Attachment 2
Palo Verde to Westwing Double Line Outage Probability Analysis
SRP
Tatyana Len Dhaliwal [email protected] <mailto:[email protected]>
Executive Summary
This report details the mitigating factors and a double contingency outage analysis of the
Palo Verde to Westwing lines 1 and 2. This outage is considered of such low probability
of occurrence and recurrence, that it warrants submittal to the WECC Phase I
Probabilistic Based Reliability Criteria (PBRC) Performance Category Evaluation (PCE)
Process. Under this process, a project with an accepted Mean Time Between Failure
(MTBF) in the range of 30 to 300 years may be adjusted to Category D, but with the
added condition of “no cascading” allowed. A project with a MTBF in excess of 300
years is considered an “Extreme Event” in the same sense as all other events in the NERC
Category D.
This report follows the Probability Reliability Evaluation Work Group (RPEWG)
recommended steps provided in Appendix I, Figure 10 RPEWG Recommended Analysis
Steps.
Analysis of the Palo Verde to Westwing Lines 1 and 2 double contingency (N-2)
qualifies to be moved to Category D based on the following statistical analysis and
mitigating factors:
1)
2)
3)
4)
5)
6)
An MTBF estimated by a traditional statistical reliability analysis method is
on average once in 2824 years.
In the 11 years of accurately recorded outage history in electronic format,
there has never been a double contingency outage of the Palo Verde to
Westwing lines. Evidence suggests that since both lines were in service, this
outage has not ever occurred.
Both Westwing and Palo Verde switchyard use breaker and a half
arrangement.
As a result of the Rudd line installation, the Palo Verde to Westwing lines 1
and 2 outage is no longer the most critical outage.
According to UFSAR, the failure of this line at the crossing over the PV-WW
1 and 2 is no longer postulated under the revised 10CFR50.59 rules.
Therefore grid studies need not address this scenario.
The Robust design features are overhead ground wires, lines are built 130 feet
apart (centerline to centerline) with towers designed to fail in the middle. The
failure and fall of one tower does not jeopardize the continued safe operation
of the other tower.
7)
8)
9)
10)
11)
12)
13)
14)
15)
16)
Palo Verde to Westwing 500kV lines are located outside the areas of
consideration for air traffic. The elevation of the lines in beyond and beneath
the criteria FAA defines for consideration as an obstacle or hazard.
The isokeraunic level near Palo Verde and Westwing is one of the lowest in
the Western US, ranging from 1.0 strikes per square mile per year near Palo
Verde to 2.5 strikes per square mile per year near Westwing switchyard.
The risk of earthquakes in Maricopa County is the lowest in the Western US.
The risks of flood, snow, and fire are negligible.
The PV-WW foundations are over designed in the range of 137 to 199%.
The lattice tower design is conservative for weather related loads.
Lines are designed with state of the art spacer dampers to control conductor
motion.
The insulation level exceeds EPRI’s guidelines.
Electronic protection is provided by redundant microprocessor based
technology with communication via fiber optics and digital microwave
systems on independent paths. A third microprocessor based relay system
operating in current differential scheme is provided for backup protection.
SRP aggressively maintains the lines with twice yearly patrols, bird guard
systems in place, an insulator-washing program, and a spacer damper
replacement program.
In summary, based on an MTBF estimated by traditional statistical reliability analysis of
2824 years and excellent design and maintenance practices, it is recommended that this
N-2 outage be moved to Category D (Extreme Events) with no other conditions or
requirements.
Description of the Palo Verde to Westwing Lines 1 and 2 Path.
SRP (Salt River Project) is a major multipurpose reclamation project comprising two
principal operating groups: the Salt River Project Agricultural improvement and Power
district, a political subdivision of the state of Arizona; and the Salt River Valley Water
Users’ Association, a private corporation. The district provides electricity to more than
625,000 customers in the Phoenix area. It operates or participates in seven major power
plants and numerous other generating stations, including thermal, nuclear, and
hydroelectric sources. The District serves a 2,900-square- mile area spanning portions of
Maricopa County (the metropolitan Phoenix Area), Gila, and Pinal counties in central
Arizona.
The two 500kV transmission lines discussed in this report connect the Palo Verde
Nuclear Generating Station near Wintersburg to the Westwing Receiving station, northnortheast of Sun City West. The line length is 45.1 miles. The lines are located to the
west of the Phoenix metropolitan area. Below is the photograph of the Westwing yard
looking south; the Palo Verde to Westwing lines are to the right.
Below, are two photographs of the Palo Verde to Westwing lines towers.
The photo below depicts a tower on Palo Verde to Westwing Line 1 near the Westwing
substation.
The photo above depicts a tower on Palo Verde to Westwing Line 2 near the Westwing
substation.
From the Westwing substation the lines proceed west for 19 miles, then south west for
3.1 miles, then 9.5 miles in the south direction, then 8 miles in the south west direction, 4
miles west- south west, and finally west for the last 1.5 miles to Palo Verde yard, as
shown in the diagram below:
Two to three miles outside of the Westwing substation, the Mead to Phoenix 500kV line
crosses over the two Palo Verde to Westwing lines. Here are some photographs of the
crossing:
27.6 miles from Westwing, the two Palo Verde to Westwing lines cross over the Liberty
to Mead 345kV line, as shown below:
29.6 miles from Westwing, the two Palo Verde to Westwing 500kV lines cross over the
Harcuvar to Buckeye 230kV line and the Parker to Liberty 230kV line, as shown below:
The new Palo Verde to Rudd line is in the same corridor with the two Palo Verde to
Westwing lines for several miles outside of Palo Verde switchyard. The Rudd 500kV is
the rightmost line.
Here is the photo of the Palo Verde to Westwing lines (the two rightmost lines)
emanating from the Palo Verde switchyard.
