Modelling the Scattering and Coupling of Small Wind Turbines Located Above BTS Antennas

Transcription

1 Modelling the Scattering and Coupling of Small Wind Turbines Located Above BTS Antennas Tran Vu La 1, François Le Pennec 1, Serge Elenga 1 Lab-STICC, Microwave Department, Telecom Bretagne, Brest, France, Francois.LePennec@Telecom-Bretagne.eu Bâtiment Gérard MEGIE, NHEOLIS, Aix-en-Provence, France, serge.elenga@nheolis.com Abstract The use of hybrid energy (fuel, wind and solar) can contribute to a reduction in both the carbon footprint and fuel consumption of base transceiver stations (BTS). Such solutions of BTSs combining solar panels, fuel generators and small wind turbines (SWT) have been deployed in many regions of the world in recent years. These SWTs typically have power levels ranging from 1.5 to 7.5 kw. Because of the EM impact of megawatt wind turbines (WT) on radars, televisions, GPS, etc., questions have been raised about how locating an SWT in the close vicinity of BTS antennas could degrade transmission quality in radio mobile communications. Radar Cross Section (RCS), Doppler shift and near field coupling of a Representative SWT (RSWT) are analysed to quantify its radio impact on BTS antennas. The results obtained with the RSWT are then compared with those of the 3.5 kw Nheowind-100 realistic model (Nheolis Company) to validate the RSWT design. Corresponding to the worst radio cases, all the models studied are fully metallic. Index Terms hybrid/wind-powered base station, small wind turbine (SWT),, base transceiver station (BTS), radar cross section (RCS), Doppler shift, back scattering, cellular panel antenna, near field coupling. I. INTRODUCTION In the context of increase in energy demand and environmental pollution, wind energy and solar power have become key components in what is called green energy. In practice, a number of big WTs (> 1 MW) or wind farms have been deployed all over the world during the last decade. However, there are also a certain percentage of wind farm installations that have been delayed or removed because of their different effects such as acoustic disturbance, extra mortality of birds, landscape and visual impact, etc. More particularly, EM interference of WTs on radio propagation such as radars, satellites, GPS, TV services, etc. is one of the most important reasons. Although this subject has been studied since the 1970s (on televisions) [1], studies are still ongoing [], because some important aspects like the impact of rotor blade rotation on radar signal processing and solutions to mitigate these effects have not yet been completely resolved. In addition to big WTs, the use of smaller WTs for home use, factories and telecommunication devices has been progressing. Notably, the hybrid model of wind-/solarpowered base stations is a highly satisfactory solution to help reduce fossil fuel consumption when BTSs are too far from the power supply network. This is more significant in the context of LTE-4G network deployment, where compliant base stations are demanding much more energy than older base Fig. 1. Different SWT combinations with BTSs: on a BTS mast (Hue, Vietnam), and separate from a BTS mast (Turkey). stations. According to [3], radio base station power consumption accounts for approximately 00 to 500 GWh per year per operator in some European countries. To figure, Fig. 1 presents two representative scenarios of wind-powered base stations: SWT on a BTS mast (Fig. 1a) and separate from a BTS mast (Fig. 1b). The first model is more compact and more often deployed. Hence, it has been the selected for use in this study. Compared with dimensions of big WTs, those of SWTs are much smaller because their typical power is in the range of kw. However, the EM impact produced by SWT on BTS antennas can still be significant because the distance between them is much closer than that between big WTs and affected radio devices. This proximity raises questions about EM far or near field coupling between SWT and BTS antennas, as well as about possible cellular radio communication disturbances. Moreover, the blade rotation of SWT can produce Doppler shift (frequency deviation) and possibly complicate the demodulation of the received signal. In the context of the OPERA-Net European project, studying and quantifying the SWT impact are the main objectives of this work. The structure of the paper is as follows. Section I introduces BTSs equipped with SWTs, and the possible EM interaction between them. In Section II, we discuss the parameters of RCS and Doppler to analyze the scattering of SWT. Then, we recall the representative model of SWT (RSWT) which has been presented in detail in [4]. Section III presents the study of Nheowind-100, a 3.5 kw realistic model. Also, comparisons with the RSWT are made to validate the

