. v0 = vm.Lj /5. eo = 0.20

Second oscillation Second oscillation
First oscillation
mode, mode,
mode
Hcrit = 36.36 m Hcrit = 21.73 m

∆σ 11.50 17.10 40.41

n 1.56 × 108 2.14 × 108 5.34 × 104

N Infinite Infinite Infinite 5. Conclusion

D 0.000 0.000 0.000 Free-standing structures are often subjected to cyclic vortex
z = 0.00 [m]. γMf = 1.15. γFf = 1.00. ∆σc = 80 [N/mm ] 2 oscillations. Actions from vortex shedding can be determined
by means of appropriate computation methods and adequate
EN 1991-1-4 . k = 2. v0 = vm.Lj /5. eo = 0.20
verifications can be made at critical points along the structure.
Second oscillation Second oscillation The calculation procedure for determining resistance of
First oscillation
mode, mode,
mode material to fatigue due to vortex shedding is presented in this
Hcrit = 36.36 m Hcrit = 21.73 m
paper, by assessing fatigue using the Palmgren-Miner method
8.20 28.50 67.37 of cumulative damage.
n 1.56 × 108 2.14 × 108 5.34 × 104 It is shown in the text in which way the wind profile influences
N Infinite 9.38 × 10 7
2.20 × 106 the prediction of the number of loading cycles and, finally,
how the same influences the total accumulated damage. The
D 0.000 2.281 0.026
authors’ experience shows that the results of cumulative

GRAĐEVINAR 70 (2018) 9, 793-809 807

Građevinar 9/2018 Marina Rakočević, Svetislav Popović

damage assessment do not always need to be more favourable The presented calculation approach based on vortex resonance
and economical when wind profile is used at the location where model from EN 1991-1-4 takes appropriately into account all
the structure is realized. Based on the use of the calculation significant parameters that bear influence on the estimation of
procedure presented in this paper on more than 1000 results.
concrete and hybrid telecommunication poles, the authors Inconsistencies and unknowns in the use of Eurocode, as
have established that, with regard to estimation of amplitude related to the selection of parameters and computation
and number of loading cycles, the Eurocode model shows methods, are presented in the paper. A special emphasis is
adequate results during calculation of structures situated in given to the significance of the presented iterative calculation
Central European countries. A considerable inconsistency has procedure in case of structures in which geometrical
been noted in assessment during design of steel structures conditions deviate from assumptions of the vortex-resonance
in Northern Europe, in littoral areas, and in zones with small model. In such cases, the effects of aeroelastic forces must
local roughness of terrain and laminar wind, such as airport be appropriately introduced in the analysis, by increasing the
zones. This points to the need for undertaking additional effective correlation length.
research of this phenomenon, which has to be transferred As the total damage is also influenced by oscillation in the
from the calculation aspect to the experimental area, involving wind direction, future research work should also concentrate
observation of response of existing structure in order to correct on the analysis of influences during such oscillation of
computation models. structures.
A proper calculation approach is possible only by using a
specialized program that implements the presented theory.

REFERENCES
[1] EN 1991-1-4:2005 - Eurocode 1. Action on structures. General [13]
Ruscheweyh, H.: Ein verfeinertes praxisnahes
actions. Wind actions. Berechnungsverfahren wirbelerregter Schwingungen von
[2] Model Code for Steel Chimneys - the CICIND Chimney Standard, schlanken Baukonstruktionen im Wind, Aufl. Inst. für Mech., Heft
ISBN 1-902998-16-2, Revision 2010, Zürich, Switzerland. 20, 1987.

[3] NRCC 48192, National Research Council Canada, User´s Gudie- [14] Ruscheweyh, H.: Practical experiences with wind-induced
NBC 2005 Structural Commentaries (Part 4 of Division B), ISBN vibrations, Journal of Wind Engineering and Industrial
0-660-19506-2, 1993. Aerodynamics, 33 (1990), pp. 211-218.

