当前位置:首页 >> 电力/水利 >>

DESIGN AND FIELD TESTING OF A SOURCE BASED PROTECTION RELAY


818

IEEE Transactions on Power Delivery, Vol. 14, No. 3, July 1999

DESIGN AND FIELD TESTING OF A SOURCE BASED PROTECTION RELAY FOR WIND FARMS
S J Haslam
P A Crossley N Jenkins Member IEEE Senior Member IEEE Manchester Centre for Electrical Energy (MCEE), UMIST, PO Box 88, Manchester, M60 lQD, United Kingdom (UK)

Abstract:The paper describes the design and field testing of a soufce based relay suitable for the protection of wind farms with fixed-speed induction generators. The relay provides short-circuit protection for the medium voltage (MV) power collection circuit and the MV and low voltage (LV) windings of each generator transformer. The use of the relay allows a reduction in the construction cost of a wind farm, since MV fuses are no longer required to protect the generator transformers. The paper discusses how the protection algorithm was designed based on the electrical behaviour of induction generators operating in a wind farm. The relay was evaluated at a wind farm in Wales and also in the laboratory using data from a wind farm simulator and a transient recorder. The latter, installed at the wind farm, recorded the voltage and current signals measured during local disturbances. The recorded or simulated data was subsequently replayed into a relay simulator which graphically describes the operating behaviour of the protection.
Keywords:

of a power collection circuit nearest to the utility network (referred to as the source end [l]), provides short circuit protection for the collection circuit and the MV and LV windings of each generator transformer. Consequently, MV fuses and the ring main units that hc/use them, are no longer required to protect these transformers [ 2 ] . This allows a significant reduction in the construction cost of a wind farm. However, all generator transformer faults are now cleared by tripping a collection circuit. This is considered acceptable since the probability of such a fault is extremely low.
The relay also detects faults on the LV cables that connect the generators to their transformers. However, it is advisable to continue to use LV fuses to clear this type of fault and to only use the relay as time delayed back-up protection. This prevents unnecessary tripping of a collection circuit, when the fault can be easily cleared by an inexpensive LV fuse that only disconnects a single generator. WIND FARM PROTECTION

Source Based Protection, Wind Farms, Field Testing, Multi-FunctionRelay

INTRODUCTION Wind farms in the U.K. normally consist of multiple horizontal axis, LV wind turbine induction generators (WTGs) connected via individual generator transformers to one or more MV power collection circuits. Each generator transformer is protected locally by an MV fuse housed in a ring main unit. This approach is identical to the standard method used in the UK for the protection of conventional radial distribution circuits. The MV fuses must operate for low current faults on the LV winding and also safely clear high current faults on the MV winding.
A new type of relay is described in this paper that can be used for the short circuit protection of a wind farm based on fixed-speed induction generators. The relay, installed at the end
PE-346-MPO-10-1998 A paper recommended and approved by the IEEE Power System Relaying Committee of the IEEE Power Engineering Society for publication in the IEEE Transactions on Power Delivery. Manuscript submitted June 30, 1998; made available for printing November 10,1998.

The source based protection relay uses a shaped directional operating characteristic to discriminate between “in-zone’’ faults, normal operating conditions and faults and disturbances on the utility network or other collection circuits within the wind farm; refer to Fig 1. The main operating region is based on the circular power characteristic of an induction generator [3] and its associated transformer. If the wind turbines connected to a collection circuit are assumed to be identical, which is the case for most wind farms in the U.K.,a normal generating, region can be defined for each power collection

I

I

Figure 1: Argand Representation of the Source Based Protection for Wind Farms Utilising Two-Speed Induction Generators

0885-8977/99/$10.00 0 1998 IEEE

819 circuit [4,5]. The protection uses a voltage polarised positive phase sequence current (IPPPS) detect an abnormal operating to condition. This can be expressed mathematicallyusing (1):The location and shape of the alarm region and the time delayed and instantaneous operating regions w r developed ee with the aid of a power system simulator that zmdelled the transient behaviour of wind farms during short-circuit faults [4, 6 . The protection detects and discriminates between normal 1 and abnormal conditions based on the relative position of Ippps. For an in-zone MV fault, lppps lies in the instantaneous operating region and for an in-zone LV fault, lppps lies within the time delayed operating region. During an out-of-zone fault or a utility network disturbance the current phasor can transiently enter the alarm region. However this only activates an alarm, it does not initiate the tripping of a circuit breaker.

