You are on page 1of 8

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

Porto, Portugal, 30 June - 2 July 2014
A. Cunha, E. Caetano, P. Ribeiro, G. Müller (eds.)
ISSN: 2311-9020; ISBN: 978-972-752-165-4

Vibrations of Chimneys under the Action of the Wind
Hans-Jürgen Niemann1, Francesca Lupi2, Rüdiger Hoeffer3
1
Department of Civil and Environmental Engineering, Ruhr-Universität Bochum, D-44799 Bochum, Germany
2
Department of Civil and Environmental Engineering, Università degli Studi di Firenze, 50139 Florence, Italy
3
Department of Civil and Environmental Engineering, Ruhr-Universität Bochum, D-44799 Bochum, Germany
email: hans-juergen.niemann@rub.de, francesca.lupi@dicea.unifi.it, ruediger.hoeffer@rub.de

ABSTRACT: Vibrations of chimneys under the wind action occur in-line with the wind as forced, random vibrations and cross-
wind as vortex-induced vibrations. The former become important at extreme wind speeds, regular vortex shedding causes the
latter and occurs at any wind speed. In the case of vortex resonance, when considerable vibration amplitudes develop, the
structural displacements alter the fluid force, which is known as an aero-elastic coupling of force and structural response.
Regarding the random load caused by turbulent wind, Davenport's pioneering concept of the gust response factor was applied to
develop a simplified equivalent static load. Two different models are available regarding the design for vortex resonance. One is
Ruscheweyh's approach to deal with vortex resonance as a forced vibration and to incorporate the aero-elastic effects in model
parameters derived from experimental data. In the second model, the design for vortex resonance is based on Vickery & Basu's
model of a self-limiting response process. The Eurocode contains both models but relies primarily on the first model. The new
CICIND model code counts solely on the second approach. Experimental observations indicate the relative importance of
parameters such as the Scruton number, the intensity of turbulence and the mode of vibration. Regarding the in-line wind load,
the mechanical model also enters into the discussion. In some cases, the chimney dimensions indicate that a beam behaviour is
neither realistic nor on the safe side. Then, a shell model has to be applied and the wind load cannot be considered as a force.
Rather, pressures have to be utilized. The paper discusses the effect of these parameters and deals with the principal results
concerning the along wind action and the design for vortex shedding.

KEY WORDS: Chimneys; gust wind load; vortex resonance; fatigue.

1 INTRODUCTION distinctive feature of vortex excitation with respect to other
Vibrations of chimneys under the wind action occur in the aeroelastic phenomena. The two models differ in the manner
along-wind and in the across-wind direction. they incorporate aeroelastic effects. One is Ruscheweyh's
Along-wind vibrations consist of a background, quasi-static approach, which introduces the concept of the effective
contribution which reflects the stochastic and broad-band correlation length, whose value depends on the vibration level
nature of the incoming flow. Depending on the natural and is calculated by iteration. The model parameters are
frequency of the structure, a dynamic contribution due to derived from experimental data. The other one is the
resonance to turbulence may also arise. Vickery&Basu model, which follows Scruton's approach [2]
Across-wind vibrations are due to the alternating vortex and treats the aeroelastic lift force as a component in out-of-
shedding in the wake. Vortex resonance occurs as the wind phase with the motion, acting as a negative aerodynamic
velocity approaches the critical value and, in case damping. The non-linear self-limiting behaviour of the
considerable vibration amplitudes develop so that the response is treated by the Vickery&Basu model in a linearized
structural displacements alter the fluid force, aeroelastic manner, by introducing - as model parameters - the negative
forces may arise. Turbulence of the incoming flow also plays aerodynamic damping parameter for small amplitudes of
a role, both on the stationary cylinder, because it reduces the oscillation and the self-limiting amplitude on the rms of the
coherence of the lift force along the cylinder axis and widens response.
the bandwidth of the lift spectrum, as well as on the vibrating The paper applies the design load models for the along wind
cylinder. action and for vortex resonance to two real samples of
All the design load models to account for along-wind industrial chimneys (one is made of reinforced concrete and
vibrations are based on the gust response concept, originally one is made of steel, see Table 1 and Table 2, respectively)
developed by Davenport [1]. Essentially, it amplifies the mean and discusses the principal results.
wind force by applying the gust response factor G. The latter Table 1. Geometry of sample chimney n.1 (concrete)
is a structural parameter which allows defining an equivalent
static load, intended to reproduce the effects of the stochastic z [m] z/h r [m] t [m]
wind loading process on the most important structural 213.0 1.000 11.675 0.350
stressing. 211.5 0.993 11.675 0.350
Two different models are available regarding the design for 205.0 0.962 11.675 0.350
vortex resonance. They include the modeling of aeroelastic 180.0 0.845 11.675 0.350
forces and ensure a self-limiting response cycle, which is the 165.0 0.775 11.675 0.350

