Wind farms located within the radar line of sight of radars may interfere with the radar performance and degrade the ability to distinguish turbines and returns from targets of interest. Objections concerning the radar interference are affecting the development of numerous wind farms around the world. Several mitigation measures have been suggested to resolve this problem, including signal processing, the use of additional gap-filling radars and radar cross section (RCS) reduction measures on the turbine itself.
To reduce the wind turbine RCS, different measures can be taken for different parts of the structure—for example careful shaping of the tower and nacelle and by applying radar absorbing materials (RAM) to the blade [1, 2]. The application of RAM on blades is considered as a viable solution, and prototypes have been fitted on commercial turbines . One of the challenges facing the treatment of the blades with RAM is the potential degradation of the lightning protection system . The efficiency of the receptor based lightning protection system is reduced if the blade is fully treated with RAM . Therefore, a solution with partial RAM treatment is needed.
The work presented in [4, 5] considered a generic 40 m wind turbine blade fitted with a receptor based lightning protection system. It suggested a possible solution where the blade is partially covered with RAM while leaving untreated areas around the lightning receptor. These untreated areas are referred to in this paper as the clearance area.
This paper investigates the effect of varying the clearance area around the lightning receptors on the total RCS of a RAM treated blade. It presents RCS modelling results for different clearance patch sizes using the WinR (Wind turbine RCS) model. WinR is an RCS modelling tool developed at the University of Manchester, which imports the geometry of the blade for meshing and then use Physical Optics for RCS computations.
2 COVERAGE WITH RAM
When modelling the RCS of a turbine blade, the orientation of the blade with respect to the radar, the tilt and pitch angles and bending because of wind loading are all factors that will affect the RCS of the blade . The radar scattering is also dependent on the detailed geometry of the blade, wind speed and direction and other operational and external factors. The RCS modelling presented in this paper assumes no blade tilt or bending and is calculated by varying the yaw angle around a single blade at rotation angle of zero as shown in Figure 1. This orientation of the blade is chosen so that the maximum RCS level can be identified at approximately 180 o and 360 o yaw angles when the suction and pressure sides are illuminated, respectively. In reality, the RCS of a blade may be lower because of wind load bending, pitch and tilt angles. However, it is important to identify the peak RCS levels as these are the causes of ‘blade flashes’ in the radar return, which can have a significant influence on the radar.
For the purposes of this paper, all RCS calculations are compared with a baseline, which assumes that the untreated blade is made of a carbon fibre shell. Also, since some RAM solutions are designed to absorb at narrow band of frequencies, the model was used to predict the RCS at 3 GHz. This approximates to the frequencies used by air traffic control radars. It is worth noting that the electrical size of structure at radar frequencies extends over several hundreds or thousands of radar wavelengths. Thus, slight changes to the reflectivity of certain parts of the blade (the areas around the lightning receptors for example), geometry or radar frequency may result in significant changes in the details of the RCS profile, whereas the mean level and peak RCS values remain relatively unchanged. This is due to the changes in the coherent addition of the scattered waves from the structure, which may add constructively or destructively depending on the phase and amplitude of the scattering from the meshed blade surface. In order to overcome the rapid changes in the details of the RCS profile, the RCS is modelled at a number of frequencies within a narrow bandwidth ( ± 0.2 GHz). The average RCS is then computed and used in the analysis of the RAM treatment. This approach gives a smoother RCS profile making it easier to assess the effects of varying the clearance area.
An important aspect of RAM absorption performance is the dependence on the angle of electromagnetic incidence. This is particularly important for RAM covering complex shapes with curved surfaces and sharp edges. The RAM used in this modelling exercise is assumed to have a standard Salisbury screen angular sensitivity and a normal incidence reflection coefficient of -25 dB, which is a representative of RAM solution for the use on a wind turbine blade [7, 8].
A common approach for protecting the blades in accordance to the International Electrotechnical Commission (IEC) standards for the lightning protection of wind turbines is the use of metallic lightning receptors on the blade surface, which are then connected internally by electrical down conductors . The lightning receptor is typically 30 mm in diameter. An area clear of RAM treatment is needed around each lightning receptor to ensure the protection system is effective and to reduce the probability of damage during lightning strikes . Depending on the materials used within the construction of the blade, the RAM and the lightning protection system characteristics, the manufacturer may have a good estimate of the clearance area needed to maintain the lightning protection efficiency. On the basis of these parameters, the modelling results shown in this paper may then be used as guidelines for the possible RCS reduction that can be achieved by choosing a particular size for the clearance area. It is important to note that the structural integrity and efficiency of the lightning protection system takes precedence over the RCS reduction approach. Hence, if a particular clearance area is selected, the modified structure must then comply with the current IEC standards for lightning testing before implementing it on operational wind turbines.
