**4. Special topics in WAKE**

#### **4.1. Wind sector management**

Wind sector management refers to a process of attempting to maximize the cumulative wind farm output through an active optimization of wind turbine energy capture. There are currently two common approaches to this technique. One form of wind sector management is concerned with the shutting down of wind turbines downstream of a machine, which is creating a turbulent wake large enough to increase fatigue loads on the turbine. This can be more broadly stated as the curtailment of a wind turbine or turbines during special wind conditions that could cause fatigue damage [18]. The second approach refers to the curtail‐ ment of an upstream wind turbine that is producing influential turbulence to increase the production of downstream turbines and therefore increase the overall production of the farm [14]. Kjaer et al. [19] briefly discussed the concept of stopping a turbine for the purpose of preventing damage upstream or downstream while Neilsen et al. [18] gave an actual method for quantification of the reduction of turbulence intensity for protection of the downstream machines. These approaches will generally result in a decrease of wind farm power production as well. The concept of increasing wind farm production by reducing ax‐ ial induction has gained most of its attention from Corten and Schaak [20] of the Energy Re‐ search Centre of the Netherlands (ECN). A patent has been granted for the strategies developed at ECN [21] after wind tunnel testing showed an overall increase in production of 4.5% in a 6 row arrangement of turbines. The concept was explained in Schepers and Pijl [22] where results from ECN's full scale experimental wind farm were also given. The re‐ sults from the full scale farm show power gains of less than 0.5% when averaged over all wind conditions. However, performance increase is most noticeable when wind direction causes alignment of turbine wakes and also when wind speeds are below optimum rated speeds. This concept was also discussed in Johnson and Thomas [14] where a theoretical study was completed and control strategy developed which showed gains in wind farm power output. Although the overall increase in power production is not large it is important to note that very little alteration is required to achieve this improvement. A strategic change in control methods with no modification to hardware has the potential to make economic sense. Future research in this area is anticipated.

#### **4.2. Wake influenced yaw positioning**

**Figure 7.** Wind rose indicating percentage of wind direction probability. Data are for the upstream turbine over the

Onshore wind farm wake propagation is reduced by complex terrain and vegetation. As stated above, onshore wake propagation has been measured up to 15 rotor diameters downstream of a turbine. While optimal wind turbine spacing has been studied [7, 15Bryony L.D.P and Cagan, J., An Extended Pattern Search Approach to Wind Farm Layout Optimization, Proceedings of ASME IDETC: Design Automation Conference, 2010, 1-10.] further work on the limit of minimum wind farm footprint to maximize prof‐

Wind sector management refers to a process of attempting to maximize the cumulative wind farm output through an active optimization of wind turbine energy capture. There are currently two common approaches to this technique. One form of wind sector management is concerned with the shutting down of wind turbines downstream of a machine, which is creating a turbulent wake large enough to increase fatigue loads on the turbine. This can be more broadly stated as the curtailment of a wind turbine or turbines during special wind conditions that could cause fatigue damage [18]. The second approach refers to the curtail‐ ment of an upstream wind turbine that is producing influential turbulence to increase the

six month data set for all power producing winds (3-25 m/s) [11].

itability may be necessary.

74 Advances in Wind Power

**4. Special topics in WAKE**

**4.1. Wind sector management**

Wind turbines are typically independently controlled, relying on the data collected from the meteorological station situated on the back of the nacelle to dictate response. The turbine continually adjusts the orientation of the nacelle in order to face the best consistent wind di‐ rection. They typically only initiate a yaw movement after the new wind direction has been observed for a specified time to avoid constant "hunting" under rapid wind direction fluctu‐ ations. The turbines under consideration decrease the necessary consistent wind speed dura‐ tion required to command a change in yaw position as the wind speed increases. Figure 8 shows data for a range of wind speeds. The Figures represent the downstream nacelle posi‐ tions subtracted from the upstream nacelle position where a difference of zero represents perfect alignment with the lead upstream turbine. As the wind direction measured by the upstream turbine approaches direct alignment with the turbine array, the downstream tur‐ bine increases its yaw misalignment with respect to the upstream turbine. However, there are angles showing consistently large differences in yaw position that are not direct align‐ ment. The Figures show that the nacelle direction offsets change as wind speeds decrease. A nacelle position offset with a greater positive magnitude indicates the downstream turbine remains at an angle counter clockwise from the upstream turbine while a negative offset cor‐ responds with the downstream turbine positioning itself clockwise from the lead upstream machine.