2 Fig.. RSWT design. Similarly as in the RSWT, the near field coupling between the Nheowind-100 and BTS antennas will be highlighted. Lastly, conclusions and perspectives are presented in Section IV. II. A. Indicators Representative small wind turbine model (RSWT) REPRESENTATIVE MODELS OF SWT (RSWT) There are two important parameters to analyze variable backscattering produced by an object like a WT: RCS and Doppler shift. In [5], the RCS of an object is defined as the intensity of the radar echo or the energy scattered back to the source of the wave. Mathematically, RCS is calculated as RCS E s = σ = lim 4π r r E i where E s is the scattered electric field and E i is the incident field to illuminate the target. r is the observation distance from the target. Monostatic RCS is the particular case of bistatic RCS when emitter and receiver share the same position. In this work, we focus on with monostatic RCS. Doppler shift is a frequency deviation observed in backscattered signal due to the moving of rotor blades. It is dependent on rotational rate, rotor diameter and considered angles of illumination/observation. In [6-7] Doppler shift f D (for a given radial position R from blade root) is calculated as f D v R (1) = f 0 () c where v R (m/s) is the tangential velocity at a specific radial position on the rotating blade and f 0 (Hz) is the frequency of the transmitted signal. During rotation, if WT blades move away from the receiving antenna (v R > 0), then the Doppler shift is negative. On the contrary (v R < 0), if WT blades move toward the receiving antenna, the shift is positive. B. Geometry and Dimensions Based on another study by the same authors [4] in which the parametric RSWT model is presented in detail, we select here four nominal wind powers: 1.5, 3, 4.5 and 6 kw. These values cover nearly the whole range of common SWT nominal powers, which are used for BTS. Despite the diversity of shapes, sizes and materials, as mentioned before, the great majority of realistic SWTs include a nacelle, a tail and three blades. The SWT mast is not considered at this level of the study. Fig. lays out the geometry and parameters of the RSWT. This shape is chosen as a compromise between the constraints of simulation, manufacturing and the representativeness. It means the scattering produced by this model is expected to own RCS levels and variations which are as close as possible to those of the other models having the same nominal power. Concerning the model material, perfect electric conductor (PEC) is selected as the worst case [8], while fibreglass or carbon fibre for the blades will be studied later. C. RCS Analyses RCS analyses of four RSWT models (1.5 6 kw) were done by EM simulations which are implemented by the CAD EMSS/FEKO software, using the MLFMM method (Fig. 3). RCS obviously increases with SWT size. When the SWT power doubles from 1.5 to 3 kw, maximum RCS increases 7.7 db. For the triple and quadruple wind power, the RCS increment reaches respectively 11.8 db and 14.5 db. These results are observed for the front and back of RSWT (0 and 180 ), where the back scattering is the most significant, because the effective surface of RSWT presented by the blades is the largest. At the other positions, the RCS increase also corresponds to wind power, but it is much less than in the front and back of the RSWT, because the effective surface dominated by the nacelle and tail is much smaller. Among four RSWT models owning the same geometry as shown in Fig., the common case of 3 kw is first selected to study the effect of frequency on RCS (Fig. 4). For this purpose, we choose here 900, 1800, 100 and 600 MHz, which correspond to the frequencies of GSM, UMTS and 4G/LTE. The maximum value of RCS clearly increases with frequency, because the ratio of RSWT dimensions to wavelength is more significant. In the GSM frequency, from 900 to 1800 MHz, the increment of RCS (maximum value) reaches 6.37 db. Otherwise, from the GSM frequency 1800 MHz (old BTS) to the UMTS and LTE/4G ones (current or future BTS), RCS only increases about 1- db. Their RCS variations are relatively similar. Fig. 3. Monostatic RCS of four representative SWT models (1.5, 3, 4.5 and 6 kw) at 900 MHz in the horizontal plane (xy plane)