3 VARYING CLEARANCE AREAS
The reduction in the RCS of the blade by applying RAM is heavily reliant on the total area covered. For simple shapes such as a flat plate, if only 50% of the area is covered with a perfect RAM, the RCS would be reduced by only 6 dBm 2. In practice, because of the complex shape of the blade, the affect will be different depending on which part of the structure is covered. Modelling shows that the peak RCS of a wind turbine blade can be in the order of 40–45 dBm 2 at 3 GHz. A significant RCS reduction may be needed to mitigate the effect on adjacent radars. Therefore, it is important to cover as much of the blade as possible with RAM.
The overall effectiveness of the RAM treatment on radar systems does depend on the type of radar used. It may often be the case where at certain illumination angles the RCS of the tip region is inherently smaller than the RCS near the blade root. However, when the side of the turbine is illuminated, radar systems using Doppler processing may be affected more by higher Doppler frequencies generated from the blade tips. Thus, the actual impact of RAM treated blades on particular radar can become a complex question.
The aim of this work is to provide manufacturers with guidelines to estimate the RCS reduction that can be achieved through RAM treatment of the blade surface while leaving a clearance area around the lightning receptor. Figure 1 shows a typical 40 m blade with three lightning receptors placed at 20 m (LR3), 30 m (LR2) and at the tip of the blade (LR1). It also shows the clearance radius, Rc, of the untreated area around each receptor. By varying Rc, the total treated area changes, which then affects the total RCS of the blade. Figure 2 shows RCS modelling results for an untreated blade, a fully treated blade and a blade with three receptors and Rc of 1 m.
The results show that both fully treating the blade with RAM and leaving a clearance of 1 m around receptors give good reduction in RCS in the regions near the leading and trailing edges. When the leading and trailing edges of the blade are illuminated, the geometrical surface area of the first 35% of the blade root make up more than 90% of the total illuminated blade surface area. From the RCS view point, the blade root has large radii of curvature making it a very reflective portion of the blade. As the root portion of the blade is fully covered for all the clearance radii simulated, a significant reduction in the RCS is achieved when illuminating the leading and trailing edges.
It can be noted that at some angles, the RCS of the RAM treated blade may exceed that of the untreated blade. This rare phenomenon occurs because of the change in the coherent summing of the amplitude and phase contributions from the blade surface. This effect will only take place in regions outside the peak RCS values. As this paper aims to investigate the effect of RAM treatment on the peak RCS values of the blade, this phenomenon is not considered to be a significant concern.
When addressing the peak RCS regions, namely the suction and pressure sides, the RCS of the fully treated blade showed good reduction, whereas introducing the 1 m clearance reduced the peak RCS by 12.2 and 2.6 dBm 2 in the pressure and suction sides, respectively. The reduction in the total blade RCS is dependent on the size of the clearance area. Figure 3 gives a summary of RCS modelling showing the peak RCS level when illuminating the suction side, the trailing edge, the pressure side and the leading edge for different Rc values. The peak value for each orientation is taken as the maximum RCS value in the shaded region illustrated in Figure 2. The horizontal axis shows the variation of the clearance radius Rc from 0 to 3.5 m, where 0 represents a fully treated blade, whereas the shaded region is the RCS of the untreated blade.
The RCS of the trailing and leading edges remains relatively unchanged for all values of Rc because of the RAM treatment of the blade root section. On the other hand, the RCS of the suction and pressure sides rises significantly as Rc increases to 1.5 m. For Rc values larger than 1.5 m, the peak RCS at the suction and pressure sides becomes close to that of the untreated blade. Lightning protection requirements may demand larger clearance areas around the receptors, which will reduce the effectiveness of the RAM treatment. For three bladed turbines within a large wind farm, this may still cause degradation in the performance of nearby radar systems. Hence, further reductions might be required. This may be achieved by using different Rc for each receptor along the blade or using a fewer number of receptors. This is a question of detailed lightning protection design and may be subjected to lab testing using the IEC standards for lightning protection testing prior to implementation .
