A robust method for single phase fault location considering distributed generation and current compensation C. Orozco-Henao, J Mora-Flórez Member, and S. Pérez-Londoño, IEEE.
Abstract-- In this paper, a methodology to face the problem of fault location considering distributed generation is presented. As one of most relevant characteristic of the proposed method, the fault current is estimated using the measurements of voltage and current obtained at the power substation and at the node where the DG is located. This last measurements avoid the modeling effort required to represent the power generator, and also it possibilities to take into account their variations in fault and prefault steady states. The proposed methodology uses a method based on the estimation of the fault impedance which considers the effect of distributed generation and the variation of the load current. The methodology presented is validated in the IEEE 34nodes test system, which is modified to connect one generator in the longest feeder. Tests are performed considering single phase faults, resistances from 0 to 40 ohms and distributed generation participation of 10%. Also, this methodology is validated in three scenarios: power system considering constant impedance load models, power system considering constant impedance, current and power load models and finally a power system considering the last models load and variations at the customer demand [50%-120%]. IndexTerms— Fault location, distributed generation, load estimation, power quality, and power distribution system.
I. NOMENCLATURE /
: Voltage measured at the power substation.
/
: Current measured at the power substation. : Voltage measured at the distributed power substation. : Current measured at the distributed power substation. : Voltage at bus k. , : Fault current from bus k to bus F. ,
: Fault current from bus (k+1) to bus F.
: Shunt current at bus k during fault steady state : Current at bus F. : Fault distance per unit. This work was developed in ICE3 (Col) Research Group on Power Quality and System Stability. This work was supported by Vice-Rectory for Research and Extension of Universidad Tecnológica de Pereira under project E6-11-12. C. Orozco-Henao, is assistant researcher at the Universidad Tecnológica de Pereira, Colombia (Tel/fax: +57 6 3212044, e-mail:
[email protected]). J. Mora-Flórez, is associated professor at the Universidad Tecnológica de Pereira, Colombia (Tel/fax: +57 6 3212044, e-mail:
[email protected]). S. Pérez-Londoño, is associated professor at the Universidad Tecnológica de Pereira, Colombia (Tel/fax: +57 6 3212044, e-mail:
[email protected]).
978-1-4673-2673-5/12/$31.00 ©2012 IEEE
: Self line impedance phase i, with i=(a,b,c). : Mutual line impedance to phase i, j, with (i≠j). : Load admittance matrix connecting at bus i. : Pre-fault votage at bus k. : Nominal voltage at bus k. : Coefficient related to constant impedance component. : Coefficient related to constant current component. : Coefficient related to constant power component of active power : Distance estimated from the power substation to the faulted node. nt: Total section analyzed : The last estimated fault distance ú : Length of last analyzed line section.
F
II. INTRODUCTION
rom the current deregulated electric markets, the interest on the presence of alternative sources at electric power distribution systems has increased. These generation centers have introduced new problems on the power system, due to the presence of phenomena such as bidirectional currents, which directly affect protection systems [1]. These type of problems caused a high fault rates and also high fault currents, which are reflected in the service quality provided by the utilities. In general the service quality is related by continuity indices, which are considered by regulatory agencies. These indices normally used to quantify the service continuity are the System Average Interruption Frequency Index (SAIFI) and System Average Interruption Duration Index (SAIDI) [2]. The most effective way to improve continuity of electricity supply and therefore these indices, is reducing by the number of interruptions and the restoration time [3]. Fault location is used to perform two tasks at the time, by determining the weak areas and quickly locate them, is possible decreased interruption and restoration time [4]. Initially, the fault location methods have been developed to be used in radial power distribution systems. Currently, these systems have introduced new features such as generation centers in different parts of the system (distributed generation, DG), which implies new features not considered by conventional strategies [5]. Furthermore, these new generation centers must be accurately modeled and for most of the cases their parameters are not available. This corresponds to a new problem to the fault location tasks, as they must determine ways to consider the effect of these new generation centers. Some authors have proposed different ways to consider the effect of distributed generation. Currently, the proposal
2
accepted in the literature uses DG represented by synchronous generator model. However, this proposal may be considered as an approximation since the model used is an approximate model of the synchronous machine and in most of the cases the exact parameters of this machine are not well known. Bretas in [6] has presented a first approach with a fault location method that uses a synchronous generator model to represent the effect of DG. However, the method is lacking of robustness, because this strategy does not consider variations in the system. Based on this work, Nunes presents in [7] an interesting strategy that introduces the effect of DG on the load current, using the synchronous generator model with all its parameters known. A different concept is presented by Marvik in [8] which show a localization strategy that introduces the effect of distributed generation by two correction factors. However, this methodology uses the subtransient reactance and generator internal voltage, which also requires of the generator parameters. The proposal presented in this paper introduces the effect of distributed generation in the estimation of the current fed to the fault from the DG source. This current is calculated from the values of voltage and current measured at DG, which avoids the use of the generator model and as a consequence avoids the use of its parameters. Additionally, variations in the load to implement a robust method of fault location are considered. This allows it to be an alternative to locate single phase faults which is simple and easy to apply in real power distribution systems. The paper is presented in six sections. In the section III presents the mathematical development of the proposed fault location method. Next, the section IV presents the proposed algorithm for fault location method; in the Section V the tests and results in a prototype distribution system are analyzed. Finally, Section VI presents the major findings associated with this research.
A. Analysis for single fault phase to ground Considering a single fault between nodes k and k+1, modeled as shown in Fig 2 the equation set (1) for i (i= a, b, c) is obtained. ,
,
,
,
,
,
,
,
,
k+1
F
k
Fig. 2.Equivalent power system model in the case of single-fault.
,
,
…
0 0 0 0 0 0
0 0
The proposed method is developed to consider the main feeder shown in Fig 1, which is composed by different laterals, measurements of the fundamental rms values of voltage and current at the main and DG substations. The main feeder presents a fault between the nodes k and k+1.
1,
(1)
0 0
From the first row of (1), (2) is obtained. This depends on the fault resistance (RF) and the fault distance (m). ,
…
,
…
, ,
2
1,
If (2) is divided by III. METHODOLOGY PROPOSED TO FAULT LOCATION
…
,
,
,
, is possible obtain a expression for fault resistance , which considers the imaginary components, as is presented in (3). ∆ ,
(3)
1,
∆ 2 /
,
, /
k
F
k+1
q
Where, ∆
,
,
4)
If (3) is replaced in (2) the fault distance (m) is easy obtained, as presented in (5).
Fig. 1.Model of the faulted power system.
The values of voltage and current are estimated for every segment of the main feeder using those values measured at the power substation. Getting these values in each node, is possible reduce the analysis only for the line section where the fault is located, as proposed in [9].
,
,
, ,
,
(5) ,
3
IV. PROPOSED ALGORITHM TO ADJUST THE FAULT CURRENT The fault location proposed algorithm is aimed to determine the fault current using the values of voltage and current measured at DG substation. The method assumes a fault placed in any line section of the power system; hence the algorithm must run section by section. The fault current between nodes (k+1) and (F) is assumed as the fault current calculated from DG to the line section in analysis. The generalized algorithm to estimate the distance fault is presented in Fig.5 and in the literals from A to G. Start
many radials feeders as ending nodes at the power system. B. Estimation of the values of voltage and current at the DG lateral The use of values of voltage and current measured at the main substation and DG has represented a huge advantage to the fault location, since this avoids knowing the state of systems before to the fault. In this way, is possible to involve the DG effect in the fault current. The strategy assumes a fault at the section line between the nodes (k) and (k+1), as is presented in Fig. 1. If the power system is analyzed in detail, is possible to in the fault section is observe that the current , represented by the current obtained in that section from the fault values of voltage and current from Generator 2 (DG). Using the measurements in DG and assuming a configuration of the power system, as is depicted in Fig (6), the values of voltage and current in all of the nodes at the feeder are estimated using (24) and (25).
Obtain information about system
configuration Estimate the values of voltage and current at the nodes of the DG lateral
Determinate the set of descriptors and the fault type Calculation of fault distance Analysis for each line section, considered as section in fault. Current load estimate
Si
Does this line section belong to DG radial.
/
No Estimate load impedance as seen in the line section (9) and (10)
Assume I (k +1, F) as currents estimated with DG measures with (6) and (7)
2
,
/
,
k
/
,
q
k+1
Estimate load current with (32)
Fig.6.Model of the faulted power system used to estimate the load current.
Refinement
The equations (6) and (7) are used to estimate the values of voltage and current at fault steady state at node (k+1).
Estimate fault resistant (Rf)and fault distance (m) Estimate the current I(k+1,F)
(6)
1 Not
|Ik(k+1,F)- Ik-1(k+1,F)|