1385

so that: 1386 .350 =  +  (4) 130.0063 In equation (5). and the integral length scales All gust wind models presently utilized in modern wind of the oncoming flow.000 0. kp is the peak factor. It depends on one hand on the wind flow parameters e.650 with: σ 1 σ 1 B= and R = Table 2.0 0.0 0. the peak factor has been chosen as kq = 3. intended to reproduce the effects of the 2 stochastic wind loading process on the most important where Iv(z) is the turbulence intensity at height z. considered cantilevered.825 0.650 G = 1 + 2k  I z B  + R (5) 0. d is the diameter of the chimney. The wind force. The load fluctuations due to wind turbulence provide a component: wz = c z c) z  ∙ C ∙ A+. zs is the height at which the reference flow 17. the mean wind 2.0 0. Davenport [1] approximations.0 1. It can be dependent (quasi-static and dynamic) components.000 0.800 0. vertical structures and their response Furthermore. fluctuating in short-term maximum occurring within the 10-min interval of time: the extreme mean wind. in particular the lowest natural frequency.600 expression for the gust response factor is obtained: 25.145 11.516 11.403 0. namely (σq/qm) ≅ 2Iv and qm ≅ ρV²/2.350 the intensity of turbulence Iv = σv/Vm if the structural 55.∙ q z broad band excitation of the structure. Geometry of sample chimney n.7Iv) 1. the turbulence intensity. It is defined as the the mean qm of the wind velocity pressure q(t). express the wind force acting on a structure or a structural σE.000 11. The gust response 16. E E + k  σ σ G= = = 1 + k (3) The use of the profile of the peak velocity pressure (7) is E E E one of the merits of this model.208 11.530 0.0063 account the reduction of the load effect due to the non- simultaneous occurrence of the load peaks over the structural surface whereas R accounts for the amplification of the load 2 ALONG WIND ACTION effect due to resonance with wind turbulence. Em: {1 + (0. B takes into 0.675 0. namely the mean (static) and time corresponds to the gust wind speed in a 1-sec gust.0 0.0 0.610 11. line-like structures such as chimneys: Design wind load models in the Standard Codes are wz = G ∙ w z (1) revisited applications of Davenport's approach. the following 30.423 11.Proceedings of the 9th International Conference on Structural Dynamics. it accounts on the other hand for loading codes go back to the classical Davenport approach.0 0.258 11. It structural parameters such as the size of the structure and its amplifies the mean wind force applying the gust response dynamic behaviour.0 0. (9) σER and σEB. factor G.0 0. z is the height above the ground. He defined shown. which is the ratio of The Eurocode introduces the following formulation to the peak of the response fluctuation to its standard deviation.775 0. EURODYN 2014 σ σ σ 151. and the damping. because the profile of the wind force over the building height results in a more realistic where σE is the standard deviation (or rms-value) of the image of the local gust loads at each level. whereas the wind force in equation (1) is an (8) equivalent static load.800 0. The rms response is split into resonant and quasi-static (background) components. In it.1 Design models for gust wind load profile.350 E  E E 90. given as a line load. In the Eurocode formulation. The mean wind force is a real q z = "1 + 7I z$ physical quantity.403 0.675 0.350 110.0 0. It to the wind action.8}.530 0. The structural stressing.550 behaviour is linear and the aerodynamic transmittance is 44. not a load parameter.0 0.800 0.0063 parameters are utilized for determining G.675 0. fluctuating response.352 11.056 11.350 Davenport showed that both components are proportional to 75.675 0. The peak velocity pressure is the sum w z = C z ∙ dz ∙ q z (2) of the mean qm and the standard deviation σq amplified by the peak factor kq as in equation (7): q z = q z + k ! σ! in which CD is the aerodynamic drag coefficient. transition from equation (7) to (8) involves two convenient To derive the gust response factor.2 0.600 12. qm(z) is the 10-min mean of the (7) ρ& V  z velocity pressure at level z.g.709 11.825 0.9 0. for slender. Introducing factors R and B.425 0.675 0. G is the Davenport approach is that the Eurocode bases the wind force gust response factor.400 0. an important difference to the In equation (1).0 0. The mean wind force wm(z) is based on on the peak (or gust) velocity pressure qp.. that the exact value of equation (7) exceeds the the gust response factor G as the ratio of the peak wind effect. approximation of equation (8) by a factor of Ep to the mean response.2 (steel) (6) E 2I E 2I z [m] z/h r [m] t [m] 40.0063 factor is a structural.117 11.5.0 0.600 quasi-stationary. is appropriate the related mode shape.

