Assessment of Flicker Emission in a Grid Connected Wind Farms

Wind farms in Tamil Nadu's Coimbatore district have identified several issues, including flickering emissions, frequent generator tripping, and power evacuation issues caused by a weak grid. Taking into consideration the scarcity of research on flickering emissions, this work focuses on the causes of short-term flicker severity (Pst) in wind farms that export generated power to industrial loads. To identify the scenarios that cause flickering, simulation models of fixed speed wind farm and variable speed wind farm with controllers were developed using DIgSILENT power factory software. The flicker emissions were measured at the wind farm substations using Fluke and Dranetz PX5.8 power quality analyzers in accordance with the IEC 61400-21 standard. To validate the simulation model, the results from the flickermeter during the simulation and the field measurements were compared. According to the results of this research, both fixed and variable speeds produce flicker emissions that exceed the IEC standard limit, that causes the power electronics-based industrial drives to fail to operate. The controllers were developed to improve the performance of wind farms that will benefit the current and future wind energy-efficient conversion systems.


Introduction
As the penetration of wind power into the grid is increasing, the power quality problem is becoming a major issue, causing concern for the power sector. Flickering is one of the main issues concerning the power quality of wind electricity generation [1]. Due to voltage fluctuations, the flickering is induced in the load flow. In a grid connected wind farm, flicker emission happens because of the fluctuations in power due to the variations in wind speed, wind shear, and tower shadow effects [2][3][4]. It depends on mean wind speed, short circuit capacity, and grid impedance angle [5][6][7][8][9][10].
The Variable Speed Wind Generator (VSWG) with permanent magnet synchronous generator (PMSG) has become more famous in recent days due to its' parameters like high power factor, high efficiency, high power density, and easy control [11][12][13]. To optimise the extraction of wind energy, variable speed turbines use the maximum power point tracking (MPPT) algorithm. In the MPPT algorithm, the powercontrol mode (PCM) is used to optimise the active power on the generator side [14]. The authors proved that the power control mode is superior to the speed control mode for power quality mitigation. Presently, new varieties of wind turbine employ power electronic devices and the converters control schemes were developed at the generator side and the grid side which helps in mitigating the flicker [7,[15][16][17][18][19][20][21].
In the case of Fixed Speed Wind Generator (FSWG), a squirrel cage induction generator connected directly to the grid was utilized. In this type of configuration, both the active and reactive power variations contribute to the flicker emission and thus subjected to the aero dynamical behaviour of the turbine [22]. The causes of various power quality issues were simulated and measured at the fixed speed wind farm substation where frequent tripping of the generators occurred [23]. In this paper, the analysis of short-term flicker emission was not studied. In [24][25][26][27][28], a static-synchronous 1 3 compensator (STATCOM) was considered to mitigate the flickering by controlling the reactive power flow from the grid. The level of flicker emission caused by the FSWF drastically reduced when the STATCOM was connected at the point common coupling (PCC).
For measuring and assessing the power quality behaviour of grid connected wind electric generators, the International standard IEC 61400-21 has worked out plans [29]. The standard tells the parameters which has to be calculated for determining the power quality issues. The field study with power quality analyzers gives important information related to the integration of large wind farms with the grid [30,31].
A field study was conducted in the FSWF and VSWF substations located at Coimbatore, which directly feed the industrial load. It was reported that there was frequent tripping of generators and large flicker emissions in the substation. The authors were motivated to study the flicker emission in the substation. In this paper, realist model of both the wind farms were developed and simulated using the DIgSILENT power factory software. Various effects of sudden load variation and different types of faults on flicker emission were considered for a given wind speed, shortcircuit capacity and grid impedance angle of the substation are analysis using flicker meter tools in this software. For the effective control of above events, converter side controllers for the VSWG and STATCOM for the FSWG were developed and incorporated in this paper.
The research work is organised as follows: in Sect. 1, a detailed analysis of flicker emission based on the recorded and simulation data from various wind farm substations is presented. Section 2 discusses the mathematical models of various components of VSWT. In Sect. 3, the model of a fixed-speed wind generator is described. Section 4 deals with the simulation models of two typical wind farm substations in DIgSILENT power factory software. In Sect. 5, the measured data obtained from the two wind farm substations is analysed. The flicker emissions in the wind farm were evaluated for various events such as load variation, faults, and interruption. Section 6 considers the comparison of simulation results with those recorded in the substation. Section 7 discusses the conclusion.