4 REDUCING THE NUMBER OF RECEPTORS
The number of receptors used on a blade is not dictated by the IEC standards for lightning protection of wind turbines . The IEC standards do specify the testing procedures to ensure a particular design is suitable to implement on operational wind farms. In the modern turbine designs, it is a common practice to use three receptor points on each blade for lightning protection . These receptors, as shown in Figure 1 as LR1, LR2 and LR3. This design has been used for many years, and statistical data prove its suitability and efficiency.
Alternatively, experience from a large number of wind turbines has shown that more than 88% of lightning attachments occurs within the outermost 1 m of the blade tip . Bertelsen  mentioned that practical experience has shown that blades most often are struck in the tip region, and adequate protection may be achieved without the need for additional receptors. Such a solution is considered to be more feasible from a mechanical point of view, since the blade skin will be stronger and lighter with fewer physical penetrations. Bertelsen  also concludes that the rear of the nacelle is struck more often than the side receptors inboard the blade tip. This is difficult to simulate in a laboratory testing environment because of the size limitation of testing facilities, which may not be large enough to accommodate a full rotating blade and nacelle.
The impact of RAM treatment of turbine blades on the efficiency of lightning receptors is still under research and testing. For the purposes of this study, it is assumed that the receptor efficiency can be maintained, and the RAM treated areas will not alter the probabilities of lightning attaching to the receptor points. On the basis of this assumption and on the work presented in [11, 12], it is also assumed that the number of receptors may be reduced while maintaining the integrity of the lightning protection system. This will allow more of the blade to be covered with RAM, and therefore, further RCS reduction can be achieved.
With the two receptors approach, only the LR1 and LR2 receptors are considered while removing LR3 from the blade shown in Figure 1. When considering a blade with one receptor only LR1 is included in the modelling. Figures 4 and 5 give the peak RCS levels with varying Rc values for the two receptors and the single receptor blade, respectively.
Comparing the RCS for the blade with three receptors and that with two receptors only show no significant improvement in the reduction achieved by removing LR3. This is due to the high reflectivity of the flat area around LR2, which dominates the RCS of the full blade when the suction and pressure sides are illuminated. When LR2 is removed and a single receptor is considered (LR1), the results show that good reduction of 13 dB can still be achieved with Rc values up to 3 m. The RCS of all the modelled scenarios shows a relatively constant level of RCS for the trailing and leading edges because of the RAM treatment of the root section of the blade as mentioned earlier.
When addressing the impact of the rotating blades on radar systems that rely on Doppler processing to distinguish between static clutter and moving targets, the movement and the RCS of the blade tips becomes an area of concern. The Doppler signature generated by the rotating blades is most significant when the turbine is illuminated side on (90 o and 270 o yaw). In such orientation, the tips of the blades move at high speeds that are similar to small aircrafts or commercial airlines that are on a landing course. Thus, the reduction of the tip RCS might be crucial in order to mitigate the interference with Doppler based radars.
5 RAM VERSUS BLADE TIP RCS
Air traffic control (ATC) and other safety critical radar systems rely on the Doppler signatures of radar returns to identify targets of interest. The returns from fast moving objects are processed, whereas the static or slow objects are suppressed typically by the moving target indicator filter. Wind turbines with their large rotating blades are designed to track and face the flow of wind. Depending on the operation conditions and the orientation with respect to the radar, the blades rotation causes the tips to move at high speeds causing a high Doppler shift in the radar returns. This may cause the returns from the blades to break through the Doppler and moving target indicator filters causing detection and initiation of false tracks or alteration of existing tracks.
The current generation of horizontal axis wind turbines has a distinctive blade profile. Although the exact geometry of the blade may vary depending on the manufacturer, a taper between the wide blade root and the very narrow tip is a norm. This means that the radii of curvature become increasingly small towards the tip of the blade. For RAM solutions with high angular sensitivity, the performance will gradually degrade as the curvature becomes steeper towards the tip. Hence, covering the very tip of the blade with RAM may have limited effect on the RCS of the leading and trailing edges near the blade tip. This can be seen when modelling only the last 6 m of the blade and varying the clearance areas around the tip receptor as shown in Figure 6.