System Configuration and Fault Analysis
Palo Verde – Westwing Double Line Outage Reclassification
The Palo Verde – Hassayampa Hub area is a major electrical energy trading hub in the
United States. There are currently Independent Power Producer generators at the Palo
Verde hub in addition to the existing Palo Verde Nuclear Generating Station. This year
there are 230kV & 500kV transmission system enhancements and new generator
interconnectors synchronizing to the transmission system.
Generators
Palo Verde Nuclear(existing)
Duke Arlington Valley(existing)
Pinnacle West Red Hawk(existing)
Sempra Mesquite(2003)
PGE National Harquahala(2003)
TECO Panda Gila River(2003)(network)
Total Generation Value
Current
Total
3861MW
600MW
1000MW
525MW
1148MW
1560MW
8694MW
Current
Equivalized
3861MW
600MW
1000MW
525MW
1148MW
780MW*
7914MW
Jun/2003
Equivalized
3861MW
600MW
1000MW
1050MW
1148MW
1040MW*
8699MW
*The TECO Panda Gila River interconnection is not a radial connection to the Palo
Verde – Hassayampa hub. This interconnection has a 230kV tie into the Gila Bend –
Liberty circuit. This generation interconnection has a reduced interaction, 0.5 to 1 as
compared to the direct interconnectors, at the Palo Verde – Hassayampa hub, as
measured in enhancing generating capability and requirement of curtailment. As
evidenced by the above list, there is much existing and new interconnecting generation at
the Palo Verde – Hassayampa hub.
There are currently 6 - 500kV circuits emanating out of the Palo Verde – Hassayampa
hub; Palo Verde – Westwing #1 & #2, Hassayampa – Jojoba – Kyrene(Palo Verde East)
& Palo Verde – Devers, Hassayampa – North Gila(Palo Verde West) and Palo Verde to
Rudd lines. Operating studies were performed in 2002 to determine the amount of
generation that could be safely interconnected to the Palo Verde – Hassayampa hub
without Remedial Action Scheme (RAS) and with RAS. Spring 2003 studies were
performed with the same methodology, to determine the amount of generation that could
be in service with and without RAS. The conclusions of this Spring 2003 study is a little
different in that there is a thermal limit that is the upper bound of allowable safe
generation. This limit is a simultaneous loading of the 500kV lines to the east and to the
west of the Palo Verde – Hassayampa hub. Only 7,301MW of equivalized generation
may operate at one time, this is the thermal limit. The thermal limit is an n-0 limit. The
interconnection of a 500kV circuit from Palo Verde to Rudd, a local load serving circuit
to serve SRP and APS load was energized in June of 2003 and has increased the thermal
capacity of the 500kV circuits emanating from the Palo Verde – Hassayampa hub.
The amount of MVARs that are produced or absorbed by the Palo Verde nuclear units
determines the amount of generation that has to be used in the RAS for safe operation. As
the amount of MVARs absorbed by Palo Verde increases, the amount of interconnector
generation that has to be used in the RAS increases. The most limiting condition studied
was with the Palo Verde 500kV bus at 525kV and the Palo Verde units absorbing
800MVARs. These constraints come from the operators of the Palo Verde Nuclear
Generating Station.
New interconnector generation, a 4th Panda Gila River unit and the 2nd Mesquite unit are
synchronizing after the Rudd line in-service date. Therefore, the rightmost column in the
above table will also have the new 500kV circuit in service.
The simultaneous outage of the Palo Verde – Westwing circuits is initiated by a single
line to ground fault at the Palo Verde bus. This is a Category C outage per the WECC
Table W-1. Category B, three phase faults applied at the Palo Verde bus with a single
circuit contingency outage are less restrictive than this Category C outage. Of the three
standards: transient voltage dip, minimum frequency or post transient voltage deviation,
the transient voltage dip is the limit in the studies. The 500kV bus voltage is monitored at
Palo Verde, Kyrene and North Gila.
In the WECC Table W-1, circuit outage frequency is associated with the different
categories. Also in the WECC Standards is the application of Probabilistic Based
Reliability Criteria (PBRC). Applicable to this analysis is the WECC Standard I.A.S5,
which is the application of the PBRC process for adjustment of the Performance
Category of Table W-1. That is, this study is to prove that the outage frequency of the
Palo Verde – Westwing lines simultaneous double circuit outage is greater than what
classifies this particular outage to Category C. The proof must show that the MTBF > 30
years. This will classify this outage as Category D. Further, by WECC Standard I.A.S6, if
the MTBF > 300 years then cascading is allowed.
During the course of the 2003 Summer Palo Verde Transmission System Operating Study
a new system configuration was taken into account. This new system configuration is
shown in Figure Appendix II – Figure I.. The new Palo Verde to Rudd 500kV line is a
joint APS/SRP project to interconnect the Palo Verde Switchyard and Rudd Receiving
Station in the Southwest Valley to serve the Phoenix area load.
The thermal capacity of the Palo Verde transmission system was studied with the total net
generation of 9,595 MW. For the Hassayampa to North Gila three-phase fault at the
Hassayampa 500kV Bus with the outage of the Hassayampa to North Gila 500kV line a
large system voltage excursion in the California power system, in particular in the Devers
area, could be experienced.
As a result of the Rudd line installation, the Palo Verde to Westwing lines 1 and 2 outage
is no longer the most critical outage. The 2003 Summer Operating Study proved that the
new critical outage is the Hassayampa to North Gila 500kV line outage. According to
SRP’s Power Operations Center the Hassayampa to North Gila 500kV line is out of
service on scheduled maintenance for about 2 days per year. Each individual scheduled
maintenance outage is limited to 1 day. In addition, the line may be out on an
unscheduled maintenance for a couple more days during the year, but the unscheduled
maintenance amount varies from year to year. The combined scheduled and unscheduled
outage time, according to SRP’s Power Operations, will be 150 to 200 hrs per year total
for Palo Verde – Westwing #1 , Hassayampa – Jojoba – Kyrene(Palo Verde East) & Palo
Verde – Devers, Hassayampa – North Gila(Palo Verde West) and Palo Verde to Rudd
lines.
