Professional Documents
Culture Documents
Abstract—Composite system preventive maintenance scheduling ownership can continue to be held by the original owners
is a challenging task in both vertically integrated and deregulated in the vertically integrated structure. Distribution companies
utility systems. The functional separation of generation and trans- (DISCOs) move energy at the retail level and may aggregate
mission in the new market environment creates operational and
scheduling problems related to preventive maintenance. Mainte- retail loads. These entities must work cooperatively to provide
nance schedules must be coordinated through an independent en- cost-effective and reliable electric power supply. Independent
tity designated as the Independent System Operator (ISO) to as- entities designated as ISOs coordinate the activities of the
sure reliable and economical service. The methods adopted by an GENCO, the TRANSCO, and the DISCO to achieve the overall
ISO to coordinate planned outages are normally based on tradi- goal of serving the customers.
tional load flow and stability analysis and deterministic operating
criteria. This paper proposes a method designated as the main- The ISO should collect all generation and transmission
tenance coordination technique (MCT) to coordinate composite planned outage requests from market participants, i.e., the
system maintenance scheduling in a deregulated utility system. GENCO and the TRANSCO within a certain time (usually
Index Terms—Composite system, deregulation, maintenance declared by the ISO) prior to the start of the outages. It should
scheduling, reliability. review all submissions of planned outages based on relia-
bility criteria and the time/date of request for maintenance
and then decide whether to permit, deny, or adjust planned
I. INTRODUCTION outage schedules to preserve the system reliability. The electric
utility industry is moving to new planning criteria in the new
C OMPOSITE system preventive maintenance scheduling is
a challenging task in both vertically integrated and dereg-
ulated utility systems. Electric power systems have tradition-
market environment where broader engineering considerations
of transmission access and risks must be explicitly addressed.
ally been organized and operated as regulated monopolies. In Specifically, the likelihood of the occurrence of worst possible
these cases, an electric power utility or entity owns and oper- scenarios must be recognized in the analysis and the acceptable
ates the generation, transmission, and distribution systems and risk levels incorporated in the decision-making process [1].
therefore controls all aspects of system planning, design, and Intense competition in power markets will result in more com-
operation. The power industry is now undergoing considerable plicated facility maintenance scheduling and create additional
changes due to deregulation. The main aim of restructuring is pressure on the GENCO and the TRANSCO to create optimal
to let market forces drive the price of electric supply and re- maintenance schedules for their facilities. It is imperative
duce the net cost through increased competition. Restructuring to develop efficient decision-making tools for the GENCO,
creates an open market environment by allowing competition in the TRANSCO, and the ISO to create the most appropriate
power supply and allowing consumers to choose their supplier maintenance schedules in a competitive situation.
of electric energy. Maintenance is an important part of asset management [2].
In the new structure, generation companies (GENCOs) can It is commonly divided into two categories: preventive main-
be separately owned and compete to sell energy to consumers tenance and corrective maintenance. The former is also called
and are no longer controlled by the same entities that control the planned maintenance or scheduled maintenance and deals with
transmission system. Transmission companies (TRANSCOs) scheduled outages. The latter usually includes repair and re-
are regulated to provide nondiscriminatory connections and placement and deals with forced outages or random failures.
comparable service for cost recovery. A TRANSCO has the role The basic objective of preventive maintenance is to prevent or
of building, owning, maintaining, and operating the transmis- forestall future random failures of the system facilities by re-
sion system in a certain geographical region to provide services moving these facilities from service at an appropriate time and
for maintaining the overall reliability of the electrical system. conducting diagnostic tests and element replacements. An op-
The Independent System Operator (ISO) handles the operation timized maintenance schedule can improve system reliability,
and scheduling of TRANSCO facilities. The use of TRANSCO reduce operating costs, and result in savings in capital invest-
assets is under the control of the regional ISO, although the ment for new facilities [3].
