MEASUREMENTS AND SIMULATION FOR POWER QUALITY OF A WIND FARM

Transcription

1 Research Article MEASUREMENTS AND SIMULATION FOR POWER QUALITY OF A WIND FARM 1 S.P. Shukla, 2 Sushil Kumar Address for correspondence 1 Department of Electrical Engineering, Bhilai Institute of Technology, Durg, Chattisgarh (India) sps_bit@rediffmail.com 2 Department of Electrical & Electronics Engineering, Bhilai Institute of Technology, Durg, Chattisgarh sk1_bit@rediffmail.com ABSTRACT This work investigates the impact of windfarm on the distribution network power quality. The quality of wind power is decided by IEEE Standard and IEEE Standard The experiments were carried out at the Vankuswade wind power project, Satara, (M.S.), India. The results are then compared by MATLAB simulation software (Version R2008a). The system is also simulated for system stability considering various faults on the simulated model. The field measurement results give important indications about the real effects of the integration of large interruptible renewable energy sources within the power system. The simulation results show an agreement with the measured response. KEY WORDS Wind power, Power Quality, Voltage Sag, Voltage Swell, Stability. INTRODUCTION Wind power generation has experienced a very fast development worldwide mainly due to environmental reasons. As the wind power penetration into the grid is increasing quickly, the influence of wind turbines on the power quality is becoming an important issue 1,2, 11. The generation of wind power occurs with the operation of multiple wind turbines in windfarms. The integration of these wind parks into the power system may cause power quality concerns. The important issue is how much the power quality will be affected by power production and connection of WTs to the grid. The purpose of this work is (i) to analyze the power quality in the electrical vicinity of a wind turbine at low voltage levels, and (ii) the interaction of wind farm with the power system at the point of common coupling with the grid at medium voltage level (Fig.1) 8. Slow voltage variations, flicker, voltage sags, transients and harmonics are measured by means of a modern digital measurement system. Important measuremed results are discussed. Then a wind farm consisting of Doubly Fed Induction Generators connected to the distribution network is simulated for various normal and abnormal conditions. The two results are then analyzed.

2 Fig. 1. Wind turbine with DFIG. Wind Farm Site Description Measurements are carried out at Asia s largest wind farm installed at Vankusavade/Chalkewadi sites in Satara district and Gudhepanchgani in Sangli district of Maharashtra State (India). The wind electric- generator under study is BHEL/NORDEX make, 3-Phase, 415 Volt, 82/250 kw, 82/424 Amp, 8/6 Pole, 30/40 Turbine RPM, 757/1008 Generator RPM, - connected asynchronous machine. The wind turbine is horizontal axis and installed at 30m hub height. There are 8 nos. of machines installed at the 2 MW demonstration wind power project of Maharashtra Energy Development Agency (MEDA) at Chalkewadi. The generator is connected directly to the grid through 0.415/11 and 11/33 kv step-up transformers as shown in Fig. 2. Three switching capacitor banks (two of 37.5 kvar and one of 25 kvar) are connected across the generator (i.e. on L.V. side) to compensate the reactive power drawn from the grid. Capacitors are not provided on H.V. side [2]. The site has good power availability with an average wind speed above 5m/s. Each machine is active yaw and pitch regulated with power/torque control capability. Experiments are performed at a wind farm sites to study the interaction of wind turbine (induction type) generator with the utility grid. A DFIG has a wound rotor that is connected to the network through a pulse-width modulated IGBT frequency converter which controls the excitation system in order to decouple the mechanical and electrical rotor frequency and to match the network and rotor frequency (Fig. 1). The wind turbine rotor is coupled to the generator through a gearbox which adapts the two different speeds of rotor and generator. The control system usually keeps the power factor to unity, but the windfarm can also exchange reactive power with the rest of the network.

3 Fig. 2. Single line diagram of the test wind farm at Chalkewadi. Power Quality Measurement Set-Up and Methodology When power quality measurements are performed on a system with wind production, due to the presence of electronic devices and frequency converters, the voltages and currents are usually nonsinusoidal quantities. In this application the measurement system has to be carefully chosen and, in particular, it has to be composed by transducers with high bandwidth, analog conditioning blocks, analog to digital converters a digital signal processing and, a storage unit. Low Voltage measurement system The following hardware components have been used: Power monitoring instrument Dranetz PX5, 8 channels, 4 voltage and 4 current, 256 samples/cycle, RMS Accuracy: ±0.1% of Reading, ±0.05% Full Scale, over 7KHz bandwidth, measures flicker according to IEC , complies with IEEE 1159, IEC Class A and EN50160 [1,5]. Instrument voltage probes that can measure from 1 to 600V. Three current probes LEM FLEX RR3035, current ranges of 30/300/3000A, bandwidth: 10 Hz to 50 khz, Accuracy: 1%. The measurement system was placed at the wind turbine terminal, between the generator and the LV side of the step-up transformer according to a 3-Φ, 3 wire star connection. Three channels have been connected to voltage and current probes. The neutral has been connected to common and has been the reference for the three channels in order to measure the phase-to-neutral voltages. A simplified diagram with only one phase is provided in Fig. 3. The measurement system permits measuring the currents and voltages instantaneously. Parameters configured for the measurement are: mean, maximum and minimum, RMS voltage and current values each minute; mean, minimum and maximum active and reactive power and power factor each minute; total harmonic distortion and individual harmonics calculated each minute; flicker measurements, calculated as per IEC

