Sei sulla pagina 1di 6

Distributed Generation Impact on Voltage Sags in Distribution Networks

Juan A. Martinez-Velasco
Dept. Eng. Elctrica - ETSEIB Universitat Politcnica de Catalunya Barcelona, Spain

Jacinto Martin-Arnedo
Depto. Distribucin ITC2 LHospitalet de Llobregat, Spain

Abstract Characteristics of voltage sags caused by faults in distribution networks depend on the design of the protection system and the coordination between the different protective devices. The presence of distributed generation (DG) changes the radial nature of distribution systems, so it can affect the performance of the protection system, and consequently the characteristics of voltage sags. In addition, DG will help to maintain the during-fault voltages of healthy phases. This paper explores the impact that DG can have on voltage sags characteristics by means of a small test distribution network and presents a simple method to compare and rank the performance of protection systems. Keywords- Distributed Generation, Voltage Sag, Modelling, Power Distribution, Power Quality, Simulation.

lack of directional sensing of relays for radial systems, the islanded operation of DG, autoreclosing schemes, or breaker failures [4]. DG islanding is one of the main concerns [5] - [7], and although islanded operation is not generally allowed, DG may successfully operate in island if there is a balance between load and generation. Characteristics of voltage sags in distribution networks caused by faults are highly influenced by the placement of protective devices and the coordination between them. They are also affected by the presence of DG, although generators embedded in a distribution network can help to maintain during-fault voltages at the healthy phases. This paper explores the impact that DG can have on the characteristics of voltage sags considering the types of protective devices installed in the system and the coordination between them. The connections of the substation transformers and the DG interconnection transformers are important aspects that can affect the design of the different protection systems, as well as their coordination. Each transformer connection has advantages and disadvantages, and no universal agreement has been reached about the best selection. A thorough analysis of each connection is out of the scope of this paper. The goal of this paper is twofold. On one hand, it is aimed at analyzing the characteristics of voltage sags that can be caused in a distribution network with DG. On the other hand, it presents a simple method to compare and rank the performance of protection systems taking into account the probability of sensitive equipment (SE) trip. A model of a very small distribution network, including the protective devices of the network, has been created to analyze the DG impact on the characteristics of voltage sags caused by faults; its main features are detailed in Section II.

I.

INTRODUCTION

The growing penetration of distributed generation (DG) from renewable and traditional sources has raised some concerns about its impact on the operation of distribution systems. Although costs associated with DG technologies are still high, they can be a solution for those situations in which high power supply reliability is needed, or when the construction of transmission lines and large power plants is not supported by end-users [1] - [3]. The installation of DG sources raises new challenges to the operation of radial distribution networks. DG does not change the radial topology of these networks but the power will no longer flow in a single direction. This has an impact on the protection of distribution networks, and this impact depends on size, type and location of DG. According to a report of the IEEE Power System Relay Committee, the list of topics to be revised would include the
This work is supported by the Spanish Ministerio de Investigacin y Ciencia, Reference ENE2005-08568/CON.

II.

TEST SYSTEM

Fig. 1 shows the diagram of the test system used in this paper; it is a very small distribution network whose substation transformer is grounded at the lower voltage side by means of a zig-zag reactor, which limits the current caused by a singleline-to-ground fault to 800 A. Fig. 2 shows the time-current curves of the protective devices installed in the system. Although simulation results presented in this paper will be based on these curves, the study will also consider characteristic curves and coordination between protective devices different from those depicted in the figure. For modelling guidelines used to represent the test system see [8].

important aspect, and although its effect will not be very important on voltage sag characteristics, it cannot be neglected when analyzing the transient performance of the system. Rather than considering a single value for the rotor inertia, a wide range of values has been used in simulations. Since simulations will not be longer than 3 seconds, the effect on the primer mover will be neglected and a constant mechanical energy input will be assumed in calculations. To analyze the effect that voltage sags can have on SE, this will be represented by the ITIC curve. Remember that this curve is applicable to single-phase equipment, and only a twodimensional characterization of sags (magnitude and duration) is required.
TABLE I. ELECTRICAL SPECIFICATION OF THE SYNCHRONOUS GENERATOR Parameter Value 50 Hz 2.0 MVA 6.0 kV 4 0.0041 pu 0.1200 pu 1.7000 pu 0.2383 pu 0.1847 pu 3.1949 s 0.02872 s 1.4000 pu

L1
2.5 km 4.0 km

L2

M0
BR
5.0 km

M1

F1
3.5 km

F2

M2
2.0 km

M3
IP GP

Rated frequency
DG

Rated power Rated voltage Number of poles Armature resistance: Ra

F4

F3

3.0 km

4.5 km

L4

L3

Armature leakage reactance: Xl d-axis synchronous reactance: Xd d-axis transient reactance: Xd

HV equivalent: 110 kV, 1500 MVA, X/R = 10 Substation transformer: 110/25 kV, 8 MVA, 8%, Yd Distribution transformers: 25/0.4 kV, 2 MVA, 6%, Dy DG transformer: 25/6kV, 2 MVA, 8%, Yd Lines: Z1/2 = 0.61 + j0.39, Z0 = 0.76 + j1.56 /km BR = Circuit Breaker, F1, F2, F3 and F4 = Fuses GP = Generator protection, IP = Interconnect protection

d-axis sub-transient reactance: Xd d-axis open-circuit transient time constant: Tdo d-axis open-circuit sub-transient time constant: Tdo Zero-sequence reactance: X0

Figure 1. Diagram of the test system.


10

Vref + VG 25 1 + 0.05s
-3.5

3.5

1 + + EFD

1 1 + 0.1s

Fuses
1

Breaker

0.02

0.1s 1 + 0.5s

Time [s]

Figure 3. Block diagram of the excitation control.

III.
0.1

VOLTAGE SAG STUDIES IN RADIAL SYSTEMS

Iscmax

0.01 10 100 1000 10000

The first approach to the impact of DG has been based on a simplified analysis of the test system. The retained voltage during three-phase faults at the point of common coupling (PCC) must be estimated considering the test system topology. The PCC is defined as the point of the network where the SE current path is separated from the fault current path. Fig. 4 shows some of the various configurations that have to be considered in the test system. The expressions to obtain the retained voltage, with and without DG, can be deduced from the results presented in some references, see [9] or [10]. Zf is the positive sequence impedance of the line section between the PCC and the fault plus the fault impedance. From the parameters of the test system components, the following conclusions can be derived:

Current [A]

Figure 2. Time-current characteristics of protective devices. (Iscmax=Maximum short circuit currentAt secondary transformer terminals)

The only distributed resource installed in the test system is a synchronous generator, whose ratings and electric parameters are shown in Table I. The block diagram of the excitation control is presented in Fig. 3. No information is provided about the mechanical system of the generator, for which a singlemass representation will be used in simulations. This is an

S
Zs
PCC

tolerance is represented by the ITIC curve, only voltage sags caused at the LV side of load nodes will be presented and analyzed. With the connection used for distribution transformers there will not be swells at the LV side [8]. The study and the simulations have been carried out assuming that the circuit breaker opens the three poles, irrespectively of the short-circuit type, while fuses are of current-limiting type and open only the faulted phases. As for the coordination between these protective devices two possibilities are analyzed: fuse saving (fuses are slower than the breaker) and fuse blowing (time-current characteristics are those depicted in Fig. 2). A consequence of the protective device operation is that voltage sags will not be always rectangular, since the coordination between these devices can produce multiple events with different retained voltages. Characterization of multiple events (magnitude and duration) becomes an important issue [11]. The following paragraphs analyze the voltage sag characteristics caused respectively with fuse saving and fuse blowing. Fuse saving: The network is protected by the breaker only, and fuse models are not included in simulations.

Zf

SE

a) Without DG (PCC is M1 for L1/L4; PCC is M2 for L2/L3)

S
Zs Zi Zf

DG
Zg

SE

b) With DG, fault on lines L1 and L4, to be applied to L2/L3


Za
PCC

Zc

Zb

DG

SE

c) With DG, fault between M0 and M3, PCC is M1 for L1/L4 Figure 4. Equivalent for calculation of voltage sag magnitudes in the test system.

To obtain a during-fault voltage at the LV side of the load nodes (L1, L2, L3 and L4) of about 80% of the rated voltage, without DG in the system, the fault location should be no less than 10 km from the corresponding PCC, even when the fault impedance is a resistance of about 10 . Therefore, all symmetrical faults will cause equipment trip. The stiffness ratio Zg/Zs (the ratio between the shortcircuit impedances of the main source and the DG measured at the MV side of the corresponding interconnection transformer terminals) is about 30. Therefore, the above conclusion is also valid with DG; that is, there is no fault location with which the during-fault voltage at any SE location will reach 80% of the rated voltage. IV. SIMULATION RESULTS

Utilities can choose between a breaker instantaneous reclose or a delayed reclose. An instantaneous reclose will last several cycles, due to the mechanical inertia of breaker contacts, while a delayed reclose can last up to several dozens of seconds because ionized gases formed during the fault could have no time to dissipate during this period. If the circuit breaker is the only protective device and opens its contacts, a three-phase interruption is caused to all load nodes and all SE will trip, since a real reclosing interval, either instantaneous or delayed, will be longer than the ITIC ride-through time with zero voltage, which is about 0.02 seconds. In fact, with this protection system design no more than one reclosing interval need to be simulated. Fig. 5 shows some plots that correspond to voltage sags caused by different types of faults at L3. Due to transformer connections, voltage drop at the LV side phases is less than 20% with a single-line-to-ground fault. ITIC equipment can withstand such voltage sag up to 10 s; therefore, if the protective device operations were delayed and opening of contacts was not allowed before 10 s, SE could withstand faults with duration shorter than 10 s; this is even more evident with a high fault resistance, see Fig. 5b. In fact, substation grounding can allow the protected equipment to withstand longer duration single-phase-toground faults. A similar conclusion can be derived for two-phase faults, although in this case one or two phases will drop more than 20%, depending on the fault resistance value. Table II provides a summary of the main results taking into account the type of fault. The table shows the clearing times for each fault type, and the consequences as a function of the fault duration, tf. It is evident that the best performance is achieved when the clearing time tc and the fault duration tf are shorter than 10 s, and tc > tf, for two-phase and singlephase-to-ground faults.

The goal is to analyze the characteristics of voltage sags caused by faults using a time-domain simulation tool and compare the voltage sag performance of the test system with and without DG. A. Without Distributed Generation Voltage sag characteristics at LV nodes can be different from the characteristics at the MV nodes. Since it is assumed that only LV sensitive equipment is installed and its voltage sag

250

P hase B

TABLE II. Fault type

VOLTAGE SAG CONSEQUENCES IN THE TEST SYSTEM Clearing time tc - Fig. 2 (Fault resistance = 0) Fault at M0 = 186 ms Fault at L3 = 500 ms Fault at M0 = 230 ms Fault at L3 = 560 ms Fault at M0 = 230 ms Fault at L3 = 600 ms Fault at M0 = 900 ms Fault at L3 = 2050 ms Consequences SE at the three phases will trip if tf > 20 ms; that is before tc. SE at one or two phases will trip if tf <10 s < tc, but the three phases will trip if tf > tc. SE at one or two phases will trip if tf <10 s < tc, but the three phases will trip if tf > tc. There will not be equipment trip if tf < 10 s < tc, but the three phases will trip if tf > tc.

200

Voltage (V)

150

100

P hase A

P hase C

3LG/ 3L 2LG

50

0 0 500 1000 1500 T im e (m s ) 2000 2500 3000

2L

a) Single-phase-to-ground fault Fault resistance=0, duration=2 s


250

1LG
P hase B

200

150

P hase A

P h a se C

100

sions derived from the fuse saving study are still valid. A different approach could be followed if the fuse melting time was longer, since then a more accurate comparison to the voltage tolerance curve should be needed, see [12]. B. With Distributed Generation The protection of small dispersed synchronous generators is much more complex than the protection of a distribution network. In addition, not only the generator protection but also the interconnect protection has to be taken into account. Interconnect protection satisfies the utility requirements to allow the connection of the generator to the grid, while generator protection provides detection of internal shortcircuits and abnormal operating conditions [6], [7], [13]. Note that the operation of the synchronous generator will not have any effect on the coordination between protective devices of this test system. For a study on the effect that DG can have on protective device coordination see [14].

Voltage (V)

50

0 0 500 1000 1500 T im e (m s ) 2000 2500 3000

b) Single-phase-to-ground fault Fault resistance=10 , duration= 2 s


250

P h a se s B and C

200

Voltage (V)

150

100

P h a se A

50

0 0 500 1000 T im e (m s ) 1500 2000

c) Two-phase-to-ground fault Fault resistance=0, duration=1 s


250

P h a se B
200 Voltage (V)

P h a se C

150

Rather than analyzing the performance of the synchronous generator and the interconnect transformer protection, the aim of this study is to estimate what conditions they should fulfill to obtain a better voltage sag performance when the generator is connected to the network. Neither the model of the generator protection nor the model of the interconnect protection have been included in simulations. The main results and conclusions from using fuse saving and fuse blowing are presented in the following paragraphs.

100

P hase A
50

0 0 500 1000 T im e (m s ) 1500 2000

d) Two-phase-to-ground fault Fault resistance=10 , duration=1 s Figure 5. Voltage sags at LV side of L3 without DG Fuse saving. Breaker protection, reclosing time = 0.4 s (Fault location = M2).

Fuse blowing (Fig. 6): According to curves shown in Fig. 2, fuse operation will be so quick that no trip will be caused to SE on the healthy tapped lines; but even some equipment at the LV side of the faulted line will not trip with a single-phase-to-ground fault. If that fault is located along the main feeder, between M0 and M2, the conclu-

Fuse saving: Fig. 7 shows some simulations results. It is evident that the presence of the synchronous generator can improve the voltage sag performance of the test system if it is not disconnected from the network during the fault. According to Fig. 7, SE would not trip during a singlephase-to-ground fault after the breaker opened contacts. In fact, only equipment connected to one or two phases will trip during a two-phase-to-ground fault. This behavior can be achieved if the protection of the generator and its interconnection to the network do not operate/open during the fault. At least two aspects have to be analyzed when considering the possibility of delaying DG protection: the transformer and the generator have to withstand the shortcircuit currents, and islanding conditions must be fulfilled.

250 200 Voltage (V) 150 100 50 0 0 500 1000 Time (ms) 1500 2000

250 200 Voltage (V) 150 100 50 0 0 500 1000 1500 Time (ms) 2000 2500 3000

Figure 6. Voltage sags without DG Fuse blowing (Sags at LV side of node L3, Single-phase-to-ground fault at L3 Fault resistance = 0, Fault duration = 1 s, Reclosing time = 0.4 s)

Figure 7. Voltage sags at LV side of L1 with DG Fuse saving. (Single-phase-to-ground fault Fault duration = 2 s, Fault location = M2, Fault resistance = 0, Reclosing time = 0.4 s).

The transient response of the synchronous generator depends on the fault characteristics (type, location, duration) and the moment of inertia of the rotor. But even with a large value of this parameter (e.g. 0.8 million kg.m2) the oscillations of the active power output can be significant. For a generator capacity of more than 30 kW, IEEE Std 1547 recommends a clearing time of 0.16 s when the voltage range is below 50% of the rated voltage and a clearing time of 2 s when it is between 50 and 88% of the rated voltage [5]. As a rule of thumb, one can assume that for the test system, the first rule applies when the fault is symmetrical, and the second one when the fault is asymmetrical. As for unintentional island, the interconnection system should detect it and cease to energize within 2 s of the formation of the island. For the test system, an island can be created only if the breaker opened its contacts. If the 2-s rule is applied to the test system, then the conclusions derived from the study without DG are valid, although now the presence of the synchronous generator will improve the voltage sag performance when the fault duration is shorter than 2 s. Fuse blowing: If the fault is on a tapped line, fuse operation will avoid trips of SE on the healthy tapped lines. The trip of some equipment at the LV side of the faulted tapped line is also avoided with a single-phase-to-ground fault. If that fault location is on the main feeder, the conclusions are again those derived with fuse saving, although an islanding condition can be created by a breaker opening. V. SENSITIVE EQUIPMENT TRIP PROBABILITY

sh is the percentage of each fault type in the test system.


[P] is a constant 4-element column vector whose values indicate the percentage of SE phases (0, 1, 2 or 3) that will trip with a fault; that is [P] = [0, 0.333, 0.667, 1]T [B] is a matrix with nL rows and 4 columns; each element of a row gives the percentage of test system line length in which a fault will cause a given percentage of SE at one load node to trip. The elements of any row of this matrix must add unity. From the simulation results presented in the previous sections, the percentage of SE trip caused by a certain fault type without DG is approached as follows: Fuse saving: If the fault duration is shorter than 10 s and the breaker does not open, the trip probability at all load nodes is assumed to be 0% for 1LG; 66.6% for 2LG and 2L, 100% for 3LG and 3L. If the breaker opens, those percentages for any load node are all equal to 100%. If the fault duration is longer than 10 s, then the percentages, irrespectively of the protective device performance, are 100% for all fault types. Fuse blowing: If the fault is downstream a fuse, the trip probability at load nodes located in healthy lines is 0% for all fault types. For the load node located downstream the fault, the trip probability is 100% for all fault types, except for 1LG fault, for which 33% of SE will withstand. Since the during-fault voltages at a load node depend on the load location and the fault characteristics (type, location, resistance), these percentages should be seen as average values. The percentages of SE trips, as well as the criteria needed to separate the different cases, will obviously be different with DG. By default the following behavior will be assumed when DG is connected to the test system: the clearing time of the DG will be 0.16 s when the voltage range is below 50% of the rated voltage and 2 s when this range is between 50 and 88% of the rated voltage; with a fault duration shorter than 2 s, the DG will maintain the voltage at healthy phases during asymmetrical faults, irrespectively of the load index in the test system;

The aim is to establish a method for estimating the probability of SE and ranking protection system designs. The trip probability of single-phase LV sensitive equipment at load nodes can be estimated as follows:

[N] = sh (i) [Bi ] [P]


4 i =1

(1)

[N] is a column vector with nL (number of load nodes) rows, in which each element is the single-phase SE trip probability at the corresponding node. i is the fault type (i = 1 for 1LG, 2 for 2LG, 3 for 2L, 4 for 3LG/3L).

with a fault duration longer than 2 s, the DG will be disconnected, and the main supply will be capable of maintaining, with any type of fault, the same voltage level that existed before DG disconnection, except during reclosing intervals, for which a zero voltage must be assumed. The fault frequencies assumed in this work are respectively 75% for 1LG, 12% for 2LG, 6% for 2L, 7% for 3L/3LG. The following example will illustrate the application of the proposed methodology. Assume that the DG is connected to the system, fuse blowing is applied, the fault duration is shorter than 2 s and the time-current curves are those depicted in Fig. 2. The matrices [B] for this case would be as follows
20.5 1 22 [B1 ] = 24.5 20 21.5 10 1 11.5 [B3 ] = 24.5 9.5 11 4 0 10 0 2.5 0 1 11.5 [ B2 ] = 0 4.5 0 24.5 9.5 0 3.0 0 11 0 10.5 4 10 0 10.5 2.5 1 11.5 [ B4 ] = 0 10.5 4.5 24.5 9.5 0 10.5 3 11 0 0 10.5 4 0 10.5 2.5 0 10.5 4.5 0 10.5 3 0 0 14.5 0 0 13 0 0 15 0 0 13.5

Fuse blowing improves the performance of the test system with respect the protection system designs mentionned above, but this improvement is not very significant with most scenarios analyzed in this work. An obvious conclusion from this study is the necessity of reliable information on fault characteristics (type and duration), since their knowledge can contribute to a better selection and setting of protective devices. VI. CONCLUSIONS

The resulting probabilities for this case are [N] [0.2039, 0.1579, 0.2192, 0.1732]T Using the same weighting factor for all load nodes, the index would be now 0.1885.

This paper has analyzed the impact of a synchronous generator on the characteristics of voltage sags caused by faults in a small distribution network. According to the conclusions of the study, DG can have a positive effect on the voltage sag performance of distribution networks if the fault duration is no longer than 2 s. A proper selection and setting of protective devices, as well as fuse blowing, can reduce the number of SE trips. Some care is advisable when analyzing the conclusions. For instance, voltage sag characteristics at the LV side of load nodes could be very different from those obtained in this study if transformer connections were changed. On the other hand, the design of the protection system of a distribution network will not be based only on the voltage sag performance; there are many other aspects to be considered for selecting the type of protective devices, their placement and coordination. REFERENCES
[1] H.Lee Willis and W.G. Scott, Distributed Power Generation. Planning and Evaluation, Marcel Dekker, 2000. M. Godoy Simoes and F.A. Farret, Renewable Energy Systems, CRC Press, 2004. G.M. Masters, Renewable and Efficient Electric Power Systems, John Wiley, 2004. IEEE Power System Relay Committee, Impact of distributed resources on distribution relay protection, August 2004. IEEE Std. 1547-2003, IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems. C.J. Mozina, Interconnect protection of dispersed generators, 2001 IEEE PES T&D Conf., pp. 707-723. IEEE IAS WG Report, Application of islanding protection for industrial and commercial generators, 59th Annual Conf. for Protective Relay Eng., 2006. J.A. Martinez and J. Martin-Arnedo, Voltage sag studies in distribution networks. Part I: System modeling, IEEE Trans. on Power Delivery, vol. 21, no. 3, pp. 1670-1678, July 2006. M.H.J. Bollen, Understanding Power Quality Problems. Voltage Sags and Interruptions, IEEE Press, 2000, New York. J.C. Gmez and M.M. Morcos, Coordination of voltage sag and overcurrent protection in DG systems, IEEE Trans. on Power Delivery, vol. 20, no. 1, pp. 214-218, January 2005. IEEE P1564, Recommended Practice for the Establishment of Voltage Sag Indices, Draft 6, January 2004. J.C. Gmez and M.M. Morcos, Coordinating overcurrent protection and voltage sag in distributed generation systems, IEEE Power Eng.Review, vol. 22, no. 2, pp. 16-19, February 2002. D. Reimert, Protective Relaying for Power Generation Systems, CRC Press, 2006. A. Girgis and S. Brahma, Effect of distributed generation on protective device coordination in distribution networks, Proc. Large Eng. Syst. Conf., pp. 115-119, 2001.

Tables III and IV present the results obtained for the scenarios analyzed in this work, taking into account the time borders that separate the different protective device settings.
TABLE III. Protection system Fuse saving Fuse blowing TABLE IV. Protection system Fuse saving Fuse blowing SENSITIVE EQUIPMENT TRIP PROBABILITY WITHOUT DG Fig. 2 curves tf 2 s 0.988 0.494 tf < 10 s
tc < tf tc > tf

[2] [3]

tf > 10 s
tc < tf tc > tf

[4] [5] [6] [7]

1.0 0.494

0.190 0.187

1.0 0.494

1.0 0.494

SENSITIVE EQUIPMENT TRIP PROBABILITY WITH DG [8] tc > tf 2<tf


<10 s

Fig. 2 curves tf 2 s 0.190 0.189

tf < 2s 0.190 0.189

tf > 10 s 1 0.536

tf < 2s 0.190 0.189

tc < tf 2<tf
<10 s

tf > 10 s 1

[9] [10]

0.190 0.189

1 0.536

0.536 [11] [12]

The main conclusions from these results can be summarized as follows: Since the majority of faults are single-phase-to-ground, and ITIC sensitive equipment can withstand voltage sags caused by these faults during 10 s, the best performance is obtained when tf < 10 s and tc > tf, with or without DG. If tc < tf, then the border value is reduced to about 2 s, which is the maximum allowed time for disconnecting DG from the faulted system.

[13] [14]

Potrebbero piacerti anche