where the polarising quantity is the positive phase sequence voltage (Vpps). For an uncompensated wind farm generating normally, lpps lags the polarising voltage by >goo. This corresponds to a normal generating region (NGR) lying in quadrant 3 of the argand representation; refer to Fig. 1. The shape of the NGR is dependent on four primary factors: the number of WTGs in the protected zone; their generation capacity; the amount of reactive power compensation (RPC); and the operational range of the wind farm voltage. Although these parameters vary for different wind farms, the boundary between the NGR and operation during in-zone disturbances can be defined using a single straight line [4]. For wind farms using WTGs with a pole switching capability (i.e. allowing generation at two rotational speeds), the boundary consists of two lines. This is shown by the c,omplex current distribution in Fig. 2, where the NGR corresponds to a collection circuit with fourteen 100/300kW WTGs. The deviation in the straight line boundary is due to the additional reactive demand from the WTGs when they generate at the lower power rating. The NGR shown in Fig. 1 incorporates the two-speed WTG boundary and is encapsulated by the stable zone of the protection in quadrant 3. To compensate for transient variations in the wind farm voltage, the stable zone is slightly larger than the NGR for added stability. In quadrant 1, the white rectangular region of the stable zone prevents the protection operating if reactive power is being exported to the utility network when the WTGs are not generating. Similarly, the white rectangular region in quadrant4 keeps the protection stable during the start-up of WTGs, when there is little or no active generation on the collection circuit.

THE FIELD TRIAL
In January 1996, the wind farm protection (WFP) relay was installed at the Cemmaes wind farm in Wales; see Fig. 3. The main purpose of this site trial was to confirm, over a monitoring period of 12 months, that the relay remained stable during all types of disturbances that originate from the utility network. It was recognised that the probability of an in-zone fault was extremely low, which precluded any verification of the relays sensitivity. As expected, no in-zone faults did occur. The 7.2MW dual circuit wind farm extends across a 4km elongated plateau, 4OOm above sea level. It contains twenty four fixed-speed, horizontal axis, twin bladed wind turbines. Each is pitch controlled and can generate at either high speed (300kW) or low speed (1OOkW) depending on the wind intensity. The WTGs use 500kVA transformers to step-up from the 0.66kV generation voltage to the l l k V collection circuit voltage. Each WTG has 60kVAr of RPC. The two collection circuits contain fourteen and ten wind turbines respectively and an 11/33kV, 7.5/15MVA transformer connects the wind farm to a 33kV radial distribution feeder on the utility network. The transformer is fitted with a 33kV *lo% tap changer. The 33kV fault level at the point of coupling to the network is 78MVA and the X/R is -2.0. At the wind farm substation, see Fig. 3, the main l l k V circuit breaker cubicle is equipped with a protection scheme designed to protect a U.K. utility network from the effects of embedded generation. The scheme, referred to as the G59 relay, must satisfy the U.K. Engineering Association G59/1 recommendations [7]. In addition to the G59 relay, the l l k V circuit breaker on each collection circuit is protected using nondirectional phase and earth fault overcurrent relays. Each generator transformer is protected using an LV and an MV fuse. The MV fuse is housed in a ring main unit. The WFP relay contains software that implements the WFP algorithm and conventional PPS undedover voltage, NPS overvoltage and undedover frequency protection. During the site trial the WFP relay operated in a monitoring only mode, i.e. it was not allowed to initiate tripping of a circuit breaker.

-0.01

-

j-0.02.

8

n

P -0.03-0.04
*

1
b .

B
p)

-0.05-0.06-0.07." 4 . 0 8 L . ~ 4.3 -025

.