The conclusion of this report is that the MTBF = 2824 years, which is greater than MTBF
of 300 years. Therefore, cascading is permissible if this outage were to occur. However,
the existing RAS scheme can respond to this outage. Therefore, the situation during
which the Palo Verde to Westwing 500kV line outage becomes the most limiting
contingency occurs for 200 hours per year at most. SRP proposes to maintain the RAS
scheme for this, however improbable, condition to protect against this outage. This
would then be a system safety net.
The table below shows that with any one of the circuits emanating from Palo Verde
initially out of service, a subsequent SLG fault on the Palo Verde 500kV Bus which
would take both PV-WW 500kV lines out of service will not exceed the 30% voltage dip
criterion.
From the diagram of the Palo Verde to Westwing Path presented on the third page of the
Description of Path section, one can note that the Mead-Phoenix 500kV line crosses over
the PV-WW 1 and 2 a few miles from the Westwing Receiving station. According to
UFSAR, the failure of this line at the crossing over the PV-WW 1 and 2 is no longer
postulated under the revised 10CFR50.59 rules. Therefore grid studies need not address
this scenario. (You may contact Harvey Leake at 623-393-6986 for information
pertaining to this issue).
SINGLE-LINE-TO-GROUND FAULT AT THE PALO VERDE 500KVBUS
TRIPPING THE TWO PALO VERDE/WESTWING 500KV LINES
(With the RAS scheme in service all cases are within the required 30% voltage drop
criteria)
500KV LINE IOS
PRE-RAS
APPROXIMATE
RAS
GEN TRIPPED
(MW)
POST- RAS
VOLTAGE DROP
(WORST CASE)
HASSAYAMPA/NORTH GILA
CASE DIVERGED
(AFTER 3 SEC RUN)
1550MW
22%
KYRENE/JOJOBA
CASE DIVERGED
(AFTER 3 SEC RUN)
1500MW
24%
PALO VERDE/RUDD
CASE DIVERGED
(AFTER 3 SEC RUN)
1070MW
26%
PALO VERDE/DEVERS
CASE DIVERGED
(AFTER 3 SEC RUN)
600MW
19%
PALO VERDE/WESTWING 2
9%
0MW
NA
The lines are ranked in order of worst outage to the outage with less dramatic effects.
IOS = Initially out of Service
1.
2.
3.
4.
2003 SUMMER
PV/HAA GEN NET=8192MW W/ PV = -436MVAR, HAA = -364MVAR (NET –800MVAR).
SLG FAULT ON PV 500KV BUS; WITH TWO PV-WWG CKTS OUT.
GENERATION TRIPPED AT 7CYC
MONITORED VOLTAGES: ARIZONA, CALIFORNIA, UTAH, AND NORTHWEST
Mitigating Factors
I. Aircraft Operations in the Area of the Palo Verde to Westwing 500kV Lines:
The Palo Verde to Westwing 500kV lines were reviewed by a licensed pilot for hazard
location and elevation. The pilot found that these lines do not qualify as a hazard to
Public Use or Military as defined by the FAA. The four areas of consideration to
determine if it were to qualify as a hazard are: location relative to Public Airports,
Military Airports, En-route operations and Aerial application operations. The review was
based on current FAA Standards.
Public Use Airport – The closest public airport to the transmission line is the Buckeye
Municipal Airport. It is located approximately 3.2 nautical miles (nm.) south east of the
line. The airport is a VFR (visual flight rules) airport with a runway length of 4300 ft.
The runway fits the requirement for obstacle clearance considerations because it is a
visual runway. Obstacle clearance criteria are based on the runway classification and
five other areas of consideration. The 500kV lines are located outside the areas of
consideration and the elevation of the lines is below all criteria the FAA defines for
consideration as an obstacle or hazard.
Military Airport - Luke Air Force base is located approximately 9 nm. south of the
500kV lines. The Air Force base has even more stringent obstacle clearance criteria. Due
to the distance from the Air Force base, the 500KV lines are both beyond and beneath
any area of consideration.
En-route – The transmission lines are well below any criteria to be considered as an
obstacle to an en-route IFR (Instrument Flight Rules) airway and are not located in any
common corridor for visual operation.
Aerial Crop Dusting Application – The line is located to the west and north of the White
Tank mountains. This area is not a developed agricultural area and no aerial application
activities have been observed in this area.
II. Analysis of natural hazards on the Palo Verde to Westwing 500kV line Right of Way
and at the Mead-Phoenix line and Hassayampa area crossover locations.
SRP’s Water Resource operation group prepared an evaluation of the natural hazards
that could impact the two transmission lines from Palo Verde Nuclear Generating Station
near Wintersburg to the Westwing Receiving station, north-northeast of Sun City West.
Included are two sites where the Palo Verde to Westwing line crosses two other
transmission lines. One site is where the Palo Verde to Westwing lines crosses over the
Mead-Liberty 345kV line east of the Hassayampa River and west of the White Tank
mountains. The second site is west of the Westwing Receiving station where the Mead to
Phoenix 500kV line crosses over the two Palo Verde to Westwing 500kV lines.
M-1
A. Lightning and Thunderstorms
Thunderstorms are most numerous during the summer from late June to September.
Lightning flash density tends to decrease from east to west over the Phoenix metro area.
(A study by the National Severe Storms Laboratory (1995), using data from 1984 and
1987-93, shows the following flash densities (see Appendix 1, Figures 1 and Figure 2):
Westwing area
2.5 strikes per square mile per year
Hassayampa area
1.5 strikes per square mile per year
Palo Verde NGS area 1.0 strikes per square mile per year
In contrast, the flash density over north Scottsdale, AZ in the far northeast Valley is 4.0
strikes per square mile per year.
