You are on page 1of 7
34C-TPC-14 Assessment of Safe Design Tension with Regard to Aeolian Vibrations of Single Overhead Conductors A. Leblond, yo Québue- REQ ‘arenes, Quebec, Canada ‘Absuact: This paper presents 2 probabilistic model for predicting Aeolian vibration response of single overhead conductors. Two diferent approaches are used to asses safe design tensions. One is based on the conductor fatigue endurance lint and the other is based on linear fatigue ‘damage accumulation associated to each vibration eycle which leads fo a cern predicted fatigue le of the conductor. Recently, CIGRE Task Force 22.11.04 made a large use ofthe calculations made with this model in working out its fina ‘recommendations of tafe design tensions, Keywords : Acoian Vibration, Single Overhead Conductor, Safe Design Tension. L INTRODUCTION Overhead conductors are often subjected 10 severe Acolian vibrations, These vibrations arise from alternating forees caused by Karman vortices. They are self-limiting so that the conductor's maximum displacement amplitude does not exceed one conductor diameter. However, such small amplitudes can damage the conductor through wie fatigue breaks at suspension clamps. Therefore, itis important that these vibrations bbe predicted and mitigated atthe design stage, in order to avoid high costs related to future service interruption ‘and repairs. In order to achieve safe vibration Conditions, one needs to rely on proper aeroelastic and structural modelling of the system. This paper first presents a probabilistic model for predicting Aeolian vibration response of single overhead conductors. This ‘model is then used to assess safe design tensions. Il GENERAL CONSIDERATIONS. 1c is well known that standed conductors get more vulnerable to Aeolian vibrations as tension i increased. ‘This is te whether or ot conductors ate protected by means of dampers. Hence, there is a nced to set an ‘upper limit to conductor unloaded tension that prevails for a significant period of ine {Guidance to such safe design tension with respect 10 ‘Acolan vibrations had been provided en 1962 by the so- called EDS Panel (1) working under assignment of CIGRE Swdy Commitee 6 (since disbanded), However, aout 30 years later, CIGRE Working Group (0-7809-6625-5/00$10.00 ©2000 IEEE. C. Hardy ‘CandeHardy neasona SeBruno, Québec, Canada 22.11 felt timely to revise the Panel’s recommendations and appointed a new Task Force in this regard. ‘This new Task Force, CIGRE TF 22.11.04, soon recognized that previous attempts to determine safe tensions on the basis of field experience only had been difficult and most ofthe time unsuccessful. This is due to the fact that actual conductor condition, terrain features, tension history and even, on occasions, design tensions are not known, Consequently, the Task Force decided to have recourse mainly to modelling to achieve his task. One important advantage of this approach is the possibility to consider all relevant parameters and variables in a systematic manner. ‘The first step in the Task Force's work [2] was to efine a tension parameter which would unify the results fo the largest possible extent. Parameters such as average stress, nominal stress in the aluminum strands and percent of conductor rated strength were considered. The parameter that was finally selected was the ratio ‘Hiw of horizontal tasile load over the conductor weight per unit length. Incidentally, it may be shown that F/w is approximately proportional to stress in the aluminum strands for ACSR conductors of any stzel content. The Task Force made a large use of the calculations made with the present model in working out its final recommendations of safe design tensions. Il, AEOLIAN VIBRATIONS: RANDOM PROCESS: Most ofthe models developed so far [3] have all been deterministic, assuming the vibrations to be sinusoidal In contrast, the present model takes into account the random character of Acolian vibrations by considering a given distribution of peak values of vibrations Acotian vibration recordings generally show a slowly varying wave form characterized by an amplitude- ‘modulated sinusoid. The related power spectrum is narrow-banded and involves some frequency ‘components ‘The concept of a narrow-band Gaussian random process was introduced by Rice [4] who originally used iv in modelling random-noise signals. Its applications to ‘other phenomena such as ocean waves