3 Fig. 4. Monostatic RCS of the 3 kw RSWT in the four frequencies (900, 1800, 100, 600 MHz) in the horizontal plane (xy plane) D. Blade Flash and Doppler Shift As mentioned in the big WT impact studies [9-10], blade flash is associated with a quick, significant RCS regrowth. It appears on the WT side when a blade is straight up or down. Thus, there are generally six blade flashes in a rotational period. The time between two flashes (blade flash period Tbl) is defined as T bl 1 = (3) N ω where N is blade number and ω r (rpm) is rotational rate of rotor. With N = 3 and the assumption ω r = 15 rpm (nominal ω r of the Nheowind-100), the flash period of the 3 kw model, described in Fig. 5 is T bl = 80 ms. Beside blade flash, another important impact produced by blade rotation is Doppler shift, which is a change in frequency of an incident wave due to blade motion. As shown in (), Doppler shift corresponds significantly to tangential velocity of a rotating blade v R, which is defined as v R = ω r (4) R R f D where ω r (rpm) is rotational rate of the rotor and r fd is distance between the centre of the rotation and interaction point. In [11], r fd of the big WTs is calculated as r f D = α r (5) ro to r where α values range from 0.5 to 0.7, depending on the blade Fig. 5. Spectrogram including the Doppler shift and blade flashes from the scattering which are brought by the blade rotation of the 3 kw RSWT at 900 MHz shape, and r rotor is the rotor radius [11]. For SWTs, to our knowledge, there are not any studies mentioning this. Hence, we select α = 1 as the worst case scenario. Consequently, maximum Doppler shift of the RSWT 3 kw is f D = 1 Hz at 900 MHz. It is described together with the blade flashes by the spectrogram in Fig. 5. In a rotational period T = 0.48 s, there are three negative and three positive Doppler shifts of flashes. Blade flash period is well noted T bl = 80 ms. At the flash positions, the power spectral density (right column) of the backscattered wave has maximal intensity (dark red). Additionally, the frequency deviation is maximal (close to 100 Hz). A constant residual frequency spreading about +/-40 Hz can be observed in any of the other blade positions. III. STUDY OF REALISTIC MODEL (NHEOWIND-100) Comparison with a realistic model is one of the next important steps to validate the RSWT. For this purpose, we selected the 3.5 kw Nheowind-100 model (Nheolis Company). This model was chosen because of its specific blade shape, with the goal of figuring out the broader diversity of the SWT shapes compared to those of big WT. A. CAD Design Transformation Original CAD models include many geometric details (small holes, screws, etc.) which induce a large number of meshes for EM simulations. This can make the calculation time and required memory increase a great deal, while they have no significant impact on results. Hence in the first step, we simplify the original CAD model of Nheowind-100 to delete such unnecessary details. Then, a geometric comparison between the original and modified models is given in Fig 6. It verifies the similarity of the two models. Fig. 6. Geometries of the original and modified Nheowind-100 models. Original model. Modified model Additionally, Fig. 7 presents the RCS difference resulting from the simplifications. Obviously, the RCS of the two models compared is nearly the same at most positions, notably for the maximum values. The small RCS differences observed at 90, 180 and 70 are mainly due to the six details connecting with braces and partially due to the blade layers. To highlight the benefit of modified Nheowind-100, we compare the calculation time and required memory of the two models. Consequently, while the simulations with the modified Nheowind-100 lose only 50 minutes with 1. GB of memory, those for the original model spend up to 7.5 hours with.7 GB. This confirms the relevance of the simplifications and explains the use of the modified Nheowind-100 model for the rest of the study.