Luke Air Force Base is located 17 miles east of the Hassayampa site, 10 miles south of
Westwing and 30 miles east-northeast of Palo Verde NGS. Luke AFB has an average of
26 days per year when thunder is heard (with a minimum of 13 and a maximum of 40
days) during the 1964-2001 period.
B. Damaging Winds and Dust Storms
High winds often occur with summer thunderstorms. Potentially damaging winds (50 or
more knots) from micro bursts in the West Valley are rare. Luke AFB records (1946-96)
indicate that a daily peak wind gust between 50 and 64 knots (58-74 mph) occurs only
once in three years. August is the month of greatest likelihood. Wind gusts of 65 knots
(75 mph) or more happen only during the summer and fall, July through October, with
each month having a recurrence interval of once in 33 years. The maximum peak gust
reported at Luke AFB is 88 knots (101 mph) on August 29, 1996.
Thunderstorm winds often stir up dirt from the Valley floor. Luke AFB reported an
average of 17.5 hours with blowing dust per year from 1973 through 1996. The majority
of the hours were in July, August and September. However, SRP has no recorded
transmission outages due to dust storm flashover.
C. Tornadoes
Tornadoes are a very rare event in the Phoenix area. Tornadoes not reaching the ground
are the most common ones, although still rare. One or two tornadoes per year may be
reported in the Valley as a whole. Damage potential of any Valley tornado is small,
estimated at F0 to F1 on the Fujita tornado intensity scale. The National Weather
Service’s Storm Prediction Center estimates that from 1950 through 2001, 61-80 percent
of tornadoes reported in Maricopa County were weak (F0-F1) while only 1 to 20 percent
were strong (F2 or F4).
M-2
D. Earthquakes
According to the U.S. Geological Survey International Residential Code map,
Maricopa County is in the lowest risk Seismic Design Category in the western U.S.,
category B (17 to 33 percent of g (gravity force)). Damage to transmission towers by
earthquake is highly unlikely.
E. Flood and Fire Hazards
No significant threat from fire or flood exists. Vegetation over this low desert area is
sparse and low, generally less than 6 feet high. There are a few areas with slightly
taller trees or bushes (10-15 feet high) and they pose no danger to the power plant or
transmission line towers in case of range fire. Refer to Figure 3 and 4 of Appendix 1
for typical vegetation under the Palo Verde to Westwing lines.
A service hydrologist at the Phoenix National Weather Service office surveyed the
two crossover sites within the last month. The Westwing site is about 3 miles west of
the center of the Aqua Fria river channel. Flow in the Aqua Fria is controlled by
releases from Lake Pleasant through the new Waddell Dam. During the flooding of
February 1980, a peak release of 66,600 cubic feet per second (cfs) was recorded.
This release has a recurrence interval of 47 years. A photomap of that episode shows
that the western edge of the flood flow was at least a mile from the Westwing station
and over 2 miles from the site. A dry wash, which connects to the Aqua Fria several
miles downstream, was noted between two of the towers closest to Westwing station.
Signs of previous high flow were well below the base of the towers.
The Hassayampa site is about 3 miles east of the Hassayampa River channel
centerline. Peak flow on the Hassayampa was 47,500 cfs in 1970. The line towers
appear far enough from the flood channel and would not be structurally threatened
when such high (or higher) flows occur.
F. Snow and Ice Hazard
No threat exists from snow or ice accumulation. No measurable snow was reported at
Luke AFB from 1951 through 1991. No freezing drizzle or rain has ever been
reported on the desert floor of Maricopa County.
III. Design
A. Foundations
A series of inspection reports by the project's geotechnical consultant (SH&B) from
1979 was found in the Palo Verde to Westwing files. These reports noted that the first
41 tower foundation borings were field-tested and it was concluded that the
subsurface conditions were equal to or better than anticipated to resist uplift forces. A
review of the geotechnical engineer's report design charts was done and concluded
M-3
that the footings were designed in accordance with their recommendations for
ultimate load capacity.
The second approach was to compare foundations for Towers 1 thru 71 of this line
with the foundation design for Towers 1003 thru 1075 of the Mead-Phoenix 500kV
transmission line, which are in the same vicinity. It is recognized, that the design for
Palo Verde-Westwing lines I and II was based on soils information and design
methodologies from pre-1978, with the design actually based on 1950's and 1960's
methods. Conversely, the Mead-Phoenix foundation design was based on a
combination of the latest EPRI computer foundation design methods and a
statistically based soils analysis using a much larger database than the earlier work
had available.
Design information from both sets of lines was compared in adjacent sections
(foundation in uplift was the worst case) to determine their relative uplift capacity.
For this set of towers, the PV-WG foundations were determined to be over-designed
in the range of 137% to 199%, depending upon soils type and foundation size. This
means that on the average, the tower foundations for the PV-WG transmission lines
should carry 159% higher load than required by structure design. This value includes
all overload and safety factors.
B. Weather Related Loads
A loads review of the Lattice Towers on the Palo Verde to Westwing lines was
conducted. The towers were designed for NESC Light Load District, Grade B
Construction. This is effectively the minimum design criterion for defining weather
related loads. However these loads are "deterministic" and not loads based on actual
statistically evaluated weather events. To quantify the general structural reliability of
the tangent structures, SRP compared the capacity of these structures as compared to
statistically based loads.
In general, structures with line angle loads are controlled by line tension, not weather
loads, so only tangent structures (5T2 and 5T3 Towers) were evaluated as being most
vulnerable to a weather related loads. This evaluation assumed that developed wind
loads on the towers were based on a review of the tower drawings. Wind areas were
based on rough approximations of quantity of angles in various sections of the tower.
Wind loads on conductors were based on actual spans and elevations on structures.