and metal fatigue soon followed. ‘The requirement that the spectrum of the signal be narrow-banded appears as the central idea of the concept [5] which is used below to model the 202 Acolian vibration response of single overhead. conductors. ‘According to Rice's concept, the conductor displacement (¢) ata given point may be represented by arandomly modulated harmonic with a random phase ange: W)= R(eoslo,t +00)), where @, is the central frequency, R(t) is the envelope, O(1) isthe phase angle and 1 is time. R(t) and ) are assumed to vary slowly compared to y(?) Rice demonstrated that envelope R(t) has a Rayleigh Alstribution as shown in Fig. | and 6(¢) has a wniform istibusion between 0 and 2x Normalized envelope (Fij,) Fig. Rehigh dstbution of he nlp. ‘The Rayleigh distribution is defined by the density function This densiy function depends only on the ratio 4 and aplis to all narow fequeney bands inmespective of the central frequency ©. Thus, Knowledge of the rms value of the signal Jpg is sufficient to determine the distribution of vibration amplitudes in any frequency band, Comparisons {5} to experimental distributions have shown that the Rayleigh distribution is a fair representation of all distributions in any frequency band, which supports the initial assumption that Aeolian vibrations can be modelled by a narrow-frequency-band of random vibrations IV, CONDUCTOR SELF-DAMPING Energy dissipation in overhead conductors is mainly related 10 fictional damping due to small relative movements between adjacent wites. Knowledge of conductor self damping is important in predicting the steady state response of conductors. Noiseux (6), using mainly a mathematical approach, was the first (0 propose similarity Iaws for internal damping of stranded cables in transverse vibrations. ‘These laws were derived from the basic cable differential equation. Strictly speaking, they may be applied to similar cables only. Leblond and Hardy [7] developed extended similarity laws, based on the pioneering work of Noiseux, applicable to all multi- layer aluminum conductors.” For multilayer (round wires) all-aluminum-conductors (AAC) and aluminam- conductorsstel-reinforced (ACSR), they found: e128 yt pe =k (022) mo @) may where Map is conductor fee-feld loss factor, & is @ factor of proportionality, EI is conductor flexural sigidty, fis frequency, Yq, is rms valve of antinode isplacement, mis conductor mass per unit length and 4s is. nominal static longitudinal stress in the aluminum wires. Equation (3) includes the effect of tarrow-frequeney-band random vibration on the power dissipated. This effect has been analyzed in (5). It consists in multiplying the tigp factor which results from self-damping measurements (sinusoidal vibrations) by the factor A(Q.22) where T() is the gamma function. \V. WIND ENERGY INPUT. The Acolian exciton is wl (3 represented by the Acolian power per unit kegih #7 introduced in the span. This power is nome onthe form ¥, fad |, 4) { 5) ® ‘which becomes a function ofthe antinode amplitude ¥, ‘divided by the diameter D. This function isa very steep function of ¥,/D. Another way is t introduce an Acolian excitation factor defined by: P, -2. a nese ® where B is the kinetic energy of transverse vibration in the span. The first advantage of this representation is that 1, 1s nota steep function of Y,/D, while the second 203 and more important advantage is that 1, can be {formulated for random Aeolian vibrations as shown in [8]. This excitation factor is normalized into a reduced ‘excitation factor n,, given by: 2m. ne =n, © : pqD’ where py is sir density. Rawiin [9] made wind tanel measurements to determine the wind power input of a fleribe elinder vibrating a its fondamentl frequency, cloeely sinulating a balfsimscid of « sting wave “The wide range and high quality of these data allow for certain transformations such a the introduction of wind turbulence andthe random character of Acolian vibrations : a narow-frequeney-band random proces. “The reduced excitation factor my based on these transformations is presented in (8) 38 a function ofthe rommalized displacement Yiyq/D and of the wind intensity of turbulence Zo, In fact, Ny, becomes a farily of curves, a function Y,y/D, one for each valu of fy The Energy Balance Principle then allows the sk aninode vibration displacement Yn to be determined by Balancing the wind energy inp (S) othe conductor