[4] DS410 E:2004, Code of Practice for Loads for the Design of [15] Ruscheweyh, H., Sedlacek, G.: Cross-wind vibrations of steel
Structures, Danish Standard Association, 2004. stacks - Critical comparisons between some recently proposed
codes, Journal of Wind Engineering and Industrial Aerodynamics,
[5] AS/NZS 1170.2:2002, Australian/New Zealand Standard, 33 (1988), pp. 173-183.
Structural design actions, Part 2: Wind Actions, 2005.
[16] Galemann, T., Ruscheweyh, H.: Untersuchung winderregter
[6] ESDU 96030. 1996. Response of structures to vortex shedding: Schwingungen an Stahlschornsteinen. Forschungsbericht Projekt
Structures of circular or polygonal cross-section. Engineering 163 - Lehrstuhl für Stahlbau, RWTH Aachen, 1992.
Sciences Data Unit (ESDU International, London, UK), ESDU Data
Item 96030. [17] Galemann, T., Ruscheweyh, H.: Untersuchung Winderregter
Schwingungen an Stahlschornsteinen - Stahlschornsteinbau
[7] ESDU 96031- Computer programs for response of structures to fortschlictliche Lösungen, Essen : Vulkan - Verlag, 1992.
vortex shedding. Structures of circular or polygonal cross section.
Part 1. Microsoft Excel module, Engineering Sciences Data Unit [18] Ruscheweyh, H.: Interferenzwirkung zwischen Schornstein und
(ESDU International, London, UK), ESDU Data Item 96031. Gebäude auf die wirbelerregte Schwingung; WTG-Mitt, Nr. 5,
Baukonstruktion unter Windeinwirkung, 1992.
[8] Holmes, D.J.: Wind Loading of Structures, 3rd Edition, CRC Press,
2014. [19] Niemann, H.J., et. all: Vibrations of Chimneys under the Action of
the Wind - Proceedings of the 9th International Conference on
[9] Ruscheweyh, H.: Vergleich der zwei Berechnungsverfahren für Structural Dynamics, EURODYN 2014 Porto, Portugal, 30 June - 2
Querschwingungen, Deutsches Ingenieurblatt - Forschung + July 2014 A. Cunha, E. Caetano, P. Ribeiro, G. Müller (eds.) ISSN:
Technik, DIB 4, 2015. 2311-9020; ISBN: 978-972-752-165-4.
[10] Clobes, M., Willecke, A., Peil, U.: Wirbelerregung von [20] Holmes, J.D.: Response of cylindrical structures to vortex shedding
Stahlschornsteinen-zwei Grenzzustände der Tragfähigkeit und in the natural wind - 13th Australian Fluid Mechanics Conference
Vorschlag für die Bemessung, FT Windtechnologie, Band 87, Mai Monash University, Melbourne, Australia, 13-18 December 1998.
2012.
[21] Verboom, V.K., van Koten, H.: Vortex excitation: Three design rules
[11] Dyrbye, C., Hansen, O.S.: Wind Loads on Structures, Wiley, January tested on 13 industrial chimneys - Journal of Wind Engineering
1997. and Industrial Aerodynamics, 98 (2010), pp. 145-154.
[12] Hansen, S.O.: Vortex-induced vibrations of structures, Structural [22] Vickery, B.J., Clark, A.W.: Lift or across-wind response of tapered
Engineers World Congress 2007, November 2-7, 2007. Bangalore, stacks, Journal of Structural Division, ASCE, 98 (1972), pp. 1-20.
India.

808 GRAĐEVINAR 70 (2018) 9, 793-809

Calculation procedure for determining wind action from vortex-induced vibration with verification of fatigue strength of steel structures Građevinar 9/2018