NESC shape factors were used for lattice towers. A shape factor of 1.2 was used for
the static (dia < .5")
Wind loads were developed per the methods shown in the 2002 edition of the NESC
for Rule 250C application of wind loads (height effect and gust response considered).
Line tensions for winds greater than 60mph were approximated. The angles on the
tangent structures were 2 degrees are less.
M-4
Based on a comparison of approximated tower capacities with actual loads along the
lines:
The MINIMUM Tower capacity is 98mph wind gust
The AVERAGE Tower capacity is 100mph wind gust
Based on 90mph being the 50 year extreme wind event, the respective Return Periods
(RP) are
Minimum Return Period = 135 years
Average Return Period = 150 years
The probability of occurrence of a wind of any RP magnitude is 1/RP in any year
(i.e.: 1/50 probability of a 90mph wind in any year). Therefore the lattice tower
design is conservative for weather related loads. Figure 5 of Appendix I shows a
typical tower footing.
C.
1. Design – Lattice Towers and Insulators
The Palo Verde to Westwing 500kV Lines I and II are constructed on lattice towers,
typical spans of 1300 feet, 130 feet of centerline-to-centerline separation between
lines. The tallest tower is 148ft tall from the ground to the static mast. The towers are
designed to fail in the midsection. A “Hypothetical Tower Failure” analysis is
provided in a separate document. Each circuit constructed with 3-1780kCM 84/19
ACSR “Chukar” conductors per phase and each line has two 7 # 8 Alumoweld
overhead ground wires for lightning protection.
The two lines are located in the Arizona Sonoran Desert west of the Phoenix
metropolitan area. The region is a relatively low isokeraunic level, approximately 30,
that translates into a GFD (Ground Flash Density) of approximately 5.0 flashes per
square mile per year. The Design Criteria required tower footing resistance values to
be 15 ohms or less and typical values were about 5 ohms. The low footing resistance
combined with a good shielding angle and two overhead ground wire provide
excellent lightning performance.
The vegetation in the Arizona Sonoran Desert, where the two Palo Verder to
Westwing lines are located, does not grow of sufficient height to present potential
outage problems due to direct flashover or from fires. If there was a brush fire of
sufficient magnitude, SRP would likely elect to de-energize the line during a fire to
minimize any likelihood of a fault, but the possibility of a fire is extremely unlikely
due to the very small amount of vegetation that could contribute to any wild fire.
SRP’s twice yearly helicopter and ground patrols identify any and all vegetation
problems for corrective action.
The transmission lines are in a region of low contamination, V-String Insulator
Assemblies and 27 porcelain insulators per leg. This insulation level exceeds EPRI’s
“Transmission Line Reference Book, 345kV and Above” criteria from Table 10.2.1
M-5
of 24 insulator units. The three additional units provide supplemental insulation in
the event of damage or punctured units.
These two lines were designed with state of the art spacer-dampers to control
conductor motion. Approximately 8 years ago SRP removed numerous serviced aged
units, tested the units and compared performance with new units. Based on test
results, the spacer-dampers have many years of service remaining. But the twiceyearly helicopter and ground patrols identify damaged units for replacement. SRP’s
aggressive maintenance program identifies damaged insulators, conductor and spacer
dampers prior to potential line outages. SRP stocks replacement or repair material for
these 500kV line and has the trained personnel for either hot-line maintenance or
under outage conditions.
C – 2 Mead-Phoenix 500kV Line Crossing
This section evaluates the design of the Mead-Phoenix 500kV line, specifically the
span that crosses over the two Palo Verde-Westwing 500kV lines. Addressed below
are the SRP design criteria of the phase wire and overhead ground wire (OHGW)
positions.
The conductor of the Mead-Phoenix line is 3-1590kCM 36/7 ACSR/TW “Lapwing”
per phase, with a 49,600 lb. RTS (Rated Tensile Strength) per sub-conductor. Each
sub-conductor was designed for a maximum final tension of 10,165 lbs. at 30º F with
a 9.0 lb wind (~ 60 mph) and a final unloaded tension at 60º F of 8013 lbs. These
tensions are 30,495 lbs (20.5% RTS) and 24,039 lbs. (16.1% RTS) respectively for a
three-conductor bundle per phase. These design tensions are significantly less that
the N.E.S.C. (National Electrical Safety Code) requirement for these conditions of
60% and 25% respectively.
The phase conductor insulator and hardware assemblies have a built-in redundancy,
see the photographs of Figure 1 & 2 below. For example, three strain insulators in
parallel per phase with each insulator rated 80,000 lbs. SML (Specified Mechanical
Load). The assembly could have one or two insulator failures, continue in service
and still meet the N.E.S.C. mechanical strength requirements of 50% of SML.
Furthermore, all hardware (shackles, turnbuckles, extension links, straps, etc.) has a
rated strength of at least two times the final design tension of the phase wire positions
of 30,495 lbs. All hardware is redundant, with the exception of the Yoke Plates
(Rated Strength 150,000 lbs.). In other words, in the event of a hardware failure, the
remaining components have sufficient strength to hold and still meet the N.E.S.C.
minimum requirements.
M-6
Figure One
Mead-Phoenix Dead-End Conductor Assembly
M-7
Figure Two
Mead-Phoenix Dead-End Conductor Assembly
The Overhead Ground Wire is a 7 Number 8 Alumoweld, RTS of 15,930 lbs. SRP
design criteria was for a maximum final design tension of 2743 lbs., at 30º F with 9.0 lb
wind (17% of RTS) and a final unloaded tension value of 1914 lbs. (12% of RTS). These
design tensions are also significantly less that the N.E.S.C. requirement for this condition
of 60% and 25% respectively.
Reference the overhead ground wire and the dead end assemblies on the span between
towers 1066 and 1067, see Figure 3. The Final Design Tension (2743 lbs.) and the final
unloaded tension (1914 lbs.) are at 17% and 9.6% respectively of the ultimate strength of
the dead-end hardware assembly. The loading of the OHGW hardware is significantly
less than the N.E.S.C. requirement for a 50% safety factor.
A great deal of engineering time was placed on developing design criterion for the MeadPhoenix 500kV line that exceeded industry standards. SRP believes it implemented a
line design that meets the above challenge including the span that crosses the Palo VerdeWestwing 500kV lines.
M-8
Figure Three
Overhead Ground Wire Assembly
D. Protective Relaying
The protection for the Palo Verde to Westwing Lines 1 and 2 is provided by
redundant microprocessor based technology, permissive overreaching relay schemes,
communicating via fiber optics and digital microwave systems on independent paths.
Independent dual channel transfer trip systems are provided over the redundant
communication channels. In addition, a third microprocessor based relay system,
operating in a current differential scheme is provided for back up protection. This
scheme utilizes relay-to-relay communication over one of the redundant
communication paths. The microprocessor technology of this equipment incorporates
self testing and monitoring capabilities for identifying critical problems, removing
tripping functions from service, and alarming to the dispatching offices via the EMS
system, where the trouble alarms receive a high priority. The use of total redundant
systems, allows for one scheme to be out of service to correct problems, without
compromising the protection of the line. Situations can be identified and corrected
before incorrect operations occur on an unfaulted line, decreasing the likelihood of a
simultaneous double line outage.
The Palo Verde to Westwing lines are not equipped with single-pole nor high speed
reclosing relays. In the North West, where very long paths are present, generation is
remote, and the isokeraunic level is high, single pole and high speed reclosing is
advantageous. In the Valley of the Sun, where SRP is located, load and generation
are closely located. Our generation is not “remote”. Therefore tripping a large
amount of generation is not a recommended procedure. The reason SRP does not
employ these relays and single pole reclosing is because of SSR problems discussed
in the “Transmission Line Reclosing – turbine generator duties and stability
M-9
considerations” article by P.G. Brown and R. Quay presented at the 29th conference
of Relay Engineers in 1976.
IV. Maintenance
The Palo Verde to Westwing lines are built 130 ft. apart (centerline to centerline).
This is sufficient to ensure that one tower failing for any reason would not fall into or
jeopardize the continued safe operation of the other.
Probably the biggest risk to these lines is vandalism, which almost always occurs
from shooting of insulators. SRP’s practice of performing an aerial patrol in the
spring and fall of each year and then performing a ground patrol twice a year should
be sufficient to find any vandalism areas and to make necessary repairs before they
result in an outage.
There are several areas where insulator contamination from birds has been a problem,
even resulting in an outage to one of the lines several years ago. The largest birds
capable of coming in contact with the lines are small and incapable of causing a phase
to phase fault. A picture of a typical large bird is shown in Appendix I, Figure 6.
SRP has been proactive in constructing protective devices and placing them above
insulators at several locations on these two lines. These protective devices contain the
bird droppings and keep them from falling on the insulators. The protective devices
are shown in Appendix I, Figures 7 and 8. The devices are flexible enough that winds
blow the bulk of the material away before the weight is high enough to be of any
concern structurally. Pacific Gas and Electric has used these bird guards successfully
for a number of years. SRP’s solution is taken from PG&E’s lead in this area. SRP
also owns two separate power washers used for washing insulators when
contamination is sufficient to warrant it. Insulator washing, with the tools and training
available to SRP, can be done with the lines energized.
Changing of insulators, repair of gunshot conductor, replacement of other hardware
are all maintenance activities that SRP performs with the lines energized. Also, our
company has the equipment and trained personnel to perform live line bare hand
maintenance. Many companies don't have this capability. There are several
deteriorating items on these lines that SRP monitors and replaces as needed. One of
these is the spacer dampers, which SRP changes as they deteriorate and fail. At this
point, the failure rate has been low enough that a major project to replace all of them
has not been necessary.
SRP has also been proactive in the identification and marking for potential aircraft
hazards. One location that was identified was where the two lines cross Interstate 10
west of the town of Buckeye. It is conceivable that a low flying airplane could snag
one of the shield wires and pull it across both lines. The lines at this location are
under 150 ft. above ground, which is well below the FAA minimum height for
airplanes. However, since this is a known potential route for small airplanes, SRP has
marked the shield wires at this location with aircraft warning balls. SRP will continue
M-10
to maintain these balls and replace them if and when they fail, as well as its
aggressive maintenance policy on this and other lines.
V. Substation Configuration
As can be seen from the two one line diagrams in this section, both the Westwing
substation and the Palo Verde substation are breaker and a half arrangements. At
either substation, if there is a breaker out for maintenance, it would require both (1) a
false trip on one line and (2) a stuck breaker on the other line to result in a loss of
both lines because of substation configuration.
M-11
M-12
Database Selection and Probabilistic Risk Assesment
Database Selection
This analysis seeks to infer future behavior of the Palo Verde Westwing 500 kV lines
(PV-WW) by using a sample of historical outage data from select lines from a larger
population of EHV transmission lines in the Southwest.
Sample Selection - Identification of a Statistical Database
The number and length of 500kV lines that SRP owns, or participates in, is small.
Therefore, an external statistical database of like lines in the Southwest was needed to
calculate a meaningful time between failure values.
SGS provided SRP with a listing of all 345-500 kV lines in the Southwest region, which
are similar to Palo Verde to Westwing 500kV lines 1 and 2. For those companies that
voluntarily report information to SGS for Transmission Reliability Benchmarking Study,
refer to 345 500 Inventory in Tables1 and 2 of the Circuit Outage and Data Tables. The
cause for removal of the lines from the database is discussed in the foreword to the table.
SRP worked in cooperation with SGS Statistical Services (SGS) to choose an appropriate
database and to perform probabilistic and statistical calculations on this database.
In order not to compromise confidential data that SGS has been entrusted with, SGS
provided SRP only with a listing containing circuit identification, circuit name, company
name, in service date or date of “earliest known good” outage data for the circuit, length,
and its voltage. The database did not contain any outage history.
The Transmission planning group of SRP met with the SRP Weather and Environmental
Services groups to discuss the terrain and weather conditions specific to the Palo Verde to
Westwing lines 1 and 2. After this meeting SRP pooled several engineers from the
transmission, design, and maintenance groups to review the listing of the 345-500 kV
lines provided by SGS and to see whether the lines in the database match the conditions
provided by Weather and Environmental services. This group jointly determined and
struck lines from the database, because they did not match the criteria. In order not to
jeopardize the confidentiality of the relationship between SGS and its clients, the original
database versus the final database with the “cuts” is not being distributed; as some of the
participants have only one or two lines, which would be easily identifiable.
Also, FERC Form 1 reports were examined for Public Service Company of New Mexico,
APS, and Tucson Electric to find lines on frame structures. These lines were removed
from the database, since Palo Verde to Westwing line is on steel towers. This first
database was a “matching terrain, weather conditions, and structure type” of similar lines
from the Sonoran desert or like environments.
The second step was to take this “matching terrain, weather conditions, and structure
type” database and scramble the company names and circuit names and look at the outage
history. Palo Verde to Westwing 1 has experienced only two single contingency (N-1)
forced outages in the twelve year history for which computer records have been kept.
Palo Verde to Westwing 2 has never experienced a single contingency (N-1) forced
outage. There has never been a double contingency (N-2) event involving those two
lines.
Because of SRP’s low outage rates on the 500kV system, SRP chose 1.5 single
contingency outages per year per line as a cutoff for similar lines or “like” lines database.
From the matching terrain, weather conditions, and structures database SGS provided
anonymous outage data, SRP choose all the lines that had 1.5 outages per year or fewer.
SRP also included the outage data for the Mead-Phoenix 500kV line to supplement the
SGS outage database. The Mead-Phoenix line is jointly owned by 21 participants and
exhibits 2.26 outages per year, which is much higher than the .09 outages per year for the
Palo Verde to Westwing 1 and the 1.5 outages per year for the “like lines” database.
However, the terrain for this line is very similar to that of the Palo Verde to Westwing
lines and therefore it was added to the database.
All of the forced outage history for the annual data available for the circuits, was used in
the statistical database. SRP conducted a survey of industry experts and leaders in the
academia in search of appropriate methodology. A multi-variate Weibull distribution
was suggested as a method to obtain more precise results. However, currently there are
no SAS programs that employ a multi-variate Weibull, this could be a topic for further
research.
Gregg A. Spindler
SGS Statistical Services, LLC
[email protected]
520-529-8202
Probabilistic Risk Assessment Using the Weibull Distribution
Statistical analysis of failures (or survival) is routinely performed in many fields
(e.g., medicine, pharmaceuticals, insurance, aerospace and electronics). It is used to
either infer future behavior or examine relationships between a response of interest and
possible predictor (independent) variables. In either case, a sample of data is selected
from a population of interest. The data includes both failures and non-failures. Most
typically this data is collected on a time-ordered basis and is represented as time between
failures (less common, but equally appropriate, are analysis of time-to-failure for
parameters such as number of cycles, temperature, force, etc).
This analysis uses a parametric method for failure analysis. Parametric methods
seek to characterize the failure process using a known statistical distribution (i.e., the
over-all “density” and “shape” of the failure process at any given point in time as
expressed by a mathematical function – see 1.1 below). By inferring the shape of the
failure data, it is possible to attach failure probabilities of failure at a given point in time
or to infer future behavior based on the distribution. To use a parametric method, one
must confirm the empirical evidence that a specified failure distribution approximates a
known failure distribution.
This analysis seeks to infer future behavior of the Palo Verde Westwing 500 kV
lines (PV-WW) by using a sample of historical outage data from select lines from a larger
population of EHV transmission lines in the Southwest. The sample was specifically
selected to provide comparable lines; the selection criteria used by SRP for the sample is
discussed in a previous section. SGS did not participate in or advise on which circuits
were included; SRP did not know specific outage histories for any lines other than those
it owns or operates. By using historical outage data, a statistical failure distribution is
“fit” to the data. Once the failure distribution is characterized, a Monte Cairo simulation
is performed to predict single, double and triple outage contingencies. The basis for this
prediction is the Weibull Distribution.
The southwestern US does not have the number or density of EHV lines present
in other regions. Hence, few SW EHV lines share common corridors. In the sample used
in this analysis, there are only two instances of an N-2 condition (a system protection
event on two lines with a shared terminal on separate ROWs and a second from August
10, 1996 on two physically separated lines of two different owners) present in 134
circuit-years of outage history. For these reasons it is not possible or desirable to
replicate the work of Bonneville Power Administration in its Kangley - Echo Lake
Probabilistic Risk Assessment. Consequently more traditional statistical reliability
analysis methods were used.
This analysis has relied on transmission outage data collected by transmission
owners. In nine years of producing the SGS Transmission Reliability Benchmarking
Study, it has always been apparent that the transmission industry would benefit from a
higher degree of consistency and commonality in the definitions and characterization of
non-available states for transmission lines. Beginning in May 2001 and completed in
February 2003, SGS Statistical Services facilitated and co-authored the Transmission
Line Availability Data Guidelines and Definitions, a collaborative and consensus-driven
process involving 37 transmission organizations. In September 2002, the WECC
Reliability Subcommittee was briefed on this effort. We encourage WECC and its
members to seriously consider adapting the Guideline to make such analyses more
precise and easier in the coming years. A copy of the Guideline is available at:
http://pages.prodigy.net/sgsstat/.