self-damping (3). ‘VL WIND EXPOSURE CHARACTERISTICS Line exposure to winds is determined, in a quantitative manner, by the effective wind frequency of occurrence defined below. Aeolian vibrations of conductors are assumed 10 be sensitive 10 the perpendicular component of wind only. Hence, the ‘effective wind frequency of occurrence is related to this, ‘component only ‘The effective wind frequency of occurrence is based here on the annual wind rose from a nearby meteorological station. It will be recalled that wind roses provide the average annual wind velocity as well asthe annual frequeney of occurrence in either one of 16 cardinal and collateral points. Its then simply a matter of projecting that information onto the perpendicular to the projected line to finally obtain the “annual quantity” ‘of exposure as a function of effective wind velocity. ‘This ‘of exposure represents the effective wind frequency of occurrence. Let F; be the annual percentage frequency of winds ‘lowing from the ith direction, which is inclined (-1yu8 radians from the North; V;, the average wind velocity in this direction and ¢, the angle ofthe Tine with respect to North. Then, assuming wind velocity from each cardinal point to be distributed according to a Rayleigh distribution oly bv, o with 7 ® where p(V.) AV is the probability of occunence of a ‘wind velocity Jj among all winds blowing from the ith cardinal point. These statistics apply at measurement Iheight iy at the meteorological station assumed to be setin open terrain. They may be transformed to apply at conductor height /, on site by means of the well-known power law o where Vi. is the wind velocity at conductor height bowing fom the th cardinal point. Both exponent and wind intensity oftrbulence Ty. depend on tern topography. Strictly speaking, (9) is valid only if the terain onthe fine sites equivalent tthe terrain at the reteoologicsl station, But since only. low-wvind- ‘locity winds ae of importance in the present context, 4 more rigorous transformation is not seen as being necesary. Taking ito account the percentage F, ofall ssnds blowing from the ith drction end projecting on the pempendiclart in ais yields: Pn Vie with a {te Toe “Vie ng where p(Vac) Tg is the probabiliy of occurence of a wind from the ith cardinal point having a component velocity of intensity Tye perpendicular tothe line, at conductor height, among all winds blowing in all directions. Taking into consideration all sixteen wind rose pints, one finds: ia a2) io(o-t-08 1sisi6 (a) 204 where (Fac) aes the probability of ocumence of a wind having component velocity perpendicular to the line of inosity Tae, st conductor height. This probability coresponds tothe effective wind fequency of oceurene. ‘As calelations of Aeolian vibration response are usually cared out in terms of vibration frequency f tis more convenient to work with p(f)d/, which designates the probability of occurence ofw vibration having a central frequency. The later probability may ‘be obtained from the former (12) merely from a change in the independent variable provided by the Stouhal relationship: vn asst, 1 a ay Af the wind equncy of ecuence is asuned 1 be the sue enc Sucre tad oly te veal ge wind velocity Vand wind from of occurence F tie known, the probly (13) may be defined in « Citeret bt eh mote epost iW, i) with 7 replacing 7) in (13). VII, DYNAMIC STRESS IN THE CONDUCTOR ‘Once the rms value of antinode displacement Ym is calculated using the Energy Balance Principle (8), the nominal dynamic bending stress in the conductor at suspension clamps can be determined using the Poffenberger-Swart formula [3], based on the assumption that each individual wire flexes independently of the others. Based on the latter, the alternating bending stress in the conductor ata clamped extremity is Some RA EG eT FYomes CS ay as) Where Tyme is the rms value of nominal dynamic bending stress, d is outer-layer wire diameter, Ey is ‘Young's modulus for the oute-layer wire material and lpn is the minimum theoretical value of conductor flexural rigidity. Equation (15) was developed for a clamped extremity. However, when the two adjacent spans are subjected to Aeolian vibrations, there is a rocking action of the suspension clamp. Experimental evidence shows that the dynamic bending stress is almost similar on both sides ofthe suspension clamp and its value is approximated by: a6) Where Gigs