4 Fig. 7. RCS comparison between three original Nheowind-100 blades and those of the modified model at 900 MHz in the horizontal plane (xy) B. RSWT vs. Nheowind-100 An RCS comparison between the RSWT and Nheowind- 100 models is presented in Fig 8. Contrary to the RSWT, the maximum RCS of the Nheowind-100 are not noted in front or back of the model in the horizontal plane. Instead, they appear at four positions from around 40 with a period of (about) 90. These positions are slightly shifted with blade rotation (not illustrated here) but generally do not induce any significant changes. The RCS of the Nheowind-100 is more variable than that of the RSWT. Indeed, while the maximal variation of RCS noted for the RSWT is 15 db, that of the Nheowind-100 reaches 5 db. Additionally, the RCS mean value of the Nheowind-100 (9. dbm²) is much stronger than that of the RSWT (4.4 dbm). This is more obvious in the vertical side Fig. 8. Monostatic RCS of the RSWT at 900 MHz is compared to that of the Nheowind-100. in the horizontal plane (xy). in the vertical side plane (yz). plane, where the RCS mean values of the two models studied are respectively 8.8 dbm² and 1.8 dbm². Otherwise, while RSWT produces just as many blade flashes as big WTs do (six flashes per rotational period), blade flashes caused by the Nheowind-100 are stronger and more numerous. The RCS deviation between the two models studied can be explained by the difference in dimensions and the specific curve of the Nheowind-100 blades. These results also suggest that a calibration of the RSWT dimensions would be useful in reducing the difference in the RCS mean values of the two models. Such calibration procedure is ongoing and it will be available to present during the conference. C. Near Field Coupling Fig. 9. Disposition between the Nheowind-100 and the panel antenna to study tnear field coupling. Side view. Front view. The near field coupling of the RSWT with a representative GSM panel antenna [1] was presented in [4]. As well as, we consider here coupling between the Nheowind-100 and the same antenna. Fig. 9 presents the disposition, in which the panel antenna is positioned below the Nheowind-100. Relative distance between them d is measured from the antenna centre to the nacelle centre of the Nheowind-100. We study here four cases of d = {0.375 d ro, 0.5 d ro, 0.75 d ro, 1 d ro } with d ro is the rotor diameter of the Nheowind-100. For the lowest position (0.375 d ro ), the antenna is completely obstructed by one blade in straight down position. Also, it is partially shadowed for the intermediate position d = 0.5 d ro. Otherwise, the antenna is not obstructed in the other cases. Besides d, blade rotation angle θ br is also an important parameter affecting dipole obstruction, then near field coupling. We consider four positions of θ br : 0, 30, 60 and 90. It is worth noting that because of the specific geometry of the Nheowind- 100, after 10 of rotation, blade position is repeated. EM simulations were carried out at 900 MHz. Fig. 10 presents antenna radiation patterns affected by the Nheowind- 100 with different relative distances d and when blade rotation angle θ br is 0. Only the lowest distances (d = d ro and 0.5 d ro ) present a considerable impact. In the horizontal plane, it consists of a gain reduction in the regions from -60 to 60 for the shortest distance and from -30 to 30 for d = 0.5 d ro. The maximum gain diminution of d = d ro is about 5 db while that of d = 0.5 d ro reaches about db. Additionally, the impact of the Nheowind-100 is noted on the undulation of the radiation patterns, notably for the two closest positions. Because of these effects, a question about the reduction in BTS cellular coverage is raised. Likewise, in the vertical front plane the main lobe of the radiation pattern is seriously degraded and