[23] Vickery, B.J., Basu, R.I.: Across-wind vibrations of structures of [39] Ruscheweyh, H.: Windlastannahmen für turmartige Bauwerke. In:
circular cross-section. Part I - Development of a mathematical Referatensammlung. Türme, Maste, Schornsteine in Stahl. Fürth,
model for two-dimensional conditions. Journal of Wind pp. 3-1 do 3-20, Berlin, Köln: Beuth 1992.
Engineering and Industrial Aerodynamics, 12 (1983), pp. 49-73. [40] EN 1993-1-9:2005 - Eurocode 3: Design of steel structures - Part
[24] Vickery, B.J., Basu, R.I.: Across-wind vibrations of structures of 1-9: Fatigue.
circular cross-section. Part II - Development of a mathematical [41] EN 1993-3-1: 2009 - Eurocode 3: Design of steel structures -
model for full-scale application. Journal of Wind Engineering and Part 3:1 - Towers, masts and chimneys - Towers and masts.
Industrial Aerodynamics, 12 (1983), pp. 75-97.
[42] EN 1993-3-2: 2009 - Eurocode 3: Design of steel structures -
[25] Vickery, B.J., Basu, R.I.: The response of reinforced concrete Part 3:1 - Towers, masts and chimneys - Chimneys.
chimneys to vortex shedding. - Engineering structures, 6 (1984),
[43] IIW Document IIW-1823-07, ex XIII-2151r4-07/XV-1254r4-07:
pp. 324-333.
Recommendations for fatigue design of welded joints and
[26] DIN EN 1991-1-4/NA:2010-12 - Nationaler Anhang - National components - International Institute of Welding, December 2008.
festgelegte Parameter - Eurocode 1: Einwirkungen auf Tragwerke
[44] DNV-GL Recommended Practice - Fatigue design of offshore
- Teil 1-4: Allgemeine Einwirkungen - Windlasten
steel structures; DET-Norske-Veritas AS Germanischer Lloyd SE
[27] NF EN 1991-1-4/NA:2005 - Eurocode 1 : Actions sur les (DNV GL AS), April 2016;.
structures — Partie 1-4 : Actions générales — Actions du vent
[45] Repetto, M.P., Solari, G.: Dynamic along-wind fatigue of slender
Annexe nationale à la NF EN 1991-1-4, Actions générales —
vertical structures, Engineering Structures, 23 (2001) 12, pp.
Actions du vent
1622-1633.
[28] Scruton, C.: An Introduction to Wind Effects on Structures - Oxford
[46] Repetto, M.P., Solari, G.: Diagnosis of Fatigue Collapses of
University Press, Oxford, UK, 1981
Slender Structures due to Aerodynamic Wind Actions; BBAA
[29] Rumman, W.S.: Basic structural design of concrete chimneys. VI International Colloquium on: Bluff Bodies Aerodynamics &
ASCE, Journal of the Power Division, 96 (1970), pp. 309-318. Applications, Milano, Italy, 2008.
[30] Ruscheweyh, H.: Dynamische Windwirkung an Bauwerken, Band [47] Repetto, M.P., Solari, G.: Crosswind Response Induced Fatigue of
2 - Praktische Anwendungen. Bauverlag GmbH, Wiesbaden und Slender Structues, 8th ASCE Specialty Conference on Probabilistic
Berlin,1982 Mechanics and Structural Reliability, University of Notre Dame,
[31] Dickel, T., Rothert, H.: Anmerkungen und computerorientierte Indiana, 2000.
Berechnungshilfen zum Querschwingungsnachweis nach DIN [48] Pasto, S.: Fatigue-Induced Risk Assessment of Slender Structures
4131 und DIN 4133, Bauingenieur, 69 (1994), pp. 403-408. with Circular Cross-Section at Lock-In; Dissertation, Department
[32] Verwiebe, C.: Fatigue Problems of Steel Chimneys due to Vortex of Civil Engineering of the TU Carolo-Wilhelmina at Braunschweig,
Excited Vibrations in the Second Mode - IDN - CICIND REPORT, 20 2005.
(2004) 2. [49] Clobes, M., Willecke, A., Peil, U.: Vortex-induced vibrations of
[33] Guide for the assessment of wind actions and effects on slender structures considering long-term wind profile statistics,
structures, National Research Council of Italy, June 2010. Proceedings of the 8th International Conference on Structural
[34] Dickel, T., Rothert, H.: Berechnungshilfen für einfache Fälle des Dynamics, EURODYN 2011 Leuven, Belgium, 4-6 July 2011, ISBN
Querschwingungsnachweises nach DIN 4131 und DIN 4133, 978-90-760-1931-4.
Bauingenieur, 69 (1994), pp. 239-246. [50] Simiu, E., Scanlan, R.H.: Wind effects on structures. Wiley, New
[35] Dickel, T.: PC-Programm EASYKAM: Nachweis der York, 1986.
Querschwingungen DIN 4133/4131, Schwabach 1988. [51] Vegard, H., Tahsin, T., Elif, O.: Full-scale CFD investigations of
[36] Dickel, T.: Zur Verwendung von Eigenformen in DIN 4131 und DIN helical strakes as a means of reducing the vortex induced forces
4149, Tipp des Monats 10/04, Prüfamt für Baustatik, Nürnberg, on a semi-submersible, Ocean Engineering, 137 (2017) C, pp.
2004. 338-351, https://doi.org/10.1016/j.oceaneng.2017.04.014
[37] Novak, M., Tanaka, H.: Pressure Correlations On a Vibrating [52] Juan, S., Jiasong, W., Shengping, L., Qilong, T.: VIV suppression
Cylinder. Proc. 4th International Conference Wind Effects on for a large mass-damping cylinder attached with helical strakes,
Building and Structures, London 1975. Cambrige: University Press Journal of Fluids and Structures, 62 (2016) C, pp. 125-146, https://
1977. doi.org/10.1016/j.jfluidstructs.2016.01.005
[38] https://www.dlubal.com/de/produkte/rfem-und-rstab- [53] Troen, I., Lundtang Petersen, E.: European Wind Atlas. Roskilde:
zusatzmodule/sonstige/rf-com, pristupljeno 21.01.2018. Risø National Laboratory, Denmark, 1989.

GRAĐEVINAR 70 (2018) 9, 793-809 809