ANd samy ae the nominal dynamic bending stresses calculated with (15) in spans 1 and 2, respectively, on each side of the suspension clamp. ‘When the two adjacent spans vibrate atthe same level (16) shows that the bending stress at the suspension clamp isthe same as in (15). VIII, ASSESSMENT OF VIBRATION SEVERITY ‘Assessment of vibration severity may be investigated by comparing calculated bending stresses with acceptable levels. For this purpose, there are essentially two ways to assess conductor vibration severity. Each of these ways is presented in the following sections. |A. Endurance Limit Approach ‘The first and easiest way to assess conductor vibration severity is to compare the maximum predicted ‘bending stress amplitude with the fatigue endurance limit of the conductor at the suspension clamp. Vibration levels are regarded as completely sae if they do not exceed the so-called conductor endurance limit. For vibration levels below the endurance limit, an infinite conductor fetme is thus inferred. FFor example, the product of the maximum antinode amplitude Y,q0- by frequency fis shown in Fig. 2as a function of frequency for a given ACSR conductor at four different mechanical tensions. Fig. 2 also shows the fatigue endurance limit of the conductor (22 MPa [BD expressed in terms of ¥, na. by means of (15). riots minade seve Vnn byecvene) 8) “t Frequency 2) Fig. 2 Product of maximum anode ampli by frequen as & faction of reqency oar differen tension eel 205 ‘The vibrations at a mechanical tension of 109% RTS are seen to be safe since the maximum response of the conductor stands below its fatigue endurance limit. However, at a tension of 12% RTS or higher, the vibration’ level exceeds the endurance limit over a limited frequency range and the conductor may then experience fatigue damage during its projected life. For such cases, the Cumulative Damage Approach may be ‘more appropriate for assessing vibration severity. B, Cumulative Damage Approach ‘The second way to assess conductor vibration severity is to calculate the anticipated lifetime of the conductor. Each vibration cycle is assumed to cause a tiny fraction ‘of fatigue damage that cumulates linearly during the lifetime of the conductor, until, ultimately, one strand fatigue breakage occurs. This approach requires S-N fatigue endurance curves for the different classes of ‘conductors 0 be dealt with as well as certain wind Statistics. From the section entitled “Fatigue performance relative 10 f pac” in (3}, conductor fatigue curves, or $-N-curves, have been produced on a 95% survival basis [10]. To take into account the ‘damage accumulated at different stress levels during the line service life, a linear damage accumulation hypothesis (Miner's rule) is used, The total damage D, ‘of the conductor in one year of service is therefore expressed as: p=. N, an where m, is the number of eyeles expected at a stress level oy during one year of service and N, is the number of cycles to failure at this level. According 10 this rule, D,=1 at failure. The number of cycles 1n(f) at frequency fin the stress band day, centered at 65 i MN=STN) Hen)don, 8) where T(f) isthe vibration duration at frequency fin secondsiyear and p(64,) doy, is the probability of cccurence of a bending stress equal to Gy; which is assumed to be distributed according to a Rayleigh Aistibution. Vibration duration T(') at frequency fis given by: T(f)=31.S10° PL. (setyean) where p(f) Af is defined in section VI. The total damage D,(f) at frequency fis: a9) doy. 20) where (0) is the number of cycles up to failure at a siess level 0, at Gye Hs the maxima valve of bending stress predicted at frequency f- Taking into account all the frequency bands Af, the total damage D, of one yeas: pe=aisnoD sei f + (os), Beihi0,. ‘The total damage Dy is very sensitive to stress level I can be shown that D,, is proportional to For en L(F¥sm © where C is defined in (10) example, Dz is proportional to ¥28, and wo f** for both ACSR_and AAC conductors. The anticipated conductor lifetime expressed in years is defined as the reciprocal of the total damage D, IX. ASSESSMENT OF SAFE DESIGN TENSION ‘The Endurance Limit Approach may be used not only for assessing vibration severity but also to assess safe design tension. To this end, the maxima of the curves in Fig. 2, identified by points A, B, C and D, may be ploted as a function of mechanical tension. The