-02 -0.15 4.1 Active Current Import (kA)

-

-0.05

0

Figure 2: Complex Current Distribution of 14 100/300kW Generators

820

THE CEMMAES WIND FARM

- Fixed-SpeedGenerators 300/100kW r b m 24 Wind T u

- Pitch Regulated - 6OkVAr Localised RPC
33111kV
Utility Network

I$,
le$
*IT&

Collection circuit-2

V-y-y-W-Y-V-V-V-Vl
48$
1 0 4

2 0 4

21&

2 2 4

23&

24$

\

Normally

Open Point

-y--y-y-v04%
I

Collection Circuit-1

&

Y

Figure 3: Schematic Representationof the 7.2MW Cemmaes Wind Farm Including the Location of the WFP Relay and the DAQ System

To maximise the benefits from the site trial, all the protection functions in the relay were designed to operate under conditions defined by ALARM or TRIP settings. The former, which activated an alarm contact, was more sensitive than the latter. This allowed the stability of the various protection functions to be assessed and iteratively optimised during the period of the trial. To monitor the power system voltage and current signals, and to record the operating performance of the WFP relay, a data acquisition (DAQ) system was installed at the substation; see Fig. 3. This consisted of:- an industrial PC with a 12 bit DAQ card; a signal conditioning unit; an unintermptible power supply; and a marshalling unit. The alarm and trip contacts of the relay were used to trigger the DAQ system, which then recorded each of the analogue signals at a sample rate of 1kHz over a data window of 4.0s or 10.0s. PERFORMANCE ASSESSMENT OF THE PROTECTION During the 12 month evaluation period, in excess of 200 events were recorded by the DAQ system. These mainly consisted of disturbances such as voltage dips, momentary interruptions, sustained under-voltages and occasionally loss of mains. Although the majority of the disturbances were of unknown origin, the nature and location of a few were established, three examples are described in the paper. To aid the performance assessment of the relay, a graphical replay facility was developed. This allowed data, recorded at the wind farm or from a simulator in the laboratory, to be plotted on an argand diagram that represents the operating behaviour of the wind farm protection, see Fig. 1. The

trajectory of the voltage polarised PPS current phasor is displayed using a locus with time increments of one cycle (2Oms at 5 H ) 0z.

(i)A Sustained Under-Voltage Disturbance
The first example is a sustained under-voltage disturbance resulting from a single phase fault on the utility network; see Fig. 4. At fault inception, the monitored phase-currents exhibit a transient response which persists for -4OOms. The undervoltage protection in the G59 relay then trips the main 11kV wind farm circuit breaker. This results in an increase in the voltage imbalance seen on the utility network. The graphical replay of this disturbance is shown in Fig. 5. The SCALE: 0.150kA (1.5,l.O) corresponds to the size of the x and y axes in quadrant 3, the x-axis is 1.5 x 0.150kA 0.225kA and the y-axis is 1.O x 0.150kA = 0.1 5kA.

-

Prior to the disturbance, the locus remains in the stable zone of the protection. At fault inception, the PPS current phasor transiently enters the alarm region. This corresponds to reactive power flowing from the WTGs to the fault. The effect is brief and the locus re-enters the stable zone, initially moving towards quadrant 4. After a short period (0.34s) the trajectory changes direction. This change is related to the acceleration of the active WTGs, caused by the under-voltage condition and the drive train dynamics of the wind turbines. The new trajectory moves away from quadrant 4 and further into quadrant 3. This direction persists until the wind farm is disconnected from the utility network, forcing the phasor to the origin. Significantly, the PPS current phasor does not enter either the time delayed operating region or the instantaneous operating region, indicating that the protection remains stable during out-of-zone faults close to the monitoring position.

821

g P
0

10.00 5.00 0.00 -5.00 -10.00 I 0.00
I

= f
z

4

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.8C

Time (seconds)

g
0

02 .5

E
Q)

00 .0
-0.25
-0.50 ! 0.00
I

f

8

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.8C

Time (seconds) Figure 4: Instantaneous Phase Voltage and Current Signals at the Cemmaes Wind Farm During an Asymmetric Voltage Disturbance