After determining the individual component maintenance re-
quirements, it is necessary to coordinate all the maintenance re-
Manuscript received May 21, 2004. Paper no. TPWRS-00617–2003. quests in terms of their impact on the system. Considerable re-
The authors are with the Power System Research Group, Uni- search has been done at the generation level[4]–[6]. Relatively
versity of Saskatchewan, Saskatoon, SK, Canada S7N 5A9 (e-mail:
Roy_billinton@engr.usask.ca; ram264@mail.usask.ca). little work has been reported in the available literature on com-
Digital Object Identifier 10.1109/TPWRS.2004.840449 posite system preventive maintenance scheduling and coordina-
0885-8950/$20.00 © 2005 IEEE
2 IEEE TRANSACTIONS ON POWER SYSTEMS
tion [7]–[9]. This is an interesting and important topic in both violating the system reliability. At the present time, the usual
vertically integrated and deregulated utility systems. procedure to assess the impact of a possible maintenance event
This paper proposes a method designated as the maintenance is to conduct power-flow studies under these conditions. These
coordination technique (MCT) to coordinate composite system studies are deterministic, in that they utilize an criterion
maintenance scheduling in a deregulated utility system. The to assess the acceptability. The criterion implies that the
MCT is demonstrated on the two test systems using com- system can withstand the loss of any single element without en-
mercial software known as MECORE [10]. The MECORE countering any difficulty. These studies do not include the likeli-
program is a Monte Carlo-based composite system reliability hood of element failures and the actual adequacy of the system
evaluation tool. It was initially developed at the University under the maintenance event. The criterion itself is consistent
of Saskatchewan and subsequently enhanced at BC Hydro. with regard to ; the resulting system risk is, however, quite
It can be utilized to conduct a variety of composite system variable. Incorporating the likelihood of element failures will
reliability and reliability worth studies. MECORE is based highlight actual high-risk situations when scheduling facility re-
on a combination of Monte Carlo simulation (state sampling moval for maintenance. In order to do this, the ISO should estab-
technique) and enumeration techniques. The program can be lish system and load point reliability criteria that can be used to
used to simulate the system component states and to calculate assess the adequacy of the system and load points when specific
annualized indices at the system peak load level. A hybrid facilities are removed from service. Most scheduling is done on
method utilizing an enumeration approach for aggregated load a weekly basis [3], although longer or shorter intervals can be
states is used to calculate annual indices using an annual load used if required. The peak load for the week is assumed to be a
curve [10]. constant value over the period. This is forecast in advance of the
actual occurrence. The risk criteria are a management decision.
The studies described in this paper use the annualized values
II. COMPOSITE SYSTEM MAINTENANCE
of the base case system and load point expected energy not sup-
COORDINATION TECHNIQUE
plied (EENS) as the criterion risks. The risk for the maintenance
It is difficult to coordinate all the component maintenance period in question—in this case a specified week—is also annu-
requirements in the new utility environment. Generation and alized in order to compare it with the annualized criterion risk.
transmission maintenance requests are decoupled in a deregu- This results in the same maintenance decision as expressing the
lated environment. The decision when to maintain a generator annualized criterion risk on the specified period basis and com-
is determined by the individual GENCO rather than by the op- paring it with the specified period risk. Generating unit mainte-
timal cost of maintenance and repair in the overall system. This nance outages may extend over several weekly periods. These
is not the case in a conventional vertically integrated utility, outages are considered in each weekly period together with the
where there is no competition between the various supply facil- relevant transmission outages in the particular period.
ities. This paper assumes that the generation and transmission As noted earlier, weekly maintenance periods or windows
facilities are completely decoupled and that the ISO is respon- have been used in the studies described in this paper. Shorter
sible for the overall integrity of the system. This could require periods can be used, if desired, or if it is necessary to determine
some companies to delay their maintenance to a more suitable the risk in an extended period involving mixed length mainte-
time. The process obviously has to be transparent and one in nance requests. In this case, the risk (EENS) is calculated for
which no individual company is treated unfairly. This may re- each segment or subperiod in the overall period in question. The
quire the ISO to utilize additional information in the form of risk over an extended period is the sum of the risks in the sub-
forecast operating costs or price signals, such as the forward periods. The subperiods can be of any length. The focus in the
price for a given future week, to assist in deciding which main- initial application described in this paper is on a single period of
tenance requests should be accepted and which requests should one week. The MECORE software used in the studies and dis-
be delayed. Excessive delays will result in increased system risk cussed further in the following section is a state sampling tech-
at future times. There may also be other engineering considera- nique. More detail could be included in a sequential simulation
tions external to the reliability model that dictate certain restric- approach but with a considerable increase in computation time.
tions on maintenance activity. The focus in this paper is on a Relatively little work has been reported on sequential simula-
future week in which any external considerations are known or tion applied to composite system reliability assessment [11].
are forecast. The actual occurrence of an external event during a The system and load point reliability are a function of the
particular week may require a previously forecast maintenance system load level, and therefore, there will be some periods
schedule to be cancelled in response to the resulting situation. (weeks) in which certain equipment removals are acceptable and
This paper presents a procedure to assist in the maintenance some periods (weeks) in which their removals lead to violation
scheduling of generation and transmission facilities in a bulk of the risk criteria.
