46 CIGRE Session - CIGRE 2016, Paris (France). 21-26 August 2016
Summary:
Overhead lines may suffer a myriad of different loads, but particularly when the stability of compact lines with pivoted insulated cross arms is studied (Paper B2-106), the wind gust modeling becomes crucial. Generally speaking, cable structures present a high non-linear behavior, so the effects of wind gusts on them may trigger much more dangerous phenomena than those deeply studied in the others. The instability of the abovementioned vees in power transmission lines or the dewirement in catenary lines for railway systems.
Regarding the codes and standards which tackle the design criteria, both for overhead electric lines (EN 50341) and general structures, the most widespread wind loading criteria are based on static calculations. These static prescriptive loading are usually based on experience in such way that they include different factors which try to reproduce the real conditions of the wind gusts. For instance, the wind velocity is defined according to the height of the cables or the roughness of the terrain, while the wind load on them depends on certain factors stated by the span length, the nature of the gusts, etc.
Due to the increase of computational strength, performance based design criteria are being recently included in codes and standards (Eurocode 1). Thus, dynamic calculations for wind loads could give more realistic results in certain complex structures, specially those mainly composed by cables. Regressive models or power spectral density (PSD) based models are nowadays used for the synthetic generation of stochastic wind series which are included in computational dynamic simulations. Davenport [1], Kaimal, von Karman or Harris are in well known in this topic, allowing to generate a random component of the velocity of the wind that must be added to the mean value, which depends on the location, the terrain, etc.
Nonetheless, the stochastic nature of wind gusts should reflect its turbulence and intermittence in the proper time scale. That is revealed as a key factor when cable systems are tackled due to their mass and stiffness are completely different than other civil structures. For that reason, traditional PSD based models may be under discussion in wind gust modeling. At large scales those models tend to follow a normal distribution, so they would be accurate enough to analyze a dynamic structural behavior regarding time periods of minutes or even hours when other topics like wind power generation is forecasted. However, a period of seconds, which would be the real nature of the wind gusts that can provoke the oscillation of a cable, would not follow such normal distribution.
Stochastic models based on Lévy processes can reproduce the intermittency observed in real wind at different time scales, which involves several stochastic processes and parameters that are not related to any physical characteristics, so their determination requires an adjustment by means of real data. Some preliminary results show that wind behaves following a Normal Inverse Gaussian distribution at small time scales, whereas it approaches a Gaussian distribution when time scale increases.
The first conclusion of this contribution is that dynamic simulation of wind loads on overhead lines give more realistic results due to the special nature of cable structures. So, the estimation of the efforts transmitted to the vees and towers will be pretty different. The second conclusion is that depending on the time scale, different stochastic models must be tackled, particularly traditional PSD based models should be discarded at small time scales in benefit the of Lévy processes.
Publication date: 2016-08-21.
Citation:
P. Rodríguez, J.C. Polo, L.F. Alvarado, A. Carnicero, J.R. Jiménez-Octavio, C. Sanchez-Rebollo, Wind gust modeling, 46 CIGRE Session - CIGRE 2016, Paris (France). 21-26 August 2016.