and their variation with respect CICIND 0. is not reflected in any of the codes.1 due 1. Moreover.2 mm) and to a rough concrete chimney surface (ks = Figure 3. The dynamic factor 0.smooth structures for which the resonant contribution R is much 0.8 0.1 the drag coefficient CD. able to reproduce 0 5000 10000 15000 20000 25000 30000 35000 40000 Wind force [N/m] the peak bending moment at the base of the chimney.9 [6]).7 wm and the gust wind load wg.6 EN . The local 0. Differences in the results are both due to respectively).1 a free-end (Lupi. the surface roughness.2 any profile appropriate to reproduce the distribution of the real 0. [7]). The CICIND wind load model (equation (11)) is unique EN .9 WT resonant amplification of the gust effect.1 0 given by the Eurocode. is the Eurocode modification to the Davenport approach.6 to the height z above the ground: z/h 0. Bending moments on the sample chimney n.3 height for determining it. despite the WT" stands for the Eurocode approach (9) applied to the non.7 EN .8 CICIND is always > 1 in principle. 0. ratio. However. the three-dimensionality of the chimney and 1 of the flow past it is responsible for the spanwise variation of 0.3 0. the gust load may be given 0. [4]. Proceedings of the 9th International Conference on Structural Dynamics. [5].5 0.8 EN . equations (9) and (11).1. furthermore the gust load profile does not account for variations of the diameter d and of Figure 2.1 after 1) wind tunnel tests (WT).8 in that it gives individual expressions for the mean wind load EN . equation (10) is obtained.4 Figure 1 shows the aerodynamic force coefficient 0. the distribution is evident in the CICIND result. the CICIND 0 adopts a linear profile with wg = 0 at z = 0.rough z/h smaller than the background contribution B.0 mm).WT 0. i. as well as the peak velocity pressure 0 0.2 By replacing in equation (9) the expressions of cs and cd 0.1.rough 0.smooth chimney reduces the wind force depending on the slenderness. aforementioned differences in the gust load profiles. z (11) 0. Their 0. EURODYN 2014 where cs is the size factor covering the reduction of the load 1 effect due to non-simultaneous gust peaks and cd accounts for 0. the so-called tip z/h 0. accordance between the different models is good. differences in the response are predominantly due to the 1387 .3 distribution along the height applied to the chimney sample 0.6 CICIND increase of CD in the tip region of a chimney. In fact. In particular.6 0.5 is ≤ 1.3 By separating both components.4 product is named structural factor and zs is the reference 0. Figure 1. it decreases as the loaded areas become larger. The flow over the free end of the 0. It is compared to the distribution 0 0 100000 200000 300000 400000 500000 600000 obtained by applying the CICIND Model Code and the Wind-induced bending moments [kNm] Eurocode model to both a smooth concrete chimney surface (ks = 0.1 0. For measured in the wind tunnel (EN-WT) and to those provided circular cross sections.smooth 0. In-line wind loading on the chimney sample n. and the aspect and 4) CICIND Model Code. application of the 2) wz = ∙G∙ ∙ C ∙ A+. The values along the height may differ. The label "EN. The aforementioned superimposition in different loading models and to the different distribution of equation (11) of the mean load and the triangular gust load aerodynamic coefficients.2 n. the factor is particularly sensitive to by the Code itself for 2) smooth or 3) rough cylinder surface the Reynolds number.2 0.4 0.5 effect. with either smooth or rough surface) Other formulations are for example proposed in the 1 CICIND model Code for industrial chimneys ([3].4 0. 0. The uniform distribution of CD resulting from wind tunnel tests.5 wz = w z + w. different loading models Figure 2 shows the differences in the in-line wind loading profile due to different gust wind load models (Eurocode and Figure 3 shows the resulting bending moment on the sample CICIND load models. The size factor cs 0. it is close to 1 for stiff 0. 0.7 EN . close to 1. It clearly shows that CD the first term.rough however it locally increases in the tip region.7 0. Aerodynamic force coefficients on the chimney 1 + 7I z ρ& V  z sample n. EN . according to different loading models and CD distributions: 1) Further differences in the code stipulations concern the Eurocode model applied to aerodynamic force coefficients aerodynamic shape factor CD (termed cf in the Eurocode). chimney n.1 bending response due to gustiness.WT the aerodynamic coefficient. as derived by wind tunnel tests on a circular cylinder with 0. according to Figure 1.e.- (10) 1 + 7I z  2 CICIND Model Code and 3) Eurocode (EN.9 EN .9 1 in equation (8).