resulting curve joining points A, B, C and D isa straight line. ‘Safe design tension arses from the intersection of this curve with the fatigue endurance limit of the conductor : 11.1% RTS or (Hi) = 811m. ‘The Cumulative Damage Approach may also be used to calculate safe design tension. Fig. 3 shows the projected lifetime of the conductor as a function of the ‘mechanical tension, For example, safe design tension comesponding to a projected lifetime of 50 years is 13.2% RTS or (HW)uyq = 961 m which is higher than the previous value calculated with the Endurance Limit ‘Approach. This makes sense on account of the former approach allowing for a certain number of ‘vibration cycles above the conductor endurance limit while the Tater does not. “Hydro-Québec has been one ofthe four organizations, ‘worldwide, that contributed to the calculation of safe design tensions for CIGRE TF 22.11.04, Using the model described above, the utility actually offered contributions which were developed according to either fone oF the {wo above-mentioned approaches to assess the severity ofthe computed vibrations regimes. 206 Proecteietine (eas) opp ee cana tention ORS) : Fig 3 Projected ite a oneson of mechanical enon ‘The safe design tensions caleulated according to both ‘approaches are shown in Figs. 4 and S as a function of ‘wind intensity of turbulence. Regarding the Cumulative Damage Approach, a Rayleigh distribution for the component of wind speed perpendicular 10 the conductors was used, with a mean or 4.5 km/h. The 98% sure predicted lifetime was set at 50 years. These plots apply to multilayered ACSR, AC, AAC (6201) ‘and ACAR conductors. Also shown in the figures are the tensions recommended by the EDS Panel [1], wanslated in terms of Hw, for different conductor classes. As the effect of turbulence was not taken into account by the EDS Panel, their recommended tensions appear as flat horizontal lines. The safe tensions for different conductor sizes and types, as calculated with the Hydro-Quebec’s model, appear as narrow bands stepping up as intensity of turbulence increases, reflecting the fact that conductors are_less susceptible to Acolian vibrations when subjected to more turbulent winds. As noted above, the ‘Cumulative Damage Approach leads to more permissive ‘iw than the Endurance Limit Approach, The predicted safe Hiw for AAC (6201) conductors are seen to be from 30 to 50% lower than that of AAC ‘or ACSR conductors. This is the result of calculations performed on the basis of lower fatigue endurance limit for the AAAC conductors, according to [3] ‘may also be noted that the predicted safe tensions in terms of Hiw according to the Cumulative Damage Approach stand in the same range as the values recommended by the EDS Panel for AAC and low steel ‘content ACSR's. However, for high steel content ACSR’s and particularly for AAAC (6201), the calculated safe Hw is much more conservative. 1600 1200 siontunit weight (mn) Horizontal Wind intensity of tbulonce (%) Fig. 4 Pode safe design tension scoring to Endurance Limit Approach for mul-ajer ACSR, AAC. AAAC201 aad ACAR ‘cdacors. 200 Hoszontaltension/unt weight (n) 10 2 0 Wing intensity of turbulence (%) gS, Predicted safe din tension scoring n Cumltive Damage Approach for mut-ayer ACSR, AAC, AAAC-S201 and ACAR ‘The above predicted safe tension in terms of Hw for undamped and unarmored conductors together with similar predictions from three other organizations were finally assessed by the Task Force against all available field experience and adjustments were made accordingly. The final result are shown in Table 1 [2] ‘where the recommended safe H/w is provided for four types of conventional, round strand conductors (ACSR, ‘AAC, ACAR and AAAC) as a function of terrain characteristics. As for the Hydro-Quebec model, the following intensity of turbulence was assumed for the different terain categories: cat: 8%; cat2: 15%; ceat.3: 22% and card: 30%, Tis important to note that tension Hin the table refers (o initial horizontal tension, before any wind and ice 207 loading and before any creep, atthe average temperature of the coldest month, The same CIGRE Task Force 22.11.08 is now working on the preparation of a guide on safe design tensions for conductors that are protected against ‘Acolian vibrations by means of end-span dampers. ‘Toble 1. CIGRE Task Force 22.11.04 recommendations (2) of safe

You might also like