(ii) Magnetising In-Rush Condition Fifteen minutes after the operation of the G59 relay, as shown in Fig. 4, the relay auto-recloses the main llkV wind farm circuit breaker and reconnects the wind farm onto the utility network. Fig. 6 shows the effect on the monitored aphase current of the in-rush flowing into the generator transformers immediately after reconnection. Fig. 7 describes the operating behaviour of the WFP relay during the in-rush. When the main llkV circuit breaker is reclosed, the PPS phasor enters the time delayed operating region for 120ms, and then returns to the stable zone. The protection remains stable, without the need for harmonic restraint, because the operating time in this region is one second.

g
2
U)

0.10 0.05 -0.00 -& -0.05 --

,

I

2 -0.10 -n
-0.15 --0.25q
I I ! I

; -0.20 -I
I
I I

I

I

I

I

I

0.00

0.10

0.20

04 .0 Time (seconds)

0.30

0.50

I

Figure 6 In-Rush Current Recorded at the Cemmaes Wind Farm

Figure 5: Replay of a Sustained Under-Voltage Netwolk Disturbance

Figure 7: Replay of the In-Rush (figure 6)

822

(iii) Loss of Mains Condition Fig. 8 shows a symmetrical disturbance which resulted in the loss of the main supply to the wind transient 33kVfault caused an initial non-rectangular voltage dip which for -4 cycles @Oms at 50Hz). During this period, contribution from the wind farm is substantial sinc WTGs were active prior to disturbance. However, d location and symmetrical nature of the fault, the phase-currents decay rapidly due to the lack of excitation in the WTGs. At time 0.18s, the initial disturbance is cleared and the voltage O m. progressively recovers over a period of 1 O s The response of the phase-currents is again transitory. Before the voltage recovers to its pre-fault level, the majority of the active WTGs are disconnected from the collection circuit using their own individual protection units and local circuit breakers. Although not discernible from the voltage waveforms, the shutdowns were triggered by a frequency excursion which exceeded 5 1Hz. With only a single active WTG on each collection circuit of the wind farm, the voltage gradually decreases until at time 0.6s the rate of change of frequency protection (ROCOF) in the G59 relay operates, opening the main 33kV and l l k V circuit breakers, isolating the wind farm from the utility network. The trajectory of the polarised PPS current locus during the loss of mains condition is shown in Fig. 9. Similar to the out-ofzone asymmetric fault in Fig. 5 , the phasor initially enters the alarm region of the protection at the fault inception. It then moves back towards the stable zone, but before it can enter this region the fault is cleared. This forces the locus into the time delayed operating region. Again the movement is transitory and the PPS phasor moves back towards its pre-fault location. The shutdown of the WTGs then forces the locus to move towards the origin of the Dhasor dot. However. the ohasor remains in
9
10.0
I

Figure 9: Replay of the Loss of Mains Condition quadrant 3 of the stable zone for -320ms indicating active generation. At time 0.6s, the wind farm is isolated from the network and the phasor moves to the origin of the plot. Although, the locus enters the time delayed operating region for looms, the time delay of the region ensures the protection remains stable. SENSITIVITY OF THE WIND FARM PROTECTION To test the sensitivity of the protection in the laboratory, a relay test set and data from a transient power system simulator were used to investigate the behaviour of the WFP relay. The
simulator generated the voltage and currents signals that would

I

be monitored at the source end of a collection circuit during inzone LV and MV faults. Fig. 10 shows the performance of the protection during the replay of five in-zone faults simulated using a detailed model of the Cemmaes wind farm. Each fault is identified by its associated number: 1. a single phase MV fault with 14 active 300kW WTGs. 2. a three phase LV fault with 1 active 300kW WTG. 3. a three phase LV fault with 14 active 300kW WTGs. 4. a resistive single phase LV fault 1 active 300kW WTG. 5. a resistive single phase LV fault 14 active 300kW WTGs. Note, each WTG operates at rated power during the faults. The trajectory for the MV fault (1) enters the instantaneous operating region 4Oms after fault occurrence, it then remains in this region until the fault is cleared. For a three phase LV fault, with one turbine (2) and 14 turbines (3) active, the trajectories enter the time delayed operating region 2Oms after fault occurrence. The trajectories for resistive single phase LV faults (4 & 5) enter this region 4 m after the fault occurs. In all O s cases, the LV fault trajectories then remain in the time delayed region until the fault is cleared (1.Os after fault occurrence).