00 and out-of-phase with the motion. Chimney sample n. A possible -1000 modelling of the non-linear behaviour of the aeroelastic lift force is proposed by Marris [9]. Then.20 and the self-limiting amplitude on the rms of the response aL 0. wind forces are applicable. side and pressure distributions are needed for the design The typical feature of the Vickery&Basu model (which is calculations.60 damping as function of the variance of the response and two z/h 0. Meridional membrane stress n22 [kN/m] at different at a given velocity and acceleration undergoes large angular z/h levels. vibrations. therefore the model is and high ring stiffness of the wall imply a beam-like suitable for typical ranges of turbulence intensity. the non-linearity severely complicates the 0. With thin method is also applicable to non-cantilevered structures. 3 DESIGN FOR VORTEX RESONANCE Even though the tip-effect is not included in the CICIND Two different models are available regarding the design for model.1 is taken as example). The model parameters are deformability of the circular cross-section. as the effective correlation length depends depends on the slenderness of the chimney and the on the amplitude of vibrations. concluded that.70 and removes the non-linearity by expressing the aerodynamic 0.40 beam independent model parameters. 0 30 60 90 120 150 180 shell z=0. this is compensated by the relatively high and constant vortex resonance. about 10% of reduction of stresses vortex resonance as a forced vibration and to incorporate the might be achieved through a more realistic distribution of aero-elastic effects in model parameters derived from aerodynamic coefficient. behaviour with cosine stress distribution along the Furthermore.80 computation. especially high at the bottom.Proceedings of the 9th International Conference on Structural Dynamics. Which way is appropriate iterative procedure. 4π a. The departure is coefficients.10 (Vickery&Basu.00 in the most general way as a force with components in-phase shell z=0.15 often of negligible importance. regarding the along wind action. In the second model. the model is non-linear. the wind loading may be represented Ruscheweyh's approach is based on the concept of effective by a wind force or by wind pressures non-uniformly correlation length and accounts for aeroelastic forces by an distributed over the surface. experimental data.90 However.d 1388 . e. Such a mathematical model assures that the system 1. but it extends over a relatively On the basis of these and other results (see [8]). Vickery revisits Marris's approach 0. The derivation is based on the -1500 φ [°] analogy with the Magnus effect for a rotating cylinder. This lies the basis to develop the cross-flow force experienced by the cylinder due to the angular vibration of the Max meridional membrane stress n22 wake. the CICIND model code results in a simplified but conservative design.1 of the response in resonant conditions. One is Ruscheweyh's approach to deal with force coefficient.30 0 shell z=0.45 component acts as a negative aerodynamic damping. compared to the m. resulting from wind tunnel tests. EURODYN 2014 different values and distributions of the aerodynamic beam distribution. In the chimney design.30 aerodynamic damping for small amplitudes of oscillations Ka0 0. The negative aerodynamic 0.g. beam distribution at z = 0.2 Structural behaviour: beam versus shell resonance is based on Vickery & Basu's model of a self- limiting response process. the stresses due to wind concentrate at the windward Method 1. may even exceed 30%. the spectral formulation of the lift force regards only the force on Figure 5. Meridional membrane stress n22 compared to beam the non-vibrating cylinder and allows calculating the variance distribution on the chimney sample n. wall and low aspect ratio the structure behaves as a shell. while the out-of-phase shell z=0. 0. as the load depends on the mode shape.60 As vortex excitation is characterized by the existence of a -500 shell z=0. High slenderness derived from experimental data. The approach 1500 goes back to Scruton [2].00 damping shall be able to account for the aeroelastic effects 0 500 1000 1500 2000 due to the fluid-structure interaction by reducing the total n22 [kN/m] damping of the structure (see equation (12)). the wake of a cylinder which is moving across a uniform flow Figure 4. who expressed the aeroelastic force 1000 beam z=0. In fact. Ruscheweyh's approach is included in the Eurocode as Then. the deviation in the maximum membrane stress. [11]). then.50 0.1. the design for vortex 2. the circumference. Therefore. namely the negative shell 0. it can be high portion of the cylinder height (Figure 5). However. included in its full and even extended formulation in the CICIND and in its simplified version in the Eurocode as 2000 Method 2) is to include the aeroelastic effects into the modelling of a negative aerodynamic damping. [10]. ρ& d Sc σ:  Figure 4 shows that in case of low aspect ratio and/or low the ξ= 1 − K &7 81 − 9 < => (12) thickness-to-radius (the chimney n.00 solution involves a limit cycle. The in-phase component is n22 [kN/m] 500 shell z=0.75 self-limiting cycle.