4 15, with P st (short term) interval equal to 10 min., P lt (long term) interval equal to 2 hours. Fig. 3. LV measurement set-up. Medium Voltage measurement system The following hardware components have been used: Power monitoring instrument Dranetz PX5. One PEARSON ELECTRONICS VD-305A capacitive voltage divider, nominal division ratio 2000:1, Maximum Pulse Voltage: 300 kv, bandwidth: 30 Hz to 4 MHz. Current probes LEM FLEX. In order to analyze the collective behavior of the wind power plant the measurement system has been placed at the point of common coupling (PCC) with the grid according to a single phase connection. A simplified connection diagram is provided in Fig. 4. Measurement Results and Discussion Voltage waveforms analysis is equal to 2.77% 3, 9. The harmonic emissions are quite high, as the measurements have been done between the generator and the l.v. side of the transformer. On the high voltage side of the transformer a reduced harmonic distortion level it is expected due to the damping features of the transformer. Distortion levels are usually high in DFIG turbines due to the frequency converter, depending on the commutation frequency. The distortion caused by the converter can be clearly observed in Fig. 5. Although harmonic distortion is an important issue but due to the high switching frequencies, the advanced control algorithms and the filtering techniques used in the wind farm allows reducing the distortion well down the maximum The voltage waveform measured at the value tolerated by standards 11. generator terminals is shown in Fig. 5. The THD

5 390.7V whereas at no load has been about 401V. Fig.5. Phase-to-neutral volt. at gen. Terminals. Fig.6 Phase-to-neutral volt. at PCC. Fig. 6 presents the voltage measured at the PCC between one phase and neutral, the THD is equal to 0.93%. The low distortion is due to the combined smoothing effect due to the aggregation of the generators. Compared with maximum harmonic levels for the power system specified by EN 50160, even though this standard is not applicable to this context, these values of THD are largely within the limits. Voltage Variations with Power Produced It has been observed a voltage variation at the terminals of the generator depending on the power generated. Fig. 7 shows the trend of the active power and voltage RMS for a time period with high power production At full load the phase-to-neutral voltage RMS has been about Long-term Voltage Variation Analysis The measured RMS voltage, under normal operating conditions, excluding situations arising from faults or voltage interruptions, during all observation intervals has been within the range of ± 5% of the nominal voltage. Voltage sags and swells PQ analyzer was set to register transients when thresholds values exceeded. Voltages below 95% and above 105% of the RMS nominal value have been recorded. Fig. 8 shows voltage sag caused by the WT disconnection from the grid due to excessive power production (overload condition). During the measurement, 18 voltage sags were registered at the generator terminal but no voltage sags were recorded at the PCC. Fig.7 Voltage variation compared with power generated.

6 Fig. 8 Voltage sag at the generator terminal. Flicker The torque from a horizontal axis wind turbine has a periodic component at the frequency at which the blades pass the tower (1-2 Hz) caused by a variation of the wind speed seen by the blade. Such variation depends on the combination of the tower shadow, wind shear and turbulence. The torque fluctuations are directly translated into output power flicker as there is only a partial buffer between mechanical input and electrical output. Fixed speed wind turbine can cause high flicker whereas variable speed one can limit the flicker within reasonable values 7. In the proposed measurement campaign the Flicker has been analyzed at the LV connection of a single WT and at the PCC of the wind farm. The measurements at the LV terminals of the wind turbine showed that the flicker level (Pst) is correlated with the active power produced by the generator. In particular the flicker level increases with the production and remain about constant even though the power changes as depicted in Figs Fig.10. Flicker measured at the generator terminal. Fig.11. RMS voltage at the PCC. Fig.9. Power generated by a single turbine. Fig.12. Power generated one phase at the PCC. The flicker at the PCC depends on the power fluctuations caused by all the turbines of the wind farm are illustrated in Figs Measures have revealed that the flicker

