Influence of wind speed variability and wind farm size

Two parameters, such as the wind speed variability and the number of wind turbines connected to a single large power converter (SLPC), appear to be crucial to analyse the proposed topology. On one hand, wind speed vari- ability in wind farms is difficult to analyse in a general way, since it heavily depends on multiple factors, such as the wind turbine distribution in wind power plants, the WPP location (onshore or offshore), the incoming wind directions, etc. However, it is possible to perform a statistical analysis in order to evaluate the effect of wind variation in terms of maximum available power, by considering the use of a single converter for the entire wind power plant or individual power converters for each wind turbine. On the other hand, the higher the number of generators connected to a common power converter, the worse the power extraction efficiency will be if different wind speeds among WTs are considered.

Hence, in order to analyse how the standard deviation and the number of wind turbines within a group affect the power extraction efficiency of the system, a range of simulations has been performed according to the methodology explained in Section 4.3.

A total of 10 OWPPs (NWPPA=10) composed of 2, 4, 6, 8, 10, 20, 40, 60, 80 and 100 wind turbines has been analysed. Obviously, it should be noted that offshore wind farms comprising few WTs such as 2, 4 or 6 are not an economically feasible option since the power output is too low. However, here they are not considered as wind farms, but as a cluster of machines connected to a single converter. The wind rose and Weibull distribution functions shown in Figure 4.7 has been used to define the wind conditions within all the OWPP considered for the study (NWCA=1). Likewise, the

number of standard deviations considered to assess the influence of wind speed variability within the OWPP on the power generation efficiency is set to 10 (NSTDA=10); varying its value from σ=0.5 to 5 m/s. Figure 4.13 depicts the obtained results. The ratio αopt indicates the power extraction

efficiency of the entire wind power plant and it is computed as αopt= PSLP C−V F

PM P C

(4.52) where PSLP C−V F and PM P C are the power generated by the WPP by con-

sidering the proposed SLPC–VF and the conventional MPC concept, respec- tively. 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.75 0.8 0.85 0.9 0.95 1 Standard Deviation [m/s] αopt 8 10 80 60 40 100 20 4 2 6

Figure 4.13: Dependence of the ratio αopt on both the size of the wind farm

and the wind speed variability within the wind farm.

Firstly, it can be noted that losses increase as expected, as the number of machines in the group as well as the wind speed variability among turbines increase. Compared with the MPC concept, the SLPC–VF scheme achieves between 98.42 and 87.83% of the maximum available power, if a wind power plant or a cluster composed of 10 machines or less are considered. Moreover, for low standard deviations (similar wind speeds throughout the wind power plant) and 10 WTs or less, the performance of the SLPC–VF is excellent (from 97.71 to 98.42%). However, if the standard deviation increases, the differences between the efficiency ratio αopt become higher. Thus, if a 5 m/s

standard deviation is considered, the CP losses of the SLPC–VF scheme

96 4.5. Further discussion about the SLPC–VF concept

In the case of large wind power plants, the dependence of the standard deviation on αopt is maintained independently from the number of turbines

within the wind farm. As it can be seen from Figure 4.13, for low standard deviations (0.5 m/s), the results obtained are quite satisfactory since αopt is

equal to 0.9667 for a wind power plant with 20 turbines and 0.8875 if the WPP is composed of 100 wind turbines. Similarly, when standard deviation increases, the performance of the single power converter topology decreases. However, the performance of a wind power plant with 100 turbines connected to a single power converter maintains αoptabove 0.85 (15% of CP losses) until

the standard deviation is larger than 2 m/s, which is larger than typical values of less than 1 m/s [128].

As it can be seen previously in Figure 4.13, higher number of machines connected to a common power converter implies worse performance of the system in terms of power extraction efficiency. Therefore, it will be required to analyse at what point the CP losses are so high that a single power

converter connected to all the WTs is no longer cost effective and different clusters of machines connected to a common power converter within a wind farm becomes a better option.

Thus, a wind power plant consisting of 40 wind turbines with three dif- ferent power converter layouts according to their connection between them and the wind turbines has been analysed (Figures 4.14 and 4.15). In Figure 4.14(a) the whole OWPP is connected to a single large power converter. In Figure 4.14(b) a dedicated power converter is connected to the first row of wind turbines whereas a second one is connected to the next couple of rows and a third power converter connects the remainder turbines. Finally, in Figure 4.14(c), five power converters are used to connect the entire wind power plant. In blue, a single power converter is connected to the first row of wind turbines. Wind turbines 6, 11, 16, 21, 26 and 31 (in green) and 10, 15, 20, 25 and 30 (in red) forms two more clusters of wind turbines connected to a single power converter, respectively. The fourth group consists of wind tur- bines located on the last row of the wind farm whereas the fifth one connects the remainder WTs to another power converter. As it is shown in Figure 4.15, the power generation efficiency is evaluated for different standard de- viations from 0.5 to 5 m/s (NSTDA=10) and compared between the three different WPP layouts depicted in Figure 4.14. As expected, the greater the number of clusters and power converters, the better the efficiency of the system will be. Hence, for low standard deviations (0.5 m/s), αopt is 0.9428

for case (a), 0.9497 for the WF layout (b) and 0.9537 for the case (c), when five power converters are connected to all the wind turbines distributed into five different clusters. Similarly, the resulting CP losses for high standard

wt1 wt2 wt3 wt4 wt5 wt6 wt7 wt8 wt9 wt10 wt11 wt12 wt13 wt14 wt15 wt16 wt17 wt18 wt19 wt20 wt21 wt22 wt23 wt24 wt25 wt26 wt27 wt28 wt29 wt30 wt31 wt32 wt33 wt34 wt35 wt36 wt37 wt38 wt39 wt40 (a) wt1 wt2 wt3 wt4 wt5 wt6 wt7 wt8 wt9 wt10 wt11 wt12 wt13 wt14 wt15 wt16 wt17 wt18 wt19 wt20 wt21 wt22 wt23 wt24 wt25 wt26 wt27 wt28 wt29 wt30 wt31 wt32 wt33 wt34 wt35 wt36 wt37 wt38 wt39 wt40 (b) wt1 wt2 wt3 wt4 wt5 wt6 wt7 wt8 wt9 wt10 wt11 wt12 wt13 wt14 wt15 wt16 wt17 wt18 wt19 wt20 wt21 wt22 wt23 wt24 wt25 wt26 wt27 wt28 wt29 wt30 wt31 wt32 wt33 wt34 wt35 wt36 wt37 wt38 wt39 wt40 (c)

Figure 4.14: Three different WPP layouts consisting of 1 unique WPP clus- ter (a), 3 WT clusters (b) and 5 WT clusters (c).

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.75 0.8 0.85 0.9 0.95 1 Efficiency Standard Deviation [m/s] WF layout a) WF layout b) WF layout c)

Figure 4.15: Comparison of power generation efficiency between the three different WPP layouts shown in Figure 4.14.

98 4.5. Further discussion about the SLPC–VF concept

deviations (5 m/s) are 18.25%, 14.78% and 13.12%, for cases a, b and c, respectively.

4.5.2 Influence of wind direction on power generation efficiency