the effect of turbulence is mainly attributed to the The Scruton number (Sc = 4πmξs/(ρD2)) governs the large scales. may considerably reduce the response. particular. formulation. In any case.20 υ υ St ∙ υ 19.08 the modeling. in 0. the effect of turbulence chimney sample n. This model is unique in the sense it intensity on the response is to reduce the critical Scruton includes the effect of turbulence and the dependency of the number at which the jump between high and low levels of response on V/Vcr.10 The core of the Vickery&Basu model is then the modeling of the negative aerodynamic damping. shown example.50 at the CICIND recommendation. EURODYN 2014 An important parameter.1 0. The latter may lie in the lock-in wind velocity and therefore have a pronounced effect on range (high level of amplitude). the limiting value structural damping of the chimney. the Strouhal law is calculation is performed on the sample chimney n.875 0.30 ∙ d fC ∙ d 12. which influences the size of the threshold is achieved at smaller Sc in case of high Iv.350 ymax(h)/d in: 9.2. As a general rule of thumb.2 0.60 number Sc to the aerodynamic damping 4πKa. as the Eurocode Method 1 (after Ruscheweyh) is empirical and 1389 . Cross-wind deflections as a function of turbulent if it assumed St = 0. Within this further investigated in this paper.15 0. Figure 8 is obtained by applying the CICIND model to the According to this formulation. This issue. it can be estimated that Re > 1. In smooth flow. In the limiting amplitude of the response.40 6. in the transitional range or in vortex-induced vibrations [12]. with regard to the Vickery&Basu model.20 0. the higher is the ratio Sc/4πKa. is not formulation as described in the CICIND code. the amplitude of oscillation is dictated by the ratio of the Scruton 0. Proceedings of the 9th International Conference on Structural Dynamics. By looking 0. which are interpreted as a slowly varying mean amplitude of the response.00 fC ∙ d 15 0 0.400 0. at a given level of CICIND Model Code for steel chimneys ([3]. However.2.270 5*105 would be advisable. the total damping of the ymax(h)/d structure will never become negative due to the existence. the effect of Iv reduction of such a value for higher modes of vibrations. the proneness of a chimney to experience Figure 6.30 0 0.50 lies in the lock-in range when Sc << 4πKa.60 order to reduce the sensitivity of the chimney to vortex resonance. formulation. however. the Scruton number mean wind velocity and the aspect ratio. This including the effect of turbulence intensity.24 >> 4πKa.050 25. Little information is 0.13 Small amplitudes of vibrations are instead predicted when Sc 0. In fact. the turbulence intensity. that both methods turn to be correct if properly interpreted by large amplitudes of vibrations occur when Sc/4πKa < 1. The Similarly.10 The kinematic viscosity of air ν is equal to 15*10-6 m2/s.700 V@+ ∙ d (13) Re = = St = > 5 ∙ 10H 15. is the mode shape. Mathematically. Besides that.5 (15) Figure 8 shows the extension of the lock-in range at a given Figure 6 shows the response curves as function of the Sc (for the chimney sample n. the Scruton number is varied by varying the According to the Vickery&Basu approach.2) and its dependency on the Scruton number at different levels of turbulence intensity. 0.451 therefore: 0. Physically. the lower is the negative aerodynamic very different results (Figure 9). high turbulence the forced-vibration range (low level of amplitude). Cross-wind deflections as a function of the Scruton vortex induced vibrations is primarily led by its Reynolds number Sc at different turbulent intensities Iv number at the critical velocity. it results: intensity Iv at different Scruton numbers Sc fC ∙ d > 1. The calculation is still which depends only on the natural frequency of vibration and performed by applying the extended Vickery&Basu not on the shape of the mode. applied to the same chimney sample. The reason is that the higher is the and Method 2). This requirement can be translated 0. The Eurocode design methods (Method 1 oscillation occurs. is calibrated for the first mode of the cantilever structure. as described in the CICIND Commentaries [5].525 fC ∙ d 0.05 0. [5]) proposes a mechanical damping (thus Scruton number).00 available on that and the aerodynamic damping at least 0 10 20 30 40 50 depends on the Reynolds number. In the lock-in range. The response 0. 0. In intensity also increases the bandwidth of the lift spectrum. 44. it means that the mechanical damping is much lower than the 0. The turbulence intensity. of the self-limiting cycle (see equation (12)).25 > (14) Iv(h) St 2 Figure 7. the figure proves damping. provide turbulence intensity.2 by violated and vortex separation is tuned to the structural applying the Vickery&Basu model in its full (an extended) vibration frequency for a certain interval of wind velocities. the low range of Re should be avoided. Figure 7 shows that.40 aerodynamic damping. in 0.

