Documente Academic
Documente Profesional
Documente Cultură
Abstract – A reliability study has been performed recently at based on Markov models for determining the optimal number
Hydro One to evaluate the optimal numbers of spare of spare units for a given population of transformers used in
transformers required for 250 MVA and 125 MVA, 230/115 kv, transmission stations.
3 phase auto-transformers of the Hydro One’s transmission
system. In this study, a probabilistic method based on a Markov
A reliability study was performed in 2003 [10] to determine
model and a bulk power reliability program was developed in
determining the number of spare units for each group of auto- the numbers of spare auto-transformers required for 250
transformers. The purpose of this paper is to describe the MVA and 125 MVA, 230/115 kv, 3 phase auto-transformers
assessment approach used in the study and to present the study of the Hydro One transmission system. The study used a
findings and its recommendations. benefit/cost analysis approach in determining the required
numbers of spares. The following assumptions were made in
Index Terms- Auto-transformer, spare units, Markov model, the study:
reliability assessment of bulk power system, probability of
failure, failure consequences, system availability, benefit/cost - No power flow models were used in the assessment of
analysis.
the failure consequences.
- Approximate load duration curves were used to estimate
I. INTRODUCTION
potential load cuts for multi-unit stations.
- All generators connected to the 115 kV transmission
High voltage auto-transformer stations in bulk transmission
network were not considered.
systems are normally designed to withstand the failure of one
- Power transfers between 115 kv areas through normally
unit or two units without shedding customer’s loads.
open lines were allowed before cutting any loads.
Therefore, the loss of one auto-transformer at any transformer
station may not be a concern to utilities from the reliability
The study findings suggested that two spare auto-
point of view. On the other hand, the loss of one unit at a
transformers are required for each group of the auto-
transformer station may result in generation rescheduling and
transformers.
therefore an increase in the system energy production costs.
In other situations, the loss of two units at the same station
At present, there are 53, 250 MVA auto-transformers and
may result not only in generation rescheduling but also in
31, 125 MVA auto-transformers on the Hydro One’s bulk
load shedding.
transmission system. In terms of spare units, there are 2
spare units for the 250 MVA group and one spare unit for the
The failure of auto-transformers can be of two types: minor
125 MVA group. Since 2003, there have been many changes
(or Class II failures) and major (Class I failures). Class II
to the transmission system network, system generation,
failures are of short duration and are repaired on site and are
system load and the number of auto-transformers on the
covered by system planning criteria. On the other hand,
system. The purpose of this study is to review the adequacy
Class I failures are of long duration and can not be easily
of the current spare policy for the two groups of auto-
repaired on site and failed units may be repaired or replaced
transformers following the various system changes.
depending on unit conditions resulted from the failure. It has
been a practice in the utility industry to carry spare units for a
II. METHOD OF ASSESSMENT
group of similar auto-transformers in service to avoid
extended customer interruption or impaired system operating
Two criteria are used in the current study to determine the
conditions. Determining how many spare units are needed
number of spare transformers for each group of auto-
has been a big challenge for transmission companies.
transformers. The 1st criterion assumes that spare units are
added to the population, one at a time, until a pre-determined
More work has been done in the area of the spare
level of group availability is reached or the group availability
transformer requirements for distribution stations [1]-[7]. On
saturates. The 2nd criterion is based on the minimization of
the other hand, some work has been reported in the area of
the total system cost (cost of carrying spares and outage costs
spare transformer requirements for transmission stations.
resulting from transformer failures). The required number of
References [8] and [9] presented some probabilistic methods
This method requires the calculation of probabilities of 1. Establish a system base case using PROCOSE. In this
Class I transformer failures. These failure probabilities are base case, generation is dispatched economically and all
obtained using the Markov model of [8]. 250 MVA and 125 MVA, 3 phase auto-transformers are
assumed to be in service and all critical transmission
A transformer station with more than one unit is normally network interfaces including the auto-transformer ratings
designed to withstand the loss of one unit at any time without are observed.
overloading the remaining units. However, the loss of the 2. Obtain from the base case results, the system expected
second unit at the same station could result in overloads at the MW cut and the system average hourly production cost.
same station or in nearby stations. Therefore, remedial These system quantities are computed under the
actions such as generation rescheduling and/or load shedding assumptions that the transmission system remains intact
are taken to alleviate overloads. and system generators can fail. In addition, the computed
system quantities can change under different generation
The probabilities of having one Class I failure (T1) and two dispatches.
Class I failures (T2) at one station are given by: 3. Select one transformer station and run PROCOSE cases
with one transformer and two transformers out of service
T1 = (m/N/) P1 (1) to compute the system expected MW cut and the system
average hourly production cost.
T2 = (mc2 /Nc2) P2 (2) 4. Use the results in Steps 2 and 3 to obtain the changes
Where: (only increases) in the system expected MW cut and in
the system average hourly production cost.
P1= Probability of having N-1 units in service 5. Repeat Steps 3 and 4 for other transformer stations
P2= Probability of having N-2 in service 6. Repeat the above steps for some overlap outages of two
m =Number of units at the station transformers at different stations in the same local area.
N =Number of units at all stations
m
c2 = Number of ways a pair of units can be formed from a It should be mentioned that in most cases, the loss of one
population of m units and is equal to m (m-1)/2. auto-transformer at any multi-unit station would have no
N
c2 = Number of ways a pair of units can be formed from a impacts on customers. On the other hand, the loss of two
population of N units and is equal to N (N-1)/2. auto-transformers at one station or the loss of two at two
stations supplying one 115 kv local area could result in an
The probability of the overlap of two Class I failures at overload of the remaining units or in an island. The overload
Stations i and j, T2ij, in the same transformer group is given problems can be solved by rescheduling generation in the 115
by: kv local area if any exists or by cutting loads. In case of
islanding, the entire 115 kv local area load will be lost if it
T2ij = (1 /Nc2) P2 (3) can not be transferred to neighbouring areas. On the other
hand, if transmission ties exist between the 115 kv affected
The probability of the overlap of two Class I failures at area and its neighbouring areas, the entire area load or portion
Stations i and j, T2ij, in two different transformer groups is of it can be transferred. The amount of load cuts computed by
given by: PROCOSE under a given contingency should be adjusted to
reflect the amount of load transfer.
T2ij = P1i P1j/Ni Nj (4)
III. DETERMINING THE OPTIMAL NUMBER OF
Where: SPARES
P1i = Probability of having one Class I failure in Group 1 In this study, both assessment criteria are used in
P1j = Probability of having one Class I failure in Group 2 determining the optimal number of spare units for each group
Ni = Number of units in Group 1
2
ARC = Annual risk cost in dollars The current generation system with no coal-fired plants was
T1i = Probability of having one failure at Station i used in the study. Some generating units were modelled
D = Length of study period in hours (8760 hours) deterministically and others were modeled probabilistically.
MWC1i = Increase in system expected MW cut with one
failure at Station i
3
1. A summary of unit reliability indices used in the study is 250 MVA 125 MVA
provided in Table I (Based on Hydro One’s experience Group Group
No. of units 53 31
over the last 14 years).
Class I annual unit failure rate .00292 .005044
2. Single and double outages of auto-transformers are Unit repair or lead time in years 1.5 1.5
considered in the assessment. Spare unit installation time in days 35 130
3. Spare transformer units are used at any one of the
stations considered in the study even if there are no TABLE II OTHER RELEAVANT DATA
outage consequences. The outage consequences at a
Cost of one 250 MVA spare unit $ 2.5 million
particular station are unknown until the consequences Cost of one 125 MVA spare unit $ 1.8 million
assessment is done. If transformer outages at a station are Annual discount rate in % 5
known in advance to have no impact on the computed Average customer interruption cost 10 $/KWh
quantities, then the station should be excluded in the No. of repair teams More than 1
spare assessment. No. of installation teams 1
4. Study period of one year is considered in the assessment.
TABLE III PROBABILITY OF SYSTEM FAILURES FOR 250
5. In case of islanding, 50 % of the area or pocket load will MVA GROUP
be transferred to neghbouring pockets if normally open
lines exist between them. Case Probability Probability of Probability of
of no One Failure Two Failures
Failures (P1) (P2)
Table II provides information about other relevant data 0 spare .783077917 .193404584 .023517499
used in the assessment. 1 spare .962728836 .035173341 .002097823
2 spares .983495686 .016178589 .000325725
VI. STUDY RESULTS AND DISCUSSION 3 spares .98507467 .014707156 .000218174
The results for the two criteria used in the assessment are
now discussed. Table VI provides a summary of the benefit/cost analysis
for the 125 MVA group with 0, 1, 2 and 3 spare units. It is
REFERENCES
TABLE VI SUMMARY OF COST RESULTS FOR125 MVA GROUP
[1] W. Li, E. Vaahedi and Y. Mansour, “Determining number and timing
Number Annual Annual Annual Total
of substation spare transformers using a probabilistic cost analysis
of Carrying Increase in Increase in Annual
approach,” IEEE Transactions on power delivery, Vol. 14, No. 3, July
Spares Cost in Production Customer cost in
1999.
$k Cost in $k Interruption $k
[2] J. M. Nahman and M. R. Tanaskovic, “Probability models for optimal
Cost in $k
sparing of distribution network transformers,” IEEE Transactions on
0 0 5.33 1049.47 1054.80
Power Delivery, Vol. 24, Issue 2, April 2009, pp. 758-763.
1 90 .32 170.80 261.12 [3] A. A. Chowdhury and D. O. Koval, “Development of probabilistic
2 180 .11 96.10 276.21 models for computing optimal distribution substation spare
3 270 .10 91.76 361.86 transformers,” IEEE Transactions on Industry Applications, Vol. 41,
Issue 6, November 2005, pp. 1493-1498.
Additional studies were performed to evaluate the impact of [4] A. M. Leite da Silva, J. G. de Carvalho and A. A. Chowdhury,
“Probabilistic methodologies for determining the optimal number of
varying some of the system paramters on the number of substation spare transformers,” IEEE Transactions on Power Systems,
spares. In these studies, the unit failure rate was doubled, the Vol. 25, Issue 1, February 2010, pp. 68-77.
unit repair time was increased to 2 years and the customer [5] G. A. Hamoud, “Assessment of spare transformer requirements for
interruption cost was increased to $20 /KWh. distribution stations,” IEEE Transactions on Power Systems, Vol. 26,
Number 1, February 2011, pp. 174 - 180.
[6] G. A. Hamoud, “Use of Markov models in assessing spare transformer
The results using the two assessment criteria show that the requirements for distribution stations,” IEEE Transactions on Power
number of spare units for the 250 MVA is two for all the Systems, Vol. 27, No. 2, May 2012 , pp. 1098 - 1105.
cases except one. [7] G. A. Hamoud and C. Yiu, “One Markov model for spare analysis of
distribution power transformers, ” IEEE Transactions on Power
Systems, Vol. 31, No. 2, March 2016 , pp. 1643 - 1648.
The results also show that the number of spare units for the [8] G. Hamoud,” Assessment of spare transformer requirements for high
125 MVA using both criteria is two for most of the cases voltage load stations,” PES General Meeting, 2012.
studied. [9] G. Hamoud, F. Qureshy, A. Elen and L. Lee, “Assessment of high
voltage auto-transformer spare requirements in bulk transmission
systems,” Power Engineering Society General Meeting, 2004, IEEE,
In summary, the two criteria should be used to complement June 6 – 10, pages 434 – 439.
one another when assessing the number of spares. [10] B. Parker, “Optimum Number of Spare Auto-transformers 250 MVA
and 125 MVA, 230/115 kv Categories,” Hydro One Internal Report,
VII. CONCLUSIONS AND RECOMMENDATIONS November, 2003
[11] B. Porretta and D. Kiguel, “Bulk power system reliability evaluation,
part I: PROCOSE – a Computer Program for Probabilistic Composite
The paper describes the reliability study that has been System Evaluation”, 14th Inter-RAM Conference for the Electric
performed at Hydro One to review the adequacy of the Power Industry, Toronto, May 26-29, 1987.
current spare numbers for the 230/115 kv, 3 phase auto-
transformers of the Hydro One’s bulk transmission system.