g -10.0 I

............ -.
I
I

I

I

I

I

I

I

I

I

I

I

I

I

I

c
0.80

0.00

0.10

0.20

03 . 0 0 4 0.50 .0 Time (seconds)

0.60

07 .0

-0.604 I I I I I I I I I I I I I I I I 0.00 0.10 0.20 0 3 . 0 0 4 0.50 0 6 .0 . 0 0.70 0 8 .0 Time (seconds)

Figure 8: Phase Voltage & Current During a Loss of Mains Condition

823

and Application’; NEE, February 1954, pp. 12-19 a Multi-Function Protection Relay - Its Design and Site Trial Experience’, 14th International Conference on Electricity Distribution (CIRED), IEE Conference Publication No. 438, June 1997, pp. 4.22.1-4.22.5 F. Santjer; ‘Synchronisationvon Netzgekoppelten Windenergieanlagen in einem Windp of Grid Connected Wind Energy Converters in a Wind Farm’; DEW1 Magazin No. 7, August 1995, pp. 80-86 (published in German) S.J. Haslam, P.A. Crossley, N. Jenkins, M. Burt, A. Borrill; ‘Design and Evaluation of a New Type of Protection for Wind Farms’, 6th International Conference on Developments in Power System Protection, IEE Conference Publication No. 434, March 1997, pp. 99-102 Engineering Association; ‘EngineeringRecommendation G.59/1 - Recommendations for the Connection of Embedded Generating Plant to the Regional Electricity Companies Distribution Systems’; 1991 ACKNOWLEDGEMENTS CONCLUSIONS A new type of relay suitable for the protection of wind farms has been described. The relay provides short circuit protection for a power collection circuit and all the generator transformers connected to it. The operating performance of the relay was assessed during a 12 month site-trial on a wind farm in Wales and also in the laboratory using a wind farm simulator and a relay test set. In addition, a relay simulator was used to determine whether the relay ever moved close to its operating thresholds when installed at the wind farm or when being tested in the laboratory using simulated “out-of-zone” faults. The simulator was also used to confirm that the relay would operate correctly on all types of “in zone” faults. The results in the paper support the main design objective for the new relay, i.e. it must remains stable during normal operation, non fault disturbances and out-of-zone faults, and be sufficiently sensitive to operate correctly on all realistic in-zone faults. REFERENCES J.V.H. Sanderson, M.M.M. Mahmoud, H.T. Yip, J.T. Hampson, J.J. Hunt; ‘Remote Detection of Distribution Transformer Faults on 11kV Radial Distributors’; International Conference on Revitalising Transmission and Distribution Systems, IEE Conference Publication No. 273, 1987, pp. 161-165 British Electricity Boards; ‘Report on HV Fuse Links for the Protection of Ground Mounted Distribution Transformers’; Report No. 86, 1983 J.E. Barkle, R.W. Ferguson; ‘Induction Generator Theory The authors would like to acknowledge the support of Scottish Power and the Energy Technology Support Unit as part of a DTI research initiative. We also wish to thank Wind Energy Group and National Wind Power for allowing the field test at the Cemmaes wind farm, and Omicron Electronics for the provision of a relay test set. BIOGRAPHIES
Steven J Halsam MEng, PhD, AMIEE was born in Cheshire, UK in 1971. He completed his PhD degree at UMIST in December 1997 He currently works as a technical assistant at CAE Electronics, Montreal, Canada, developing energy management systems. Peter A Crossley BSc, PhD, CEng, MIEE, MIEEE was born in Lancashire, UK in 1956. After 13 years with GEC ALSTHOM he moved to UMIST where he is now a Senior Lecturer specialising in power system protection.

S.J. Haslam, P.A. Crossley, N. Jenkins, M. Burt, A. Borrill; ‘Source Based Protection of Wind Farms using

Figure 10: Replay o SiTulated LV and MV In-Zone Faults f

Nicholas Jenkins BSc, MSc, DIC, PhD, CEng, MIEE, MIEEE was born
in Essex, UK in 1954. He is now a Senior Lecturer at UMIST specialising in the power system problems resulting from renewable or embedded generation.


赞助商链接
相关文章:
更多相关标签: