A Survey On The Reliability of Protective Systems For Power Transformers

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

A Survey on the Reliability of Protective Systems for

Power Transformers

Carlos J. Zapata Jhon F. Valencia


Universidad Tecnológica de Pereira Central Hidroeléctrica de Caldas - CHEC
Pereira, Colombia Manizales, Colombia
cjzapata@utp.edu.co jhon.valencia@chec.com.co

Abstract—This paper presents the results of a survey on the


reliability of protective systems for power transformers that
serves at distribution substations of the CHEC system. The
protective systems of seventeen 115/33 kV power transformers in
the range of 20 to 60 MVA and twenty-six 33/13.2 kV power
transformers in the range of 3 to 30 MVA were surveyed using
operating records that covers 7 years. Reliability indexes were
computed for the protective systems as a whole and for their
components and the models of arrival of false openings and their
corresponding repairs were obtained. Results show that: i.
Security is the weakest aspect of the studied protective systems,
being false openings the more frequent problem. ii. Failure to
close is the second more frequent problem of the studied
protective systems. iii. Circuit breakers are the protective system
component that require more attention because they cause most
of false openings and failures to close and have the highest repair
time. iv. Arrival of false openings and their times to repair are
stationary what means reliability of studied protective systems is
not deteriorating. v. Reliability indexes of the protective systems
for 115/33 kV power transformers is higher than the ones of the
33/13.2 kV power transformers.