Model of the Variable Speed Wind Electric System
The Variable speed wind farms under study use Permanent Magnet Synchronous Generators (PMSG) with back-to-back converters. A complete wind turbine model consisting of wind speed, pitch controller, aerodynamics, shaft and electrical components such as synchronous generator, PWM converter, and transformer were developed.
The mathematical models of the various components discussed in this section were implemented in DIgSILENT power factory software [32] to study the flicker emission. The wind speed model has four parameters, as given in Eq. (1) where V w (t) represents the wind speed, V wa (t) is the average value of wind speed, V wr (t) is the ramp component, V wg (t) is the gust component and V wt (t) is the turbulence component. The aerodynamic power P t developed by the turbine is represented in Eq. (2). This developed power depends on the rotor radius R, the wind speed V and the air density ρ [33].
The power coefficient C p depends on the blade angle β and the tip speed ratio λ and its empirical formula for a given turbine is expressed as The pitch control model was designed such that when wind speed increases above the rated value. The value of C p was adjusted to maintain the turbine power constant by varying the set value of blade angle β which was obtained from turbine characteristic. The torsion characteristics of the wind turbine and generator were analyzed from the shaft dynamic model as described in Eq. (4).
where τ d is the coefficient of wind turbine mechanical transmission system, T g is the generator electromagnetic torque [Nm] and T m is the aerodynamic torque [Nm].
The simulation model was developed for various components of the turbine in the DIgSILENT platform which was integrated with the PMSG model available in the library as shown in Fig. 1.
The Stator Voltage-oriented Reference Frame (SVRF) (d-axis aligned to the stator voltage vector) was selected to control the active and reactive power of the generator by using direct and quadrature axis currents respectively as expressed in Eqs. (5)(6)(7).
The electrical torque T e of the generator can be expressed as P gen is the generator real power pψ pm i sq Q gen is the reactive power

Model of Converter Control
The generator side PWM converter connects the PMSG to a DC bus bar in order to track the maximum power and maintain a consistent stator voltage. To control the active and reactive power import into the grid, a connection was made between the DC bus and the grid side PWM converter [34][35][36].

Generator Side Contoller
The stator voltage control is required to avoid overvoltage caused by wind turbine overspeeding. The stator voltage oriented reference frame, with the d axis aligned with the stator voltage, is preferred to keep the stator voltage at its rated value by generating correct reference current i qref . The generator side converter was used to adjust the generator's stator voltage V ac and output power P. Figure 2 depicts the generator side controller's control block diagram.
The PCM technique was used in MPPT algorithm to generate optimal refernce power P * when the speed of the wind reached below its rated value (typically 12 m/s) as given in Eq. (8).
When the wind speed falls below the rated value, an algorithm for tracking maximum power MPPT is used. The method takes the form of a P-ω table, which indicates the points of optimum aerodynamic efficiency. Based on the measured generator speed ω gen,meas , the reference power signal P ref was created. To extract huge power out from the wind, either the generator speed or the DC bus voltage can be employed. The reference currents i dref1 and i dref2 were generated using P ref and U dc , respectively. The maximum value of these reference currents was used to calculate the necessary reference current i dref , which was then used to calculate the control variable P md linked to the d-axis. The  V ac _ ref is compared to the measured V ac to create the i qref , which was then compared to the measured i qs to regulate the modulation index P mq associated with the q-axis. A rapid inner current loop controls the stator currents ids and i qs , whereas a slow outer control loop controls the actual power and stator voltage. Figure 3 depicts the grid side converter control approaches that preserve unity power factor on the grid. This converter regulates the reactive power Q grid and the DC link voltage U dc . The flow of real power P into the grid was governed by the U dc , which was determined by the d-axis component ids, while the flow of reactive power Qgrid was dictated by the q-axis component. The measured U dc was compared to Udc ref to yield i dg_ref , which was then compared to the measured i dg to control the modulation index Pmd associated with the d-axis. The measured Q grid was compared to the Q grid_ref to create i qg_ref , which was then compared to the measured i qg to regulate the modulation index P mq linked to the q-axis. Each axis has two control loops for the variables, the inner current loop being faster than the outer loop. Reactive power Q grid and DC bus voltage U dc are managed by the outer loops.

Grid Side Controller
At the generator side converter, a PI control loop with reference power P * as an input was used to generate reference current i* ds for the inner current control loop in the d-axis. A voltage PI controller compares stator voltage with the set reference value to provide the reference current i* qs for the q-axis current control loop. The grid side converter was designed to control the reactive power with quadrature axis component i qs derived from Q ref and the active power with i ds which was derived from V dc . The grid side converter can be operated to supply leading or lagging reactive VARs by setting the Q ref appropriately which help as to maintain the set voltage at the grid. It can be set to a zero value to keep the power factor unity at the grid. Therefore, variable speed wind generator generates the voltage fluctuation and flicker emission were enhanced by this control variables.

Model of the Fixed Speed Wind Electric System
The fixed speed wind electric system has the models of wind, aerodynamics, shaft, pitch angle control, and the built-in electrical components such as induction generator, transformer, and grid. The simulation model was developed for each component based on their mathematical models using DSL in DIgSILENT. The components such as wind, turbine, pitch control, and shaft of the system is like the variable speed wind turbine. Hence, they are not elaborated on in this section. The models for the electrical components such as induction machines, grids, and transformers are taken from the built-in library. Therefore, the mathematical models of these electrical components are not discussed in detail. Figure 4 shows the integration of various simulation blocks developed with the built-in asynchronous machine block.

Statcom Model
STATCOM is a FACTS shunt device that can absorb or inject reactive power into the system. Figure 5 depicts the basic STATCOM model coupled to an ac bus system through a shunt coupling transformer. The compensating current in a STATCOM is independent of the system voltage, allowing it to run at full capacity even at low voltages. This distinguishes it from other devices such as SVCs. STATCOM is chosen as a source for reactive power support because it continuously chances its susceptance while reacting fast and providing voltage support at a local node. The outputs of the controller are i d _ref and i q _ref which are the reference currents in the dq coordinates which are needed to calculate the power injections by the STATCOM as in (9) and (10).

Network Model of the Wind Farm
The schematic layouts of wind farm substations were implemented in DIgSILENT software with the built-in power system components. Figure 6 shows aggregated simulation model of the fixed speed wind farm which consists of six 0.6 MW/0.69 kV wind electric generators. The starting transient current was reduced by connecting the soft starter to the generator busbar. All the wind electric generators were connected to the 110 kV busbar through 11 kV underground cables. A special care was taken in this wind farm substation, which directly feeds the power to the rolling mills.
The variable speed wind farm substation located at Dharapuram district consist of 14 feeders (E1-E14) of different rating. The E2 feeder was taken for the flicker analysis and it consists of 12 variable speed PMSGs with back to back converter, each rated for 0.8 MW at 0.44 kV. Figure 7 shows the aggregated simulation model of the variable speed wind farm (VSWF) [35]. The generated power was connected to the 22/110 kV substation transformer through underground to feed paper mill.

