Wind farm HVRT capability improvement based on coordinated reactive power control strategy

: Voltage swells after grid fault clearance is one of the major causes of large-scale wind power tripping accidents in China. In this study, reasons that led to overvoltage phenomena were researched, and factors that in ﬂ uence the grid overvoltage during grid fault were quan-titatively analysed through simulation. A coordinated reactive power control strategy combined the control of wind turbines with reactive power compensation devices was proposed to prevent the occurrence of high voltage and improve the high-voltage ride through capability of wind power plant. Finally, the effectiveness of the proposed control strategy was indicated by the simulation results in DIgSILENT/ PowerFactory.


Introduction
With the continuously increasing wind power in the weak grid in China [1,2], the impact on the stable operation of the power grid is much notable than before [3]. The large-scale wind turbine (WT) tripping accidents occurred in 2011 made grid operators aware of the importance of wind power plant (WPP) low-voltage ride through (LVRT) capability. However, it should be noted that many of the WTs which ride through low voltage were tripped because of the voltage swell after fault clearance [4][5][6][7][8]. To safeguard the network against these problems, grid operators in China and some other countries have recently enforced stringent requirements on the high-voltage ride through (HVRT) capability of large WPPs. A comparison of the international grid codes shows that AEMC in Australia and Energinet.dk in Denmark have the most stringent regulations on HVRT, and the maximum voltage at the instant of fault existence is 1.3 pu of the nominal voltage at the point of c o m m o nc o u p l i n g [ 9].
Previous studies on HVRT are mainly focused on the control strategy of a single WT in the transient process. The HVRT capability of WTs studied in [10][11][12][13]14] shows the improvement of HVRT capability for WTs with a full scale converter with the application of STATCOM. Results have shown that it is an effective approach to meet the HVRT requirements for a single WT by modifying converter control schemes. Other research studies are focused on the hierarchical reactive power control or voltage control strategies of a WPP [15][16][17], which is mainly concerned about the steady state voltage stability.
There are different types of transient overvoltage in the power grid, such as the switching overvoltage [18], lighting overvoltage [19], and overvoltage caused by high voltage direct current (HVDC) monopole or bipolar block, etc. However, according to the survey of large-scale WT tripping accidents happened in China, the main reason is the reactive power redundancy during and after grid short circuit faults, which is caused by an inappropriate control strategy of the WT and reactive power compensator. As a result, to avoid the voltage swell after grid fault, the reactive power of the WPP must be co-ordinately controlled during this transient process, and this is the main concern in this paper. This paper is organised as follows. Section 2 presents the two main reasons that have result in voltage swell and large-scale WT tripping accidents in China in recent years. In Section 3, the influence of the WT control characteristics and grid parameters on grid overvoltage is analysed. A coordinated reactive power control strategy for the WPP is proposed in Section 4, and the result is demonstrated by means of dynamic simulations with DIgSILENT/ PowerFactory.

Causes of overvoltage in WPP
In the actual power grid with a high wind power penetration level, overvoltage can be caused by various reasons. According to the survey of large-scale WT tripping accidents happened in China, there are two main reasons that may lead to voltage swell.

Inappropriate switching of reactive power compensators in steady states
WPPs are generally located far from the load centre and connected to the weak grid in China, and the grid impendence is always in a relatively great value. A large number of active power transmission in the line will easily get to the static voltage stability limit of the system, and in which situation, small disturbance of reactive power may lead to grid voltage swells.
The equivalent schematic diagram of a WPP integration system is depicted in Fig. 1, which has a structure similar to the single machine infinite bus system.
In Fig. 1, U WPP is the voltage at point of interconnection (POI) of WPP, U grid is the voltage of grid, P WTs and Q WTs are the active and reactive power generated by the WTs, X is the equivalent impedance of the WPP transmission line, and B is the equivalent capacitance of the WPP reactive power compensator.
Choosing the reference phaseU grid = U grid /0 • , U WPP = U WPP /d, active and reactive power generated by WTs can be expressed as follows [20]: If WTs operate at unity power factor, the following equation is obtained: Suppose that the grid voltage U grid is constant, the relationship between P WTs and U WPP under different reactive power compensation capacities is shown in Fig. 2 (S B = 1000 MVA). The WPP is operating at Point A at first, where U WPP = 1.0 pu, P WPP= 1.0 pu, B = 0.33 pu. If a reactive power compensation device, such as a 20 Mvar fixed capacitor, is switched on, the operation point will be changed to Point B with a voltage swell. WTs without HVRT capability may disconnect from the grid, and with the reduction of active power, the operation point will move towards Point C. Voltage at POI will be further increased, leading to more WTs tripping. This is the chain reaction in the WT tripping accidents in Guyuan, Hebei Province on 14 May 2012 [21].