Index Terms—Circuit breakers, power transformer, power


system protection, power system reliability, relays.
Fig. 1. A typical protective scheme for a power transformer [1]
I. INTRODUCTION
The mission of a protective system (PS) is to detect 3. The more severe the damage a power transformer suffers,
abnormal operating conditions in the protection zone (PZ) to higher are the repair cost and time of repair.
which it is assigned to and to take actions that: guarantee 4. The fact that if a power transformer is severely damaged or
safety, guarantee power system operational security and destroyed, and there is no an available spare, the time for
safeguard the investment in power system assets. In the context delivering a new one is in the order of years.
of this paper the PZ is the power transformer. 5. The very high cost of this type of asset, in the order of
A PS is the set of protective system components (PSC) such millions of dollars.
us relays, instrument transformers, sensors, circuit breakers and In the technical literature, there are a lot of papers devoted
trip circuits assigned to a PZ in accordance to a given to the reliability modeling of PS but few ones specifically show
protective scheme. These schemes are specific for the type and measures of the reliability of existing PS, for example [2], [3],
size of power apparatus to be protected. Fig. 1 shows a typical [4]. It is then the aim of this paper to give information of the
protective scheme for a power transformer [1]. PS for power transformers used in distribution substations
Reliability of PS for power transformers is a matter of because this kind of information is necessary:
utmost importance due to: 1. To have an idea, in quantitative terms, of the reliability of
1. The great impact that an outage on a power transformer of a existing PS.
distribution substation could have on the customer’s 2. To know which types of PS misoperations really occur.
continuity of service. 3. For taking countermeasures for improving reliability.
2. The more a failure on a power transformer lasts the more 4. For applying reliability models and perform reliability
the severity of the damage that it can produce. studies truly tied to reality.
II. FAILURE MODES OF A PROTECTIVE SYSTEM Wanted openings → CTO
A PS can take two kinds of actions: disconnection and Wanted openings successfully performed → (CTO − FTO )
connection of the PZ. These actions arise automatically, due to Wanted closings → CTC
abnormal operating conditions in the PZ, or manually, due to
intentional or unintentional orders given by an operator. These Wanted closings successfully performed → (CTC − FTC )
actions are materialized through the opening and closing of the w a n ted o p en in g s & clo sin g s su cessfu lly p erfo rm ed
circuit breakers associated to the PZ. Requests to the PS to D =
w a n ted o p en in g s & clo sin g s
come into action can thus be: calls to open (CTO) or calls to
close (CTC).
A PS operates correctly and appropriately if it does not fail (C T O − F T O ) + (C T C − F T C )
D = (2 )
when it is called to operate and does not operate when it is not CTO + CTC
required. The basic PS failure modes are then [5], [6]: failures
to operate, which include failures to open (FTO) and failure to C. Security
close (FTC), and false operations, which include false openings Security refers to the degree of certainty that the PS will not
(FO) and false closings (FC). produce false operations. It is measured in Eq. (3) as the ratio
Failures to operate include those situations where the of wanted calls to operate which were performed successfully
opening or closing takes more than the specified time. to the number of wanted and unwanted openings and closings
which were performed. Thus:
III. PROTECTIVE SYSTEM RELIABILITY INDEXES
The aspects of PS reliability are explained in [7]-[10]; Wanted openings → CTO
definitions presented here are taken from [7]. Formulation of Wanted openings successfully performed → (CTO − FTO )
PS reliability indices presented here is derived from [2] and [4]; Wanted closings → CTC
other formulations can be consulted in [8], [11], [12], [13].
Wanted closings successfully performed → (CTC − FTC )
A. Reliability Unwanted openings → FO
Reliability refers to the degree of certainty that the PS will Unwanted closings → FC
perform correctly. It is measured in Eq. (1) as the ratio of
wanted openings and closings that were successfully performed w a n ted o p en in g s & clo sin gs p erfo rm ed
S =
to the number of exposures. An “exposure” is an event that: (1) w a n ted a n d u n w a n ted o p en in g s & clo sin g s p erfo rm ed
Truly requires the operation of the PS due to a failure or
abnormal operation condition in the PZ or because an opening (C T O − F T O ) + (C T C − F T C )
or closing is ordered manually or automatically. (2) Falsely S = (3 )
(C T O − F T O ) + (C T C − F T C ) + F O + F C
starts the operation of the PS due to misoperations of some
PSC. Thus:
D. Probability of Failure when Opening
Wanted openings → CTO It is measured in Eq. (4), in accordance to frequency
Wanted openings successfully performed → (CTO − FTO ) definition of probability, as the ratio of failures to open to the
Wanted closings → CTC calls to open.
Wanted closings successfully performed → (CTC − FTC ) PFTO = FTO / CTO (4 )
Unwanted openings → FO
Unwanted closings → FC E. Probability of Failure when Closing
It is measured in Eq. (5), in accordance to frequency
w a n ted op en ing s & clo sin g s su cessfully p erfo rm ed definition of probability, as the ratio of failures to close to the
R =
ca lls to o p era te + fa lse o p era tio ns calls to close.
PFTC = FTC / CTC (5 )
(C T O − F T O ) + (C T C − F T C )
R = (1 )
(C T O + C T C ) + ( F O + F C ) F. Mean time to Repair
It is measured in Eq. (6), as the statistical mean of the
B. Dependency
sample of repair times. n is the number of failures and ttri is
Dependency refers to the degree of certainty that the PS the time to repair of failure i .
will perform correctly when it is called upon to operate. It is
measured in Eq. (2) as the ratio of wanted openings and n
r = (  ttri ) / n (6 )
closings which were successfully performed to the number i =1
wanted openings and closings. Thus:
G. Failure Rate B. Time to repair
It is measured in Eq. (7), as the number of misoperations, i. Times to repair are generally stationary and, by this reason,
e. failures, to the interval in which the PS or PSC was a probability distribution is the most common model for
operative. n is the number of misoperations, T the period repairs. The procedure for fitting this kind of model includes
covered by the operating records and N the number PS or PSC steps of analysis of tendency, independency and goodness of fit
from which the sample was taken. test; it is explained in [16].