As seen earlier, the peak RCS of the tip can be reduced by applying RAM, and the reduction is dependent on the size clearance patch. These peak values occur when the front and the back of the blade are illuminated. In such angles of illumination, the blade may have high RCS but produces a low Doppler shift. For the parts of the blade that may result in high Doppler, i.e. the leading and trailing edges, it can be noted that a fully treated blade tip reduced the RCS of the leading edge by only 16 dB rather than the specified 25 dB reduction of the RAM. Partial treatment of the tip by varying Rc shows that the leading edge RCS slowly rises to -5 and -2 dBm 2 at Rc of 2.5 and 3 m, respectively. The RCS of the trailing edge remains relatively unchanged regardless of the size of the clearance patch. The RAM treatment of the sharp sections of the blade appears to have little effect on the RCS because of the RAM's angular sensitivity. However, this should not deter from considering RAM treatment as a potential solution. For this particular blade geometry, the RCS of the tip's leading edge is 0.3 and 0.6 m 2 at Rc of 2 and 3 m, respectively. Depending on the radar system, application and subject to further studies, this might be considered to be small enough to be ignored at extended range.
To illustrate the effect of partial RAM treatment on the Doppler signature of a blade, the model was used to simulate the Doppler returns from a single blade over a full rotation. In this modelled scenario, the blade is initially illuminated with the leading edge facing the radar. As the blade rotates, the tip faces the radar at 90 o, then trailing edge is illuminated at 180 o and the bottom of the blade is illuminated at 270 o. This particular orientation is chosen as it produces the highest Doppler returns, which may affect Doppler based radars. The results for an untreated blade are shown in Figure 7.
The results show that the Doppler plot follows a sinusoidal pattern as the blade rotate as might be expected. The returns are high at distinct regions when the leading edge of the blade is illuminated (at 0 o and 360 o) and the root section of the blade when the trailing edge region is illuminated (170–200 o). This agrees well with the measurements shown in . It also confirms that the highest RCS contribution comes from the root section of the blade at this orientation angle. This can be seen when the fully and partially RAM treated blades are modelled in Figures 8 and 9, respectively.
Treating the blade with RAM immediately shows a significant difference in the RCS levels of the regions with larger radii of curvature, i.e. the blade root and the leading edge. The regions with steep curves and sharp edges see no significant change in the RCS levels. This is also notable in Figure 9, where the partially treated blade is modelled. The untreated regions around the middle receptor still produce high RCS, which can be clearly seen in the figure. The clearance area around the second receptor produces a region of slightly higher RCS than the rest of the blade. The untreated tip appears to have little effect on the Doppler signature as anticipated. This is due to the sharp curves of the tip region, which will degrade the performance of RAM solutions because of its angular sensitivity . The use of RAM solution of better angular sensitivity may improve these results. Furthermore, it may be possible to apply a thin strip of parasitic RAM along the leading edge to reduce the RCS of the leading edge at tip region of the blade.
The use of RAM treatment to reduce the RCS of the blades is considered as a possible solution to mitigate the interference of wind farms with radars. RAM needs to be applied to most of the blade surface in order to achieve sufficient RCS reductions. However, fully treating a blade may affect the integrity of the lightning protection system fitted on the blades. Thus, a partial RAM treatment solution is considered where clearance areas around the lightning receptors are left untreated. For a typical blade with three receptors, leaving clearance areas with radii 1.5 m or more may not achieve the desired reduction.
Since lightning strikes mostly occur at the tip receptor, it was assumed that the number of receptors may be reduced without degrading the efficiency of the lightning protection system. Reducing the number of receptors allows more of the blade surface to be treated with RAM. Having a single receptor near the blade tip with clearance radius of 3 m allows the partially coated RAM blade to achieve a reduction in the peak RCS by 15 dBm 2. This may be sufficient for radar systems that employ Doppler processing since the peak blade RCS occurs at illumination angles that produces low Doppler shift. This is dependent on the radar system and its intended application and may need further investigation.
When the leading and trailing edges are illuminated, a significant reduction can be achieved by fully treating the root section of the blade with RAM. This will reduce the RCS of the regions of high scattering and low Doppler returns. The regions with high Doppler returns are located at the tip of the blade and may already have a very low RCS profile depending on the blade geometry. The RAM performance is dependent on the incident angle of the incoming wave. If further reduction to the tip RCS is needed, using a RAM with a standard angular sensitivity may not have a significant effect because of the sharp geometry of the blade tip. Using a RAM solution with wide angular properties or applying a strip of parasitic RAM along the leading edge of the blade may enhance the RCS reduction of the blade tip region.
The support of this work by the EPSRC's Supergen V Wind Programme is gratefully acknowledged. The authors would also like to thank the members of the former DTI's stealth technology for wind turbines consortium.