Inappropriate control of WTs and reactive compensation devices in transient process
On the other hand, the inappropriate control characteristics of WTs and reactive power compensation devices will result in the overvoltage during a grid fault. In Fig. 1, P WPP and Q WPP are the active and reactive power generated by the WPP, and Q RCs are the reactive power from reactive power compensation devices. As a result, P WPP = P WTs , Q WPP = Q WTs+ Q RCs , and the following equations can be obtained It can be seen from (6), the voltage of the WPP is increased with Q WPP and decreased with P WPP , which means after grid short circuit fault clearance, both reactive power redundancy caused by the reactive current over injection and the active power loss caused by the low active power recovery speed will result in overvoltage. It will aggravate the voltage swell if the control strategy and response time of reactive power compensation devices cannot meet the requirements of transient voltage regulation. These are the reasons for WT tripping accidents in Jiuquan, Gansu Province on 24 February 2011 [7].

3I n fluence factor analysis on overvoltage
Overvoltage after grid fault is mainly caused by continuous reactive current injection after the fault clearance. So the maximum voltage is strongly correlated to the control characteristics of WTs and the grid structure. Factors that influence the grid overvoltage after fault clearance is analysed through simulation in this section.
A typical WPP was modelled in the simulation software DIgSILENT/PowerFactory. WTs were equipped with a full power converter that can independently control active and reactive powers. Internal details of the WPP should be modelled with sufficient accuracy so that the currents and voltages within the WPP can be evaluated and constraint violations within the WPP can be monitored.
The layout of the WPP is shown in Fig. 3. The WPP consists of 25 × 2 MW WTs and is connected to the transmission system with a 35 kV/220 kV transformer, which is modelled explicitly. The WPP has a total of two overhead feeders of different lengths, and the distance between adjacent WTs is set to 0.5 km. A static synchronous compensator (STATCOM) of 10 Mvar and During the simulation, the WPP was producing at a level of 100% of its nominal capacity prior to the fault, and a three-phase short circuit fault was assumed to occur 1 s after the start of the simulation. The fault reactance was adjusted such the voltage at the POI dropped to around 0.2 pu and its duration was set to 500 ms.

Control characteristics of WTs
The reactive current injection capability during voltage dips is required by grid codes in many countries, which is usually expressed as in (7) I q ≥ kf (U T )I N , where I q is the reactive current of the WT during a fault, U T and I N are the terminal voltage and the nominal current of WT, and k is the factor that defines the requirement of I q at a certain U T . The response time of I q is also required, such as 75 ms after the inception of LVRT in China and 100 ms in South Africa. As there is no upper limit for I q in these grid codes, some manufactures may use a k-factor as large as possible to meet the response time requirement. Fig. 4 shows the over voltage comparison between different settings of k-factor and time delay of WT under the same voltage dip. To make the comparison more apparent, the most remote WT from the POI was selected as the study object. In Case 1, as there was no reactive current support provided by the WT during a fault, no overvoltage was detected after a fault. Case 2 shows the simulation results of good control characteristics, which used a proper k-factor with a relatively short control delay (k = 1.2, t delay = 0.01 s). The reactive current injected at the low-voltage side of the WT transformer was 1.5 kA during a fault, and the grid voltage was stable to a pre-fault value in quite a short time after fault clearance. While in Case 3, inappropriate control parameters were used (k = 1.6, t delay = 0.04 s), and after the clearance of grid fault, the WT did not stop providing reactive current and the voltage suddenly jumped over the nominal value and continue increasing to the peak value of about 1.35 pu.
The grid voltage under different active power recovery speed after grid fault clearance is shown in Fig. 5. When the active power recovered at a slow speed (0.2 pu/s), there would be a voltage swell for the lack of active power. If the active power recovered at a very fast speed (2 pu/s), there would be an obvious voltage oscillation.