UK. which includes aeroelastic 0.6 0. The response of slender line-like structures to a gusty 0. Sc ≈ 4πKa.8 1. towers and 0. Mech.1 1. In case of a line-like two methods differ considerably even at relatively high values structure. 26-28 June 1963. able to amplify the mean wind load vibration phenomena. On the wind-excited oscillations of stacks.2 1. The core 0.60 of the Vickery&Basu-based model codes is the modelling of the negative aerodynamic damping. a and tensile stresses at the windward side become the leading representative parameter.. the CICIND 0. Scruton. The scale of turbulence alone is not.60 investigation of the model independent variables (like self- EN .88 application in the Standard Codes .15 modelling is unique in the sense that it includes the effect of 0.00 aerodynamic damping depends at least on the Reynolds number. Extension of the lock-in range as a function of Iv relatively low Scruton numbers. Part A: The Shell. 2011. the parameter to be investigated. Davenport. In general.40 CICIND the intensity and scale of turbulence. 389.Method 2 limiting amplitude of vibration and aerodynamic damping) 0. 0. the latter related to the cylinder diameter).25 aerodynamic damping and increases the spectral bandwidth of 0. the scale of turbulence would be another of the cross section might become significant.9 1.4 1. the mean wind velocity and y(h)/d 0.88 results. Civ. Proc.3 1. University of Florence / Technical University Braunschweig.10 the aspect ratio. the model parameters as well as the influence of Gust wind load models to calculate the along wind response turbulence intensity should be carefully investigated.00 the lift force.. Phd Dissertation. Wind effects on Building and 0 0. The two approaches. Depending on the aspect ratio. 1964. [8] H. However. because they destroy The second issue which is addressed by the paper concerns the vortex cells. EURODYN 2014 calibrated on the basis of experimental data. [5] CICIND Commentaries for Steel Chimney Code. theoretically correct and consistent. the coherence along the chimney axis.8 0. design models for vortex resonance. As a result. further research would be advisable in the 0. Niemann. Marris.g. 2013. periodic wakes and induced is a structural parameter.30 REFERENCES 0. 1390 .9 2. Journal Basic Eng. Soc. The latter [9] A. Comparison of model codes for vortex resonance [4] CICIND Model Code for Concrete Chimneys. Therefore.Method 1 and of the parameters they depend on (like the mode shape. as the are applicable to line-like structures. Eds.6. Eng. Proc. ovalization Furthermore.20 turbulence. F. of Symposium no. Trans.Proceedings of the 9th International Conference on Structural Dynamics.J. Figure 9. the design for vortex resonance. A review of vortex streets. in the case of a chimney.05 0.0 Eurocode reveals that the full formulation in the CICIND V/Vcr model is.20 [1] A. pp.1 0. The effect of non-uniform for vortex resonance are in principle consistent and physically spanwise distribution of aerodynamic coefficient is also correct.05 0. The CICIND Chimney Standard. the Davenport approach of the gust response factor.are considered. [6] CICIND Commentaries for Model Code for Concrete Chimnes. the small scales have an influence especially on wind pressures and not forces have to be provided.2 0. 