λ = n / (N *T − r * n) (7 ) V. PROTECTIVE SYSTEMS UNDER STUDY


This study covered a population of seventeen PS for 115/33
When the repair times where missing, the failure rate was
kV power transformers and twenty-six PS for 33/13.2 kV
computed in the following simplified way:
power transformers. Operating records cover a period of seven
λ ≈ n / (N *T ) (8 ) years and correspond to the events recorded by the CHEC’s
control center [17].
H. Example The rating of 115/33 kV power transformers are in the
range of 20 to 60 MVA and the rating of 33/13.2 kV power
Let us consider the following operating data taken of two transformers are in the range of 3 to 30 MVA. Both power
identical PS: transformers classes have three winding construction connected
Yn-Yn-Δ, but the tertiary is not used. All these power
CTO= 30 FTO= 2 FO= 2
transformers serve at CHEC’s distribution substations for rural
CTC= 32 FTC= 3 FC= 1 and urban zones.
T= 5 years r= 4,5 hours The PS under study are composed by the following PSC:
1. A relay unit for overcurrent protection.
(3 0 − 2 ) + (3 2 − 3) 2. A relay unit for differential protection.
R = = 0, 8 7 6 9
(3 0 + 3 2 ) + ( 2 + 1) 3. A module for communications to CHEC’s control center.
4. SF6 circuit breakers for 115 kV.
(30 − 2 ) + (3 2 − 3) 5. SF6 and oil circuit breakers for 33 kV.
D = = 0, 9 1 9 4 6. Vacuum and oil circuit breakers for 13.2 kV.
30 + 32
7. Devices for oil and winding temperature
8. Buchholz relay.
(3 0 − 2 ) + (3 2 − 3)
S = = 0, 9 5 0 0 9. Pressure relief valve.
(3 0 − 2 ) + (3 2 − 3) + 2 + 1
10. Current transformers for 115, 33 and 13.2 kV.
The arrangement of 115/33 kV substations is main and
P F T O = 2 / 30 = 0, 0667 transfer bus at both voltage levels. The arrangement of 33/13.2
kV substations is single bus arrangement at both voltage levels.
P F T C = 3 / 32 = 0, 0938
VI. METHODOLOGY
n = FTO + FTC + FO + FC = 2 + 3 + 2 + 1 = 8 Failures The procedure implemented in this survey include the
following steps:
N = 2 PS 1. From the operating records, classify the events of the PS as
CTO, CTC, FTO, FTC, FO and FC. The events when
8 failure of PS was caused by erroneous settings (Human
λ = = 0, 8 0 0 3 Failures/year
4, 5 * 8 error) were discarded.
2*5 −
8760 2. For each event of FTO, FTC, FO and FC identify the PSC
involved in the event.
In the computation of λ , the factor 8760 is used to convert 3. Compute the reliability indexes for the PS and its PSC.
units of the mean repair time from hours to years. 4. For those failures that have enough data ( n ≥ 20 ), apply the
procedure to fit a probability model for arrival of failures
IV. MODELS FOR FAILURES AND REPAIRS and the corresponding times to repair.
A. Arrival of failures 5. Analysis of results, conclusions and recommendations.
A stochastic point process is the model for the arrival of VII. RESULTS
failures. The procedure for fitting this kind of model includes Tables I and II present the summary of operating events and
steps of analysis of tendency, independency and goodness of fit the corresponding reliability indexes for both types of PS
test to stationary or non-stationary models in accordance to the considered in this study.
tendency; it is explained in [14] and [15]. Units for λ and r are [failures/year] and [hours],
respectively. R , D , S , PFTO , PFTC are probabilities.
Failure rates of PS were computed with T = 7 years for With the symbol “--" are marked all indices that could not
both types of PS. N = 17 for 115/33 kV power transformers be computed because there were no failures.
and N = 26 for 33/13,2 kV power transformers. Eight FO were reported for the PS of 33/13.2 kV power
Failure rates of circuit breakers and current transformers are transformers which were caused was erroneous settings. As
computed taken into account that each PS have two of these mentioned in the methodology, this kind of events are not
components. considered because they are no caused by PSC.
For some PSC, although failures were recorded, the repair Tables III and IV present, respectively, the models for
times were missing and, by this reason, the corresponding r is arrivals of FO and repairs. Only this kind of failure had
not presented and marked with “N/A” (Not Available). samples with enough data to fit models.