Measurement and Analysis of Voltage Variation and Flickering
The purpose of measurement of flickering was to assess the quality of power delivered by a wind energy conversion system (WECS) at the substation. Two types of wind farms were considered for the study namely, variable speed PMSG wind farm substation located at Dharapuram district and fixed speed wind farm substation at Coimbatore district in Tamil Nadu, India. The Fluke analyzer installed at the wind turbine busbar and the Dranetz analyzer was connected to the group control breaker of 110 kV feeder in the substation. The data was collected during the period between July and September as the wind flow is at its maximum during this period. According to IEC 61000-21 standards, the acceptable value of short-term flicker severity P st is less than 1 and the voltage magnitude variation should be within ± 10%. Figure 8 show the measured waveform of voltage and flicker severity P st at the VSWF substation for a duration of seven days. The red dotted lines in the graph indicate the voltage variation limit: the line at the top is for ± 10% variation whereas the bottom line indicates 90% limit. The measured voltage fluctuation most of the time lies within ± 10% limit. These voltage fluctuations were caused by the variations of wind speed and load leading to a small value of flicker severity P st which was less than one. The peak load demand, various faults and interruption can cause the bus voltage to fall below the 10% limit which results in an increase in the P st value more than 1.
The measured variations of voltage and P st at FSWF were shown in Fig. 9 for a duration of one week. The voltage produced by the fixed speed wind generator has large fluctuation exceeding 10%limit. The measured value of short term flicker emission P st may reach a value as high as 5 exceeding the unity limit. The P st value can be used for the voltage stability analysis: If P st is less than one, the wind farm is stable with minor disturbances. But, the system stability was affected when P st was greater than one.

Analysis of Flickering Emission
The short term flicker severity P st was calculated as given in Eq. (11) and it is dependent on the variation of wind speed, grid impedance angle and short circuit ratio [4,16]. The built-in flicker meter was used to calculate the P st value at any given interval. Table 1 shows the comparison of various parameters versus flicker emission values. P st is the flicker emission from the wind turbine (short term). S n is the rated apparent power of the wind electric system. S k is the short-circuit apparent power of the grid. Ψ k is the network impedance angle (30°, 50°, 70°, 85°). v a is the annual average wind speed.

Effect of Wind Speed
The simulation took into consideration the wind speed variation from 5 to 20 m/s with a short circuit ratio 20 times the individual generator rating and about 65° grid impedance angle. The short term flicker severity (P st ) of fixed and variable speed wind farms as a function of wind speed is shown in Fig. 10.

Impact of Short Circuit Ratio of the Grid
The grid strength depends on the short circuit ratio value. Figure 11 shows the P st value for varying short circuit ratio with grid impedance angle of 65º. For both the wind farms, when the short circuit ratio increases the flicker level decreases exponentially.

Effect of Gird Impedance Angle
Across a line, the change in voltage may be expressed as where the active and reactive power flows were represented by P and Q, and the resistance and reactance in the line were represented by the R and X and V represents the voltage at the terminal. The grid impedance angle Ψ k depends on the reactance and resistance of the line whereas power factor angle Ψ depends on real and reactive power. The changes in the voltage ΔV can also be calculated as given in Eq. (13) and it depends on grid impedance angle and power factor angle.
The calculated P st value for varying grid impedance angles is shown in Fig. 12. As the grid impedance angle increases, the flicker level decreases and reaches the minimum value at 65º.