23. They are applied to a represented overestimated. it should be and define an equivalent static load which reproduces the reliable at higher values of turbulence intensity. however. which only provides results for zero those of the Eurocode and the CICIND) are revisited versions turbulence intensity and for this reason the response is often of the Davenport approach. wind speed. after Ruscheweyh and after Vickery&Basu . CICIND Report. vol.40 0. However. The figure proves that the design load models real world sample chimney.00 masts. whereas the effect of large 30% the corresponding stresses calculated by using a beam turbulence scales is usually interpreted as a slowly varying model. Shaker Verlag. Figure 7 shows the impressively high required. However.50 EN . 2011. 4 CONCLUSIONS AND FUTURE OUTLOOKS [7] F. Little information is available on that and the 0. they even excess of of the chimney. The comparison of results between CICIND and 0. When a shell model of the chimney has to be used.6 1. A new aerodynamic phenomenon and its effects on the The paper investigates the modeling of vibrations of design of ultra-high cylindrical towers..7 0.as well their Sc = 15. at low values of Sc.25 Structures. in principle. The opposite effect of critical design stresses (e. Engineers. chimneys under the wind action. 2010. The design load models (such as Vickery&Basu model). in addition to the intensity of beam-like model might be neither realistic nor on the safe side turbulence. aspect ratio and ratio of wall thickness-to-radius. International Standardizazion of Wind Actions All design load models for the gust wind load are based on on Chimeys.10 wind. Teddington. as well as in the range investigated. Lupi. Soc. 185-193.0 1. High turbulence intensity reduces the negative 0. as it must be related to the diameter stresses. 1962. depending on the joint combination of amplitudes of vibrations at small values of Sc.7 1. often provide very different Sc = 15. especially at Figure 8. 2011. 86.10 0.15 0.30 0. holds for the Eurocode Method 2 (a simplified version of the the base bending moment). are of turbulence intensity.5 1.. [2] C. Lupi. ymax(h)/d 0.W. force coefficients and not pressure coefficients. the turbulence intensity. Iv [3] CICIND Model Code for Steel Chimneys. Among Standard Codes. Am.50 effects.G.20 0.

Across-wind vibrations of structures of circular cross-section. Part II: Development of a mathematical model for two- dimensional conditions. 1983. R. Basu. 15. 49–73. [11] B.I. Vickery. 1998. Across-wind vibrations of structures of circular cross-section. Basu. 75–97. Vol. 12.O. Hansen. Journal of Wind Engineering and Industrial Aerodynamics. 15-23. Proceedings of the 9th International Conference on Structural Dynamics. [12] S.J. 1391 .I. 12. Vickery. Vol. 1983. Journal of Wind Engineering and Industrial Aerodynamics. Part 1: Development of a mathematical model for full- scale application. Vortex-induced vibrations of line-like structures. Vol. EURODYN 2014 [10] B. No 1. CICIND Report.J. R.