•
The Weibull Distribution
The Weibull Distribution is one of the most commonly used parametric statistical
distributions in reliability analysis. It is a continuous, location-scale type of failure
distribution and can assume a wide variety of shapes.i This flexibility allows its wide
application.
Prior to deciding upon use of the Weibull distribution, four statistical distributions
were evaluated to see if they fit the data. In addition to the Weibull, the lognormal,
exponential and extreme value distributions were evaluated. All four distributions are
illustrated in Figure 2-A: Comparison of Weibull, Exponential, Lognormal and Extreme
Value Distribution. None of the other distributions came close to the fit provided by the
Weibull, with “fit” being defined by how close the data (plotted points) adhere to the
distribution (the diagonal line in each panel).
The Weibull distribution, as any other failure distribution, may be characterized
by three different functions of time:
Probability Density Function (PDF): Expressed as f(t), the PDF is the relative
frequency of failures at a given point in time.ii The sample PDF is illustrated in Figure 4.
The PDF was used to generate random variates for the simulation. The Weibull PDF is
expressed as:
f ( t : µ, σ) =
β t 
 
η  η 
β −1
β

 t 
exp −   
  η 


Cumulative Density Function (CDF): Expressed as Pr(T ≤ t), the CDF is the
cumulative probability of failure of T ≤ t at a given point in time. The CDF for the
sample TBF is illustrated in Figure 2. The CDF is the standard tool which provides the
assessment of the fit of a particular distribution. The Weibull CDF is expressed as:
β

 t 
Pr(T ≤ t : η, β) = 1 − exp −    , t > 0
  η 


Hazard Function: Expressed as h(t), the Hazard function expresses the propensity to
fail at the next small interval of time given survival to time t.iii The Hazard function is
illustrated in Figure 4. A piece-wise hazard function for product life is the familiar
“bathtub curve” of failures (in this analysis the hazard function corresponds mostly with
the center portion of the “bathtub curve”). The Weibull Hazard function is expressed as:
h( t : µ, σ) =
β t 
 
η  η 
β −1
,t >0
For the three functions,
t
is a random var iable of time
µ is the exp ected value (mean ) of time
σ is the s tan dard deviation of time
β is the Weibull scale parameter
η is the Weibull shape parameter
The parameter t is the observed time for a failure; this forms the core of the
analysis. From the failure data we compute η, which is the “scale” parameter and also
corresponds to the 0.632 quantile value. The scale parameter may be though of as how
far “out” the distribution goes from t=0. The parameter β is the “shape” parameter. The
shape parameter indicates how “peaked” is the distribution. A β=1 is the commonly used
exponential distribution, which indicates a constant failure rate (hazard function). β < 1
indicates a decreasing failure rate while β > 1 is an increasing failure rate. β, η and t are
always > 0.
•
Preparation of Outage Data for Analysis
The outage data for selected lines represents a subset of the actual operating
history of each line. Unfortunately, the entire line outage history collected by each
transmission owner is not accessible in electronic format; some resides in paper logs.
Even when it is available, the older data is generally less complete or consistent than data
from the recent past. Thus, each line has either an “anchor date”, representing the date of
earliest “known, good” outage data or its in-service date for lines placed in service after
the anchor date.
An assumption in this analysis is that outage events are independent unit-events.
This assumption is made because transmission lines which experience outages are
returned to service in compliance with engineering and design standards (i.e., after a
failure it is returned to its condition defined by design standards with all known defects
removed). It is assumed that the pooled outage data from the selected sample of lines
represents the inherent, underlying failure distribution for a class of transmission lines.
The SGS Transmission Reliability Benchmarking Study collects transmission line
outage data from its participating systems each year. Because the Study results are
reported to participating systems in an anonymous format, it was not possible to provide
SRP with unblinded outage data for the full inventory of southwestern US transmission
lines. However, because each participating system is identified and its transmission
assets are public information, SRP was provided with a listing of, by name, of each EHV
transmission line represented in the SGS Study. This listing is contained in Table 1:
SGS Study Southwest US EHV Transmission Lines.
Lines were selectively eliminated from the analysis by SRP for engineering,
geographical or environmental reasons. Next, lines remaining were further reduced by
eliminating lines with average outages greater than 1.5 per year. In addition the MeadPhoenix line was included in the sample. The sample consisted of:
Number
Of
Circuits
Circuit Years
Circuit Mile-years
Sum Miles
Total Outages
17
134.06
10,409
1303
88
Next, outage history for each line was assembled (see Table 2: SGS Study
Southwest US Outage Data for SRP PV-WW Probabilistic Risk Assessment). Each
circuit’s outage data was “anchored” at the starting and ending points of outage data. The
starting point is the midnight of the first date of “known-good” outage data (for preexisting lines) or the in-service date (for new lines). The ending point is midnight of the
end of outage data (one line was retired when it was segmented and the retirement
represents the end of reporting).
Both starting and ending points of the data are considered censored observations.
Censoring is the acknowledgement that failure history prior to the anchor date is
unknown. Likewise, after the ending point of outage history, the failure behavior is also
unknown. However, the entire time interval a unit has been in service, whether failed or
not failed, is critical to characterizing the failure distribution.
Censoring and underlying outage data used in the analysis is illustrated in Table 2
and FIGURE 1: Graphical Depiction of Outage Data Used in Analysis. Starting and
ending times are illustrated with open circles with failures as Xs. Censoring may take
three forms:
•
•
•
Left Censoring: The time interval from the starting point for outage data until the
first failure.
Right Censoring: The time interval from the last failure until the ending point for
outage data.
Interval Censoring: A special case when no failures have occurred between the
starting and ending point of outage data.
Time between failures is calculated for each individual circuit from the starting
point to each successive failure up to the ending point of the data. TBF is an elapsed time
value. In addition to TBF, the duration for each outage is calculated as