Short circuit capacity (SCC) and cable length
The voltage swell also shows a high dependency on grid impendence. The relationship between the maximum voltage during the fault ride through the process the SCC of the power grid and the cable length from the low-voltage side of the WPP transformer to the most remote WT is shown in Fig. 6.
During the simulation, the control parameters in Case 3 were used, and the range of the cable length was from 5 to 30 km. SCC was set from 400 to 1600 MVA. This gives a short circuit ratio from 8 to 32 for the considered WPP rating, which covers the range from common interconnection to strong interconnection.
As can be seen from Fig. 6, the maximum voltage at POI raises from 1.05 to 1.2 pu with the decrease of the SCC, and the cable length has a significant influence on the voltage of the WT terminal. With the increase of the cable length, the maximum voltage will increase about 0.1 pu, which means the most remote WT may face much serious voltage swell than the WT near the substation.

Coordinated reactive power control strategy
The voltage swell and reactive power redundancy of the WPP is a complex issue caused by multi-reasons, and cannot be solved by simply improve the HVRT capability of WTs or reactive power control scheme of WPP. To reduce the risk of wind power trip off under high voltage, a coordinated reactive power control strategy including transient and steady state control of WT and reactive power compensation devices is needed.
For the control of WTs, the reactive current injection factor k in (7) should be calculated according to the grid impendence of each WT. The grid impendence is an important parameter in the voltage control loop since it determines how much the voltage can be changed when a certain amount of reactive current is injected from WT. For the WT located further from the POI (with a higher grid impendence), a smaller number of k should be selected. On the other hand, to avoid the overvoltage at the WT terminal after fault clearance, the control delays of reactive current should not exceed 20 ms, and k should be limited to a maximum 1.5. Inductive reactive power generating is needed if the voltage exceeds 1.05 pu after fault clearance, and the active power recovery speed is set to 1 pu/s. For the coordinated control of reactive power on the farm level, the reactive power is not provided by static var generator (SVG) but by FCs and WTs during normal operation. SVG with fast response characteristics is only used during the transient process of grid fault. Instead of connected to the grid all the time, FCs should be switched off if the voltage is below 0.5 pu or above 1.05 pu and connected to the grid by steps after voltage recovery.
The simulation results of the WPP with the traditional control strategy and the coordinated control strategy are shown in Figs. 7 and 8. Comparison between the strategies used in the simulation is shown in Table 1. Fig. 7 shows simulation results of the most remote WT from the POI. There is an obvious voltage swell of about 1.2 pu after voltage recovery under traditional control strategy and results in the tripping off of the WT. As can be seen from Fig. 7c, the WT does not stop injecting reactive current until 80 ms after fault clearance, which is the main reason that leads to the overvoltage. While under a coordinated control strategy, no voltage swell appeared and the WT has successfully ridden through grid fault. The reactive power of the WT is plotted in Fig. 7d, and in order to ensure the reactive power from SVG is 0, the WT is generating 0.1 Mar reactive power during steady state under a coordinated control strategy. Fig. 8 shows the simulation result at POI. It can be seen that there is a voltage swell about 0.1 pu and last for 0.08 ms after voltage recovery under a traditional control strategy. The active power after grid fault is 24 MW, which means that more than half of WTs are tripped because of the high voltage protection. While the WPP has successfully ridden through grid fault and avoided the   Fig. 8d. FCs are always connected to grid in the traditional control strategy, while in the coordinated control strategy, FC is disconnected when the grid voltage is extremely low and connected to the grid by steps after the voltage is stable to avoid a sharp rise of the grid voltage.

Conclusion
The two main reasons for overvoltage in WPP was analysed based on large-scale WT tripping accidents occurred in China, which are the inappropriate switching of reactive power compensation devices during steady states and inappropriate control of WTs and reactive compensators in the transient process.
A typical WPP model with WTs and different types of reactive power compensators was built in the simulation software DIgSILENT/PowerFactory. Factors such as WT control characteristics and grid parameters that influence the grid overvoltage after fault clearance were analysed. Simulation results show that the voltage at the WT terminal is proportional to grid impendence, and the grid voltage is significantly influenced by the reactive power control parameters such as k-factor and time delay during a grid fault.
A coordinated reactive power control strategy is proposed to avoid the voltage swell caused by reactive power redundancy during and after grid short circuit faults. The comparison of the traditional control strategy and the new proposed control strategy is listed. The simulation results between the traditional control strategy and coordinated control strategy have proved the effectiveness of the proposed scheme.

Acknowledgements
The authors gratefully acknowledge the financial support by the Research Project of State Grid Corporation of China: Simulation and Evaluation Technology for Renewable Power Grid Integration and Technical Standard Development (NY71-15-037).