Simulation and Study of Flicker Severity for Fixed and Variable Speed Wind Electric System
Faults, Sudden load variations and interruptions have impact on the voltage of the fixed speed wind generator. Due to changes in the voltage, the reactive power absorption from V g cos ψ cos ψ k the grid and P st value were affected. Therefore, the events like faults, sudden load variations and interruption were considered for evaluating the P st value in the wind farms for the specified variation of wind speed, short circuit ratio and grid impedance angle. As per IEC standard 61000-4-15, a built-in flicker meter available in DIgSILENT was used for the calculation of P st value during continuous operation. The RMS transient simulation was executed for 3600s (2 h). During simulation each event was applied successively for a period of 600s to evaluate the voltage changes and the value of flicker severity P st . The sequence of events considered in the simulation was: (1) An addition of 3.6 MW load at 601s.
(2) A three phase symmetrical fault with fault reactance value X f = 0 is applied to the 0.69 kV generator busbar at 1201 s for the duration of 100 ms. (3) A unsymmetrical fault L-G was applied to the generator busbar at 1801s for the duration of 100 ms. (4) The interruption may be due to the tripping of generator for low wind speed or manual disconnection of the generator. This event was applied at 2401 s for duration of 200 ms.
These events were applied to the simulation model of fixed speed wind electric system as explained in Sect. 4.   Figure 13 shows the variation in the RMS voltage waveform recorded during the simulation period. The measured Pt value by the built-in flicker meter during the simulation is shown in Fig. 14.
From Figures, it was observed that the increase in P st value depends on the percentage of voltage drop from the nominal value. The value of P st was nearly 1.4 for 12% drop in the voltage due to event1.A high value of P st equal to 5.4 was measured for 80% of voltage drop due to event 4. During the faults, the P st value lies between 3.5 and 4. The simulated values of P st for the events 1-4 were matching well with those recorded values during the field study. The developed model was therefore validated and the assumed events 1-4 were possible causes for emission of high values of P st in the wind farm under consideration.
To reduce the measured value of P st , a shunt compensation device like STATCOM was connected at the 11 kV bus bar in the simulation model of FSWF to control the reactive power flow. In Fig. 4, the red rectangular block indicates the simulation model of STATCOM. The bar chart given in Fig. 15 shows that the values of P st with STAT-COM were much less than the values of P st obtained without STATCOM.
The events 1-4 are applied to the simulation model of variable speed wind farm as explained in Sect. 4. Figures 16  and 17 show the variation in the RMS voltage waveform recorded and the measured P st value by the built-in flicker meter during the simulation respectively.
It was observed that the voltage fluctuation and the flicker emission in the VSWF were less than those that of the values measured for FSWF. The grid side converter was designed to control the reactive power flow in the busbar to reduce the voltage fluctuation. During the events 1-4 the grid side converter injects sufficient lagging VAR into the grid to keep the voltage variation within the specified limit. Hence, the flicker emission simulated values of VSWF with grid side converter controller were improved when compared with the measured values and also the performance are highly comparable with doubly fed wind generator [37,38].

Conclusion
Wind farms in Coimbatore district, Tamil Nadu have identified several events including flickering emissions, frequent generator tripping and power evacuation events caused by a weak grid. In the proposed study, one of the power quality issues such as flicker emission was considered for the analysis. The simulation models of fixed and variable-speed wind farms were developed with the controller in the DIgSILENT to evaluate the flicker level.
The power quality analyzers were installed at the individual wind turbine and the control breakers at the substation to record the short-term flicker severity value, P st as per IEC standard. Hence the measured P st values were exceeding the standard limit. With help of flicker meter tools, P st values were evaluated for various events such as such as faults, interruptions, and sudden load variation. The simulated values of P st match well with the recorded values during the field study. The developed model is therefore validated, and the assumed events were the possible causes of high flicker emission in the wind farm due to its power electronics-based industrial drives failing to operate. To bring the flicker value below the IEC standard limit, a compensation device, STATCOM was connected at the PCC in a FSWF, and a grid-side controller was effectively used for VSWF. For some events, the flicker values were not brought below the IEC standard.
According to the study, the FSWF emits large shortterm flickers because of frequent sags caused by load variation, faults, and interruption. Hence there was a need for developing advanced schemes to reduce flicker emissions. The measured readings help the researchers to improve the design and hence the manufacturing quality of high power rating renewable energy systems. This research improved our understanding of the causes of flicker emission caused by various events. Its goal was to reduce system failures and security threats while also assisting in meeting energy demand and energy optimization by improving cost-effectiveness and environmental concern.