electric system. This approach is designated as the MCT. The This is illustrated in Fig. 1, in which there are two designated
MCT is based on practical procedures used by most ISOs. As areas. The one below the criterion risk level is the acceptable
indicated earlier, an ISO should collect all the generation and area. The other area above the criterion risk level is the unaccept-
transmission planned outage requests from the market partici- able area. Fig. 1 shows the variation of the risk with increasing
pants at an agreed time (usually declared by the ISO) prior to the load. The intersection of the criterion risk and the risk profile
start of these activities. The ISO must then determine which of occurs at the critical load level. Any load higher than this will
these maintenance activities can proceed as requested without violate the risk criterion. Any load level less than this has an
BILLINTON AND MO: COMPOSITE SYSTEM MAINTENANCE COORDINATION IN A DEREGULATED ENVIRONMENT 3
TABLE II
LOAD POINT RELIABILITY INDICES FOR THE IEEE-RTS
TABLE I
LOAD POINT RELIABILITY INDICES FOR THE RBTS
TABLE III
acceptable risk, and therefore, the system configuration associ- SYSTEM RELIABILITY INDICES FOR THE RBTS AND IEEE-RTS
ated with the risk profile is acceptable at these load levels. The
risk profile shown in Fig. 1 will change as different generation
and transmission facilities are removed from service. The risk
profile will also change slightly if load forecast uncertainty is in-
corporated in the evaluation. The risk in each case will increase
with increased uncertainty.
Fig. 2. System EENS of the RBTS as a function of the load level for the
selected cases.
Fig. 3. Bus 3 EENS of the RBTS as a function of the load level for the selected
cases.
TABLE IV
AVAILABLE WEEKS FOR THE SELECTED MAINTENANCE OUTAGES BASED
ON THE SYSTEM EENS
TABLE V
COMPARISON OF THE AVAILABLE PERIODS BASED ON THE LOAD POINT AND
THE SYSTEM EENS FOR THE SELECTED CASES
Fig. 6. System EENS as a function of the load level (cases G1-40, L1-3, and
G1-40-L1-3).
TABLE VII
AVAILABLE WEEKS FOR THE SELECTED MAINTENANCE OUTAGES BASED
ON THE SYSTEM EENS
Fig. 5. System EENS as a function of the load level (cases G1-40, L1, and
G1-40-L1).
Fig. 7. System EENS as a function of the load level for the six cases.
TABLE VI
AVAILABLE WEEKS FOR SELECTED MAINTENANCE OUTAGES does not significantly increase the risk. It should be appreciated
BASED ON SYSTEM EENS that the risk associated with G1-40-L1 cannot be obtained by
simply adding the risks associated with G1-40 and L1.
Fig. 6 shows the system EENS as a function of load level for
the three cases of G1-40, L1-3, and G1-40-L1-3. The weeks in
which these maintenance removals can be conducted are given
in Table VII. The risk associated with G1-40-L1-3 is higher than
those of the other two cases, and the opportunities for mainte-
risks. Determination of the criterion risk at a load point is a prac- nance in these three cases are also different.
tical management issue. It is important from a load point view- In practice, it is possible that different generation and trans-
point to check for unacceptable conditions created using overall mission element owners may request that different facilities be
system analysis. removed for maintenance in the same time period. From their
Fig. 5 shows the system EENS as a function of the load own perspectives, these removals are acceptable. From a system
level for three cases (G1-40, L1, and G1-40-L1). The weeks in point of view, however, they may not be acceptable. For in-
which these maintenance removals can be conducted are given stance, it can be seen from Table VII that G1-40 and L1-3 are
in Table VI. It can be seen from Fig. 5 and Table VI that the risk acceptable in weeks 28 and 29 from an individual point of view,
associated with removing one 40-MW unit at Bus 1 is much but G1-40-L1-3 is unacceptable during these two weeks from
higher than that associated with removing Line 1, and therefore, a system viewpoint. This clearly indicates that an overall body
there are more opportunities available for maintenance on Line such as an ISO should coordinate the many possible requests for
1. The difference between the risks associated with G1-40 and maintenance removals.
G1-40-L1 is very small, which indicates that the G1-40-L1 risk Fig. 7 shows the system EENS as a function of load level
is dominated by G1-40. In other words, whenever one 40-MW for another six cases (G1-40, L4, L8, and some of their com-
unit at Bus 1 is removed for maintenance, removing Line 1 binations). The weeks in which these outages can be done are
6 IEEE TRANSACTIONS ON POWER SYSTEMS
TABLE X
SYSTEM EENS (IN MEGAWATTHOURS PER YEAR) OF THE MRTS AT DIFFERENT
LOAD LEVELS FOR THE SELECTED CASES
Fig. 8. System EENS of the IEEE-RTS as a function of the load level for the
selected cases.
shown in Table VIII. It can be seen from Fig. 7 and Table VIII
that L4 can be done at any time of the year. Cases G1-40 and
G1-40-L4 have almost the same response to the load variation
and therefore the same opportunities for planned outage. When
one 40–MW unit at Bus 1 is removed for maintenance, the re-
moval of Line 4 minimally increases the system risk. This is not
the case for L4-8. Case G1-40-L4-8 is quite different from either
G1-40 or L4-8. G1-40-L4-8 is unacceptable at any time. This re-
inforces the point made earlier that the risk associated with two
maintenance requests cannot be assessed by simply summing
the risks associated with each individual request.
are more sensitive to the load level than G18-400 and G23-350,
V. APPLICATION OF THE MCT TO THE IEEE-RTS
and the impact of removing an additional 197-MW unit is sig-
The RBTS is a relatively small system, and many factors con- nificant. The impact of removing transmission elements in ad-
strain removing elements for maintenance. The IEEE-RTS is dition to generating units is seen in G18-400-13-197-L2-8-9-12
relatively large compared to the RBTS. It has a strong transmis- (removing one 400-MW unit at Bus 18 and one 197-MW unit
sion network and a weak generation system. It has more room at Bus 13 as well as Lines 2, 8, 9, and 12). This condition has a
for removing elements, especially transmission lines, from the similar critical load to G18-400-13-197 and the same possible
system for maintenance than does the RBTS. The weekly peak weeks for maintenance. This is also the case for G23-350-13-
loads are given in [13]. 197-L2-8-9-12 and G23-350-13-197.
Fig. 8 shows the system EENS of the IEEE-RTS as a func- In order to stress the transmission network, the original
tion of the load level for the selected cases. The weeks in which IEEE-RTS generating units and load profile were doubled with
these outages can be done are shown in Table IX. It can be seen the transmission system unchanged. The total capacity of the
from Fig. 8 and Table IX that the six cases that involve removing modified IEEE-RTS (MRTS) is 6810 MW with a peak load of
transmission have little impact on the system EENS and can be 5700 MW. The system EENS of each case for the MRTS at the
done in any week of the year except week 51. This again in- different load levels are shown in Table X. The corresponding
dicates that the IEEE-RTS has a very strong transmission net- curves are presented in Fig. 9, except L5. The weeks in which
work. Cases G18-400 and G23-350 are sensitive to the load these maintenance outages can be done are given in Table XI.
level, and there are fewer opportunities than for the six transmis- It can be seen from Table X that the transmission system of the
sion cases. Cases G18-400-13-197 (removing one 400-MW unit MRTS is stressed significantly and that some line removals are
at Bus 18 and one 197-MW unit at Bus 13) and G23-350-13-197 restricted. For example, removing Line 5 is unacceptable when
BILLINTON AND MO: COMPOSITE SYSTEM MAINTENANCE COORDINATION IN A DEREGULATED ENVIRONMENT 7
TABLE XI
AVAILABLE WEEKS FOR THE SELECTED MAINTENANCE OUTAGES BASED
ON THE SYSTEM EENS OF THE MRTS
America. Relatively little transmission is being built or pro- [3] M. K. C. Marwali and S. M. Shahidehpour, “Long-term transmission
posed in the near future. Under these circumstances, reliability and generation maintenance scheduling with network, fuel and emission
constraints,” IEEE Trans. Power Syst., vol. 14, no. 3, pp. 1160–1165,
will degrade, and removing elements for maintenance will be Aug. 1999.
restricted significantly as load grows and additional generation [4] F. El-Sheikhi, “Maintenance Scheduling of Generating Facilities in
is added. Single and Interconnected Power Systems,” Ph.D. dissertation, Univ.
Saskatchewan, Saskatoon, SK, Canada, 1983.
[5] A. Abdulwhab, “Generating Unit Maintenance Scheduling Using Prob-
VI. CONCLUSIONS abilistic Techniques,” M.Sc. thesis, Univ. Saskatchewan, Saskatoon, SK,
Canada, 2000.
Removing system elements for maintenance can create sig- [6] , “Incorporating Generating Unit Maintenance Scheduling in Power
nificant increases in the system risk. The MCT proposed in this System Reliability Evaluation,” Ph.D. dissertation, Univ. Saskatchewan,
Saskatoon, SK, Canada, 2003.
paper was applied to the two test systems to examine the impact [7] G. J. Anders, G. Hamoud, A. Leite da Silva, and L. Manso, “Optimal
of removing elements for maintenance. The object is to deter- outage scheduling—example of application to a large power utility,” Int.
J. Elect. Power Energy Syst., vol. 25, no. 8, pp. 607–614, June 2003.
mine if a certain planned outage could be conducted during a [8] W. Li and J. Korczynski, “A reliability based approach to transmission
designated period. maintenance planning and its application in BC hydro system,” in Proc.
The analysis described in this paper indicates that different IEEE-PES Summer Meeting, vol. 1, July 2001, pp. 510–515.
[9] M. Shahidehpour and M. Marwali, Maintenance Scheduling in Restruc-
cases have different critical loads, which result in different tured Power Systems. Boston, MA: Kluwer, 2000.
opportunities for the planned maintenance. Some cases can be [10] W. Li, Installation Guide and User’s Manual for the MECORE Program,
done in any week during the year. Some cases can not be done Jul. 1998.
at any time. In certain cases, if one element is removed for [11] A. M. Leite da Silva, L. A. F. Manso, J. C. O. Mello, and R. Billinton,
“Pseudo-cbhronological simulation for composite system reliability
maintenance, another element can be removed simultaneously analysis with time-varying loads,” IEEE Trans. Power Syst., vol. 15, no.
without significantly increasing the risk. Generally, removing 15, pp. 73–80, Feb. 2000.
more components from service results in the related curves [12] R. Billinton et al., “A reliability test system for educational pur-
pose—basic data,” IEEE Trans. Power Syst., vol. 4, no. 3, pp.
moving to the left, which means that the corresponding risks 1238–1244, Aug. 1989.
increase, and the weeks available for the maintenance decrease. [13] IEEE-RTS Task Force of APM Subcommittee, “IEEE reliability
test system,” IEEE Trans. Power App. Syst., vol. PAS-98, no. 6, pp.
Planned maintenance outages, which are acceptable based on 2047–2054, Nov./Dec. 1979.
the system risk, may be unacceptable based on load point risks.
Determination of the criterion risks, particularly for load points,
is a practical management issue and can have a large impact on
maintenance scheduling decisions. It is important to appreciate Roy Billinton (S’59–M’64–SM’73–F’78) came to Canada from England in
that it is necessary from a load point perspective to check for 1952. He received the B.Sc. and M.Sc. degrees from University of Manitoba,
unacceptable conditions created by using system risk criteria. Winnipeg, MB, Canada, and the Ph.D. and D.Sc. degrees in electrical engi-
neering from the University of Saskatchewan, Saskatoon, SK, Canada.
He joined the University of Saskatchewan in 1964. He is the author and coau-
APPENDIX thor of eight books on reliability evaluation and over 800 papers on power
system reliability evaluation, economic system operation, and power system
Fig. 10 shows the single line diagram of the RBTS, and Fig. analysis.
Dr. Billinton is a Fellow of the EIC, the CAE, and the Royal Society of Canada
11 shows the single line diagram of the IEEE-RTS. and a Registered Professional Engineer in the Province of Saskatchewan.
REFERENCES
[1] R. Billinton, L. Salvaderi, J. D. McCally, H. Chao, T. Seitz, R. N.
Allan, J. Odom, and C. Fallon, “Reliability issues in today’s electric Ran Mo was born in Harbin, China, and received B.Sc. and M.Sc. degrees from
power utility environment,” IEEE Trans. Power Syst., vol. 12, no. 4, pp. the Harbin University of Science and Technology in 1984 and 1990, respec-
1708–1714, Nov. 1997. tively, and the M.Sc. degree in electrical engineering from the University of
[2] J. Endrenyi et al., “The present status of maintenance strategies and the Saskatchewan, Saskatoon, SK, Canada, in 2004.
impact of maintenance on reliability,” IEEE Trans. Power Syst., vol. 16, He worked for the Research Institute of Machinery Science and Technology
no. 4, pp. 638–646, Nov. 2001. and the China Aero Polytechnology Establishment, Beijing, China.