7 increases as the power produced decrease. In Fig. 11, part of this flicker effect could be imputable to the regulation action of the on-load tap changer (OLTC) installed in the substation transformer as well as caused by switching operations of start and stop of wind turbines. The results are in good agreement with other contributions related to measurements in similar conditions 3,4. Simulation of a Wind Farm Using DFIG Wind Turbines Model Description A 9-MW wind farm consisting of six 1.5 MW wind turbines connected to a 25-kV distribution system exports power to a 120-kV grid through a 30-km, 25-kV feeder. A 2300V, 2-MVA plant consisting of a motor load (1.68 MW induction motor at 0.93 PF) and of a 200-kW resistive load is connected on the same feeder at bus B25. Both the wind turbine and the motor load have a protection system monitoring voltage, current and machine speed. The DC link voltage of the DFIG is also monitored. 4, 6, 11 MATLAB Simulation and discussion of the results Following conditions were applied on the MATLAB model: 1. Turbine response to a change in wind speed. 2. Simulation for stability due to LG fault on the distribution system (25kV). 3. Simulation of a voltage sag on the system CASE 1- TURBINE RESPONSE TO A CHANGE IN WIND SPEED Effect of wind speed variation is shown in fig 14a & b. It increases from 8m/s -20m/s in steps. The generated power also increases, at t = 32 sec. when V = 15.5m/s P Generated becomes 9MW i.e. rated capacity of the wind farm. A further increment in wind velocity causes voltage swell of 2.5% at the generator terminals a swell of 0.25% at the converter terminals (Grid). Voltage swells are eliminated by the pitching of the blades. The pitch angle increases from 0 degree as soon as the wind speed crosses 15m/s i.e. the speed for 1p.u. generation. (a) (b) Fig. 14. (a) V, I, P& Q at Gen. Terminals, (b) V DC, Turbine and Wind speed & Pitch angle.

8 Fig. 4. MV measurement set-up. Fig. 13 MATLAB model of DFIG connected to grid. CASE 2- depicts a large drop in voltage and SIMULATION FOR STABILITY DUE TO corresponding current. But the system regains VARIOUS FAULTS ON THE 25-KV steady state without losing stability when the SYSTEM fault is over. When the same fault was applied The LG fault is applied on phase A at t=5sec. for 9 cycles, it was observed that the windfarm for a duration of 5 cycles. Its effect on load gets tripped from the network and its voltage and current is shown in Fig. 15, which contribution (P, V & I) becomes zero as shown

9 in Fig. 16 a. This happens due to drop in generator terminal voltage below 0.9 p.u. Thus the stability of the system is lost. When LLG fault is applied at t = 37sec. for 5 cycles, the generated power suddenly drops to 37% from the rated value but comes back to steady state when the fault is over as shown in Fig. 17a and b. When LLLG fault is applied at t = 37sec. for 5 cycles, the windfarm trips immediately from the network and does not regain the stability. Case-3 Simulation of a voltage sag on the system Fig. 16 b depicts that effect of LG fault on voltage sag is much severe at the generator terminals (approx.30%), then less severe on the medium voltage distribution network (approx. 10%) and least on the high voltage grid i.e. at PCC (approx. 1%). During LLG and LLLG fault similar behavior was obtained but with deeper sag. CONCLUSIONS In this paper, the power quality of a large windfarm at the low voltage level and at the PCC with the HV grid is investigated. Flicker, harmonics, and voltage sags have been analyzed and correlated with wind characteristics of the site. Then same were observed on a simulation model. The simulated and the measured results exhibit a good agreement. The investigation also shows that the windfarm works without a negative impact on the transmission network when DFIG turbines are installed. (a) Fig 15.Load (500KW) Voltage and Current after a 5 cycle LG fault. (b) Fig. 16. ( a) V, I & P at Gen. Terminals (b) Voltage at Generator, Distribution, Grid terminals.

10 (a) (b) Fig. 17. (a) V, I & P at Gen. Terminals, (b) Voltage at Generator, Distribution, Grid terminals. REFERENCES 1. IEEE Recommended Practices and Requirements for Harmonic Control in Electrical Power Systems, IEEE Standard , pp.1-8, April P. Kundur, Power system stability and control, McGraw-Hill, Ake Larsson, Flicker emission of wind turbines during continuous operation IEEE Trans. Energy Conversion,Vol. 17,No. 1,pp , P. Sorensen, A.D. Hansen and P.A.C. Rosas, Wind models for simulation of power fluctuations from wind farms, Journal of wind Engineering and Industrial Aerodynamics, Vol. 90,pp ,December IEEE Standard 1547: Interconnecting Distributed Resources with Electric Power System, IEEE Standard , p.p. 1-5, June Yuriy Kazachkov and Steve Stapleton, Modelling Wind Farms for power system stability studies Power Technology, Newsletter Issue 95, April S.P.Shukla, B.S.Narang Harmonics in wind power systems Applied science periodical, Vol.IX, No. 2, pp , May, Trinh Trong Chuong, Dragan Voltage stability of grid investigation of grid connected wind farm PWASET, Vol. 32, August, S.P.Shukla et al Transient Control Techniques in Synchronous Wind Turbine Generators CSVTU Research Journal, Vol.2, No. 1, pp 46-50,Jan S.P.Shukla, B.S.Narang Quality and Stability Assessment in a wind power system International Journal of Engg. Research & Indu. Appls. (IJERIA), Vol.2, No. VI, October 2009,pp S.P.Shukla, B.S.Narang Power Quality and Stability Aspects in a Wind Power System Applied Science Periodical, Vol.XII, No. 4, November, 2010.