5 curved blade shapes of the Nheowind-100 explain why the blade flashes produced by this model occur much more frequently and are stronger than those of the RSWT and probably most small WTs owning the same nominal power. Like the RSWT, the near field coupling between the Nheowind-100 and a representative panel antenna has been studied. The obstruction of dipoles caused by the blades is the most significant impact. It creates not only a decrease in antenna gain but also the undulation of radiation patterns. In the worst situation, the gain decreases 5 db compared with that not affected by the SWT. Notably, a regrowth of db is noted in the back side lobe. This confirms the preliminary results obtained with the RSWT in [4]. The study of blade rotation impact on panel antennas is still on going and results will also be available very soon. The Doppler of the Nheowind-100 will also be studied. The effect of using fibreglass or carbon fibre materials for the blades instead of metal will then be studied. Manufacturing and measurements of a scaled small wind turbine model are also under preparation. ACKNOWLEDGMENT This work is supported by the European project OPERA- Net (Celtic-plus). the back side lobe increases significantly about db. This is a kind of the back flash which can degrade the performance of the cells located on the back of the panel antenna. This phenomenon is called intra-cell interference. Such a back flash effect has been reported for the RSWT in [4]. It occurs when one blade is straight down and obstructs significantly the antenna dipoles. IV. Fig. 10. Radiation patterns of the panel antenna affected by the Nheowind-100 vs. different relative distances d at 900 MHz. in the horizontal plane (xy). in the vertical front plane (xz). CONCLUSION In this work, we have presented the original results concerning RCS and Doppler shift produced by fully metallic SWTs, as well as their EM coupling with BTS antennas. To better address the diversity of SWTs, the representative model RSWT was first analyzed. RCS increases with both WT power and frequency. Among four models of 1.5 kw 6 kw, the common model of 3 kw was first analysed. Doppler shift produced by the blade rotation of this model is spread around +/-40 Hz, with flashes reaching +/- 100 Hz. Maximum RCS noted for the flashes is 4 dbm² and the flash period is 80 ms at nominal blade rotation speed. In the second part, the 3.5 kw Nheowind-100 (realistic model) was studied and compared with the RSWT scaled to have the same 3.5 kw. The comparison results between the two models show that the RCS of the Nheowind-100 is much more significant than that of the RSWT. Thus, a calibration of the RSWT dimensions is ongoing in order to reduce the RCS deviation compared to the Nheowind-100. The wide and REFERENCES [1] D. L. Sengupta, T. B. A. Senior, Electromagnetic interference to television reception caused by horizontal axis windmills, IEEE Proceedings, vol. 67, no. 8, p , July [] Gavin J Poupart, Wind farms impact on radar aviation interests, Final Report FES W/14/00614/00/REP DTI PUB URN 03/194, sept [3] Opera Net Project, [4] Tran Vu La, François Le Pennec, Christophe Vaucher,"Small Wind Turbine Generic Model Design for BTS Radio Interaction Studies", 4th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 013), 8-11 Sep. 013 (paper accepted and presented). [5] E. F. Knott, M. T. Tuley, J. F. Shaeffer, Radar Cross Section, nd edition, SciTech Publishing, 004. [6] M. Cheney, B. Borden, Fundamentals of Radar Imaging Rensselaer Polytechnic Institute Troy, NY and The Naval Postgraduate School Monterey, CA, pp. 3, Dec [7] D. Sozen, M. Kartal, Scatter and Doppler Effect of Wind Power Plants to Land Radars, 14th International Conference on Computer Modelling and Simulation (UKSim 01), p , 01. [8] L. S. Rashid, A. K. Brown, Radar cross-section analysis of wind turbine blades with radar absorbing materials, European Radar Conference (EuRAD 011), p , 011. [9] R. R. Ohs, G. J. Skidmore, G. Bedrosian, Modeling the effects of wind turbines on radar returns, Military Communication Conference (MILCOM 010), p. 7 76, 010. [10] A. Naqvi, S.-T. Yang, H. Ling, Investigation of Doppler Features From Wind Turbine Scattering, IEEE Antennas and Wireless Propagation Letters, vol. 9, p , [11] G. Greving, W.-D. Biermann, R. Mundt, Radar and wind turbines-rcs theory and results for objects on the ground and in finite distances, Microwaves, Radar and Remote Sensing Symposium (MRRS 011), p , 011. [1] François Le Pennec, Matthieu Roques, Sylvain Germaine, Dosimétrie des antennes relais des systèmes de troisième génération, RNRT ADONIS sous-projet ISIS, Rapport d étude Lot 3 / Tâche 3, nov. 005.