TABLE I. SUMMARY OF OPERATING EVENTS AND RELIABILITY INDICES OF PS FOR 115/33 KV POWER TRANSFORMERS

Item CTO FTO CTC FTC FO FC R D S PFTO PFTC r λ


Communication 92 0 114 2 0 0 0,9903 0,9903 1,0000 0,0000 0,0175 N/A 0,0168
Current transformer 92 0 0 0 3 0 0,9684 1,0000 0,9684 0,0000 -- 26,900 0,0126
Circuit breaker 93 0 114 2 9 0 0,9491 0,9903 0,9579 0,0000 0,0175 71,330 0,0462
Pressure relief valve 1 0 0 0 1 0 0,5000 1,0000 0,5000 0,0000 -- 12,317 0,0084
Overcurrent relay 78 0 0 0 4 0 0,9512 1,0000 0,9512 0,0000 -- 0,900 0,0336
Differential relay 9 0 0 0 3 0 0,7500 1,0000 0,7500 0,0000 -- 2,083 0,0252
Oil temperature device 0 0 0 0 1 0 -- -- -- -- -- 0,220 0,0084
PS as a whole 365 0 228 4 21 0 0,9593 0,9933 0,9656 0,0000 0,0175 38,596 0,2103

TABLE II. SUMMARY OF OPERATING EVENTS AND RELIABILITY INDICES OF PS FOR 33/13,2 KV POWER TRANSFORMERS
Item CTO FTO CTC FTC FO FC R D S PFTO PFTC r λ
Communication 198 0 240 3 0 0 0,9932 0,9932 1,0000 0,0000 0,0125 N/A 0,0165
Current transformer 180 0 0 0 0 0 1,0000 1,0000 1,0000 0,0000 -- -- 0,0000
Circuit breaker 199 1 249 12 30 2 0,9063 0,9710 0,9315 0,0050 0,0482 182,870 0,1239
Pressure relief valve 6 0 0 0 8 0 0,4286 1,0000 0,4286 0,0000 -- N/A 0,0440
Overcurrent relay 160 0 2 2 2 0 0,9756 0,9877 0,9877 0,0000 -- 25,000 0,0220
Differential relay 20 0 0 0 5 0 0,8000 1,0000 0,8000 0,0000 -- 35,300 0,0275
Oil temperature device 0 0 0 0 1 0 -- -- -- -- -- N/A 0,0055
Winding temperature device 2 0 0 0 1 0 -- -- -- -- -- 0,220 0,0055
PS as a whole 765 1 491 17 47 2 0,9487 0,9857 0,9619 0,0013 0,0346 154.652 0,3705

As can be seen in Tables I and II, the main problem of both TABLE IV. MODELS FOR REPAIRS OF FALSE OPENINGS [HOURS]
PS is FO; it is the most frequent type of misoperation and by PS Model
r
Parameters
this reason security is the weakest aspect of their reliability. [hours]
Lognormal
The PSC that cause most of FO are the circuit breakers 115/33 kV
distribution
35,479 μ = 2,7447 σ = 1, 2839
followed by the relays and sensors. Lognormal
33/13.2 kV 150,337 μ = 4,1887 σ = 1, 2839
The second more frequent misoperation of both PS is FTC distribution
which decreases dependency; these are caused mainly by the
circuit breakers followed by communication devices and As can be seen in tables III and IV, the process of arrival of
overcurrent relays. false openings and their corresponding repairs are stationary for
Repair times of PS for 33/13.2 kV power transformer are both PS. This means that, as time evolves, there is no evidence
higher than the corresponding ones of PS for 115/13.2 kV that FO arrivals and times to repair increase (reliability
power transformers; the reason for this high difference is that deterioration) or decrease (reliability improving).
most of 33/13.2 kV substations are located in rural areas and
transportation takes a long time. VIII. CONCLUSIONS
Security is the weakest aspect of the reliability of studied
TABLE III. MODELS FOR OCCURRENCE OF FALSE OPENINGS protective systems because for both false openings is the more
Mean time Intensity frequent misoperation.
PS Model between failures function The second more frequent misoperation of the studied
[years] [Failures/year] protective systems is failures to close which decreases their
Homogeneous λ(t ) = 0,1695
115/33 kV
Poisson process
5,899 dependency.
Homogeneous Circuit breakers are the protective system component that
33/13.2 kV 4,042 λ(t ) = 0, 2474
Poisson process require more attention by the operator of the system in order to
improve reliability because they cause most of FO and FTC Jhon F. Valencia was born in Colombia. He obtained his BSc degree in
electrical engineering from the Universidad Nacional de Colombia,
and have the highest repair time. Manizales, Colombia, in 2005 and the degree of specialist in transmission and
Arrival of FO and the times to repair are stationary what distribution power systems from the Universidad de los Andes, Bogotá,
means reliability of the studied PS is no deteriorating as time Colombia, in 2009. Since 2005 he has worked for CHEC S. A, Manizales,
evolves. Colombia.
Reliability of the PS for 115/33 kV power transformers is
higher than the one of the 33/13.2 kV power transformers.
IX. REFERENCES
[1] IEEE, Guide for Protective Relay Applications to Power Transformers,
(Standard C37.91, 2008).
[2] G. H. Kjølle, O. Gjerde, B. T. Hjartsjø, H. Engen, L. Haarla, L. Koivisto,
P. Lindblad, Protective systems faults – a comparative review of fault
statistics, 9th International conference on Probabilistic Methods Applied
to Power Systems, 2006.
[3] T. Johannesson, F. Roos, S. Lindahl, Reliability of protection systems –
operational experience 1976-2002, Lund University, 2004.
[4] C. J. Zapata, H. A. Cely, “A survey on the reliability of protective
systems for large hydro generating units” in Proc. 2014 IEEE
Transmission & Distribution Latin America Conference, 2014, Medellín,
Colombia.
[5] C. J. Zapata, D. S. Kirschen, M. A. Ríos, A. Torres, Reliability
assessment of protective schemes considering time varying rates,
International Review of Electrical Engineering, No. 65, 2009.
[6] M. H. J. Bollen, P. Nassee, A classification of failures of protection, 3th
International conference on Probabilistic Methods Applied to Power
Systems, 1991.
[7] IEEE Working Group, Terms Used by Power System Protection
Engineers (Tutorial Course TP-130-0-031998-1-0), 1997.
[8] IEEE Working Group, Transmission Protective Relay System
Performance Measuring Methodology, 1999.
[9] IEEE Working Group, Protective Relaying Performance Reporting,
IEEE Transactions on Power Delivery, Vol. 7, No. 4, October 1992.
[10] C. J. Zapata, E. Ortiz, M. A. Ríos, Reliability aspects of protective
relaying for power systems, Mundo Eléctrico, No. 71, 2008. (In
Spanish).
[11] G. F. Johnson, Reliability considerations of multifunction protection,
IEEE Transactions on Industry Applications, Vol. 38, No. 6,
November/December 2002.
[12] IEEE Working Group, Proposed statistical performance measures for
microprocessor-based transmission line protective relays, part 1 -
Explanation of the statistics, IEEE Transactions on Power Delivery, Vol.
12, No. 1, January 1997.
[13] IEEE Working Group, Proposed statistical performance measures for
microprocessor-based transmission line protective relays, part 2 -
Collection and uses of data, IEEE Transactions on Power Delivery, Vol.
12, No. 1, January 1997.
[14] C. J. Zapata, J. Díaz, M. L. Ocampo, J. D. Marriaga, J. U. Patiño, A. F.
Gallego, “The repair process of five Colombian power distribution
systems” in Proc. 2010 IEEE Transmission & Distribution Latin
America Conference, 2010, Sao Pablo, Brazil.
[15] C. J. Zapata, J. Urrea, “Assessing the service rendered by a power
distribution control center” in Proc. 2012 IEEE Transmission &
Distribution Latin America Conference, 2010, Montevideo, Uruguay.
[16] C. J. Zapata, S. C. Silva, O. L. Burbano, “Repair models of power
distribution components” in Proc. 2008 IEEE Transmission &
Distribution Latin America Conference, Bogotá, Colombia, 2008.
[17] J. F. Valencia, Reliability of protective systems of power transformers of
the CHEC system, Graduation Project, Universidad de los Andes,
Bogota, 2009. (In Spanish).

X. BIOGRAPHIES
Carlos J. Zapata was born in Colombia. He obtained his BSc degree in
electrical engineering from the Universidad Tecnológica de Pereira, Pereira,
Colombia, in 1991 and his MSc and PhD degrees in electrical engineering
from the Universidad de los Andes, Bogotá, Colombia, in 1996 and 2010. Mr.
Zapata worked for eleven years for Concol S. A, Bogotá, Colombia. Since
2001 on, he has worked as a professor at the Universidad Tecnológica de
Pereira, Pereira, Colombia.

You might also like