Some patterns can be observed in the different plots although there is significant variation from one wind speed to the next. A steady increase in nacelle position misalignment for the array occurs in the wake zone, with greater offsets more likely to occur at TA +/- 5 to TA +/- 20 degrees (off wake centre). The first downstream turbine shows the least offset and the third downstream turbine shows the greatest offset. There is a large amount of variation in magnitude and profile of turbine misalignment for each upstream (free stream) wind speed range shown, however a distinct increase is evident for the nacelle position range encom‐ passing the wake zone as defined above. One possible cause of misalignment peaks are the vortex streets on the outer edges of the upstream wake profile. When free stream wind speed is not at turbine alignment (i.e. not coincident with the turbine array line) the down‐ stream turbine instrumentation may experience increased turbulence and rotational velocity in the wind. This could be due to the wake's outer edge of tip vortices passing over the wind speed and direction sensors of the downstream turbine. Similar results are evident for other arrays within the wind farm. An array of six wind turbines with identical linear alignment and spacing shows an increase in yaw misalignment within the wake region (Figure 9). The first downstream turbine has the smallest offset with progressively larger offsets down the array. There are distinct peaks in the alignment offset with an approximate return to 0 (+/- 2 deg). However, the two additional turbines for this arrangement complicate the interactions. As shown in Figure 9, the third downstream turbine agrees with the pattern in magnitude but its direction of rotation is opposite to the rest of the turbines in the array. Furthermore, the separation of nacelle position offset between the downstream turbines is less defined. This adds to the unpredictability of the yaw behaviour within the array, since it is not obvi‐ ous which turbine will show the greatest offset or at what wind direction it will occur. A potential source of some of these complications may be due to the mixing of each subse‐ quent turbine vortex street when not in direct turbine alignment as discussed earlier. This lack of distinction is further evidenced in the power coefficient profile shown in Figure 10. Although similar to the power coefficient profile given for the array of four turbines in Fig‐ ure 5, the separation of the lines for each downstream turbine is less defined. The third downstream turbine once again exhibits unique behaviour.

#### **4.3. Turbine operational sensitivity to wake**

With wake interactions accepted as an unavoidable fact it becomes useful to quantify the sensitivity of the operation of two turbines to this interaction. The purpose of quantification relates to the mitigation of negative effects and optimization of performance within the wake region. McKay et al. [23] presents the application of the Extended Fourier Amplitude Sensitivity Test method for determination of downstream turbine power output to upstream turbine operational parameters.

A global sensitivity analysis of eight fundamental operating parameters on wind turbine power output is performed. By comparing the sensitivities of normal operation to wake con‐ ditions, a better understanding of group turbine behavior is obtained. The most significant characteristic that is evident in the presented analysis is the effect that the introduction of wake has on turbine performance. For a turbine operating in the wake of an upstream ma‐ chine, power production is most sensitive to wind speed standard deviation above all pa‐ rameters included in this study, excluding wind speed itself as shown in Figure 11.

array occurs in the wake zone, with greater offsets more likely to occur at TA +/- 5 to TA +/- 20 degrees (off wake centre). The first downstream turbine shows the least offset and the third downstream turbine shows the greatest offset. There is a large amount of variation in magnitude and profile of turbine misalignment for each upstream (free stream) wind speed range shown, however a distinct increase is evident for the nacelle position range encom‐ passing the wake zone as defined above. One possible cause of misalignment peaks are the vortex streets on the outer edges of the upstream wake profile. When free stream wind speed is not at turbine alignment (i.e. not coincident with the turbine array line) the down‐ stream turbine instrumentation may experience increased turbulence and rotational velocity in the wind. This could be due to the wake's outer edge of tip vortices passing over the wind speed and direction sensors of the downstream turbine. Similar results are evident for other arrays within the wind farm. An array of six wind turbines with identical linear alignment and spacing shows an increase in yaw misalignment within the wake region (Figure 9). The first downstream turbine has the smallest offset with progressively larger offsets down the array. There are distinct peaks in the alignment offset with an approximate return to 0 (+/- 2 deg). However, the two additional turbines for this arrangement complicate the interactions. As shown in Figure 9, the third downstream turbine agrees with the pattern in magnitude but its direction of rotation is opposite to the rest of the turbines in the array. Furthermore, the separation of nacelle position offset between the downstream turbines is less defined. This adds to the unpredictability of the yaw behaviour within the array, since it is not obvi‐ ous which turbine will show the greatest offset or at what wind direction it will occur. A potential source of some of these complications may be due to the mixing of each subse‐ quent turbine vortex street when not in direct turbine alignment as discussed earlier. This lack of distinction is further evidenced in the power coefficient profile shown in Figure 10. Although similar to the power coefficient profile given for the array of four turbines in Fig‐ ure 5, the separation of the lines for each downstream turbine is less defined. The third

With wake interactions accepted as an unavoidable fact it becomes useful to quantify the sensitivity of the operation of two turbines to this interaction. The purpose of quantification relates to the mitigation of negative effects and optimization of performance within the wake region. McKay et al. [23] presents the application of the Extended Fourier Amplitude Sensitivity Test method for determination of downstream turbine power output to upstream

A global sensitivity analysis of eight fundamental operating parameters on wind turbine power output is performed. By comparing the sensitivities of normal operation to wake con‐ ditions, a better understanding of group turbine behavior is obtained. The most significant characteristic that is evident in the presented analysis is the effect that the introduction of wake has on turbine performance. For a turbine operating in the wake of an upstream ma‐ chine, power production is most sensitive to wind speed standard deviation above all pa‐

rameters included in this study, excluding wind speed itself as shown in Figure 11.

downstream turbine once again exhibits unique behaviour.

**4.3. Turbine operational sensitivity to wake**

turbine operational parameters.

78 Advances in Wind Power

**Figure 8.** Nacelle misalignment between wind turbines for 6 months averaged data. a) 9-10 m/s, b) 8-9 m/s, c) 7-8 m/s, d) 6-7 m/s, e) 5-6 m/s/ Lead upstream nacelle position is subtracted from each downstream turbine nacelle posi‐ tion [11].

**Figure 9.** Nacelle misalignment between wind turbines in array of 6 machines for 6 months averaged data. Lead up‐ stream nacelle position is subtracted from each downstream turbine nacelle position [11].

**Figure 10.** Six turbine array power coefficient profile for an upstream turbine wind speed of 8-9 m/s averaged over 6 months [11].

In essence, the method assigns a frequency to the upstream turbine operational parameters under consideration. The frequency data is input into a model which provides a power out‐ put signal related to the downstream turbine containing all of the frequency data in varying proportions. A Fourier transform is performed on the output and the resulting frequency content is ranked according to the operational parameters significance. This is shown in Fig‐ ure 11. The parameters chosen for the study along with their significance rankings are given in Table 1.

**Figure 11.** Frequency content of power signal extracted from wake condition data [23].


**Table 1.** Sensitivity indices for eFAST method applied to wake conditions [23].

**Figure 10.** Six turbine array power coefficient profile for an upstream turbine wind speed of 8-9 m/s averaged over 6

In essence, the method assigns a frequency to the upstream turbine operational parameters under consideration. The frequency data is input into a model which provides a power out‐ put signal related to the downstream turbine containing all of the frequency data in varying proportions. A Fourier transform is performed on the output and the resulting frequency content is ranked according to the operational parameters significance. This is shown in Fig‐ ure 11. The parameters chosen for the study along with their significance rankings are given

**Figure 11.** Frequency content of power signal extracted from wake condition data [23].

months [11].

80 Advances in Wind Power

in Table 1.

The comparison of results from the single turbine and wake conditions is given in Figure 12. Rotor RPM and wind speed are clearly the dominant features in the Figure followed by rela‐ tively similar magnitudes for each of the remaining parameters. Two main features become apparent. Firstly power output is more sensitive to rotor speed than wind speed under nor‐ mal operating conditions while the reverse is true under wake conditions. The turbines un‐ der study are designed to operate at an optimum tip speed ratio therefore the control systems will work to keep the rotor speed at specific RPMs depending on the wind condi‐ tions. This links rotor rpm directly to power output. In other words any changes in rotor speed will directly affect the power output of the turbine resulting in high sensitivity. The same is true of wake conditions as well. However, it has been shown above that for a tur‐ bine experiencing wake, changes in wind speed cause changes in power loss. Very high wind speeds reduce the power losses due to wake while wind speeds falling between 5 and 11 m/s can have a substantial effect on production. As a result, the power output becomes dependent on wind speed for power losses in wake in addition to the dependence of nor‐ mal, non-wake operation.

Secondly, the wind speed standard deviation's increases in sensitivity under wake. This is expected since the extraction of power by the upstream turbine leaves a turbulent rotating wake region. Therefore, variance in wind speed and direction increases. The increase is di‐ rectly linked to a loss in power for the downstream turbine, increasing sensitivity. By quan‐ tifying the sensitivity it is shown that changes in wind speed standard deviation are more critical to power production than all other parameters other than wind speed.

Utilization of this method to identify other power output sensitivities is possible. By further applying this method to more complex data sets, qualitative comparisons can be quantified, and subsequently, priorities can be placed on turbine operational parameters. This can be used for the purpose of optimizing performance or increasing turbine reliability. The results also suggest that through monitoring sensitivity indices, downstream machines may be able to determine whether or not they are in a wake. Depending on the severity of the turbulence in the wake the turbine could be controlled to mitigate negative effects or improve perform‐ ance. Additionally, the method could provide another tool to assess the efficacy of the origi‐ nal siting of existing wind farms.

**Figure 12.** Sensitivity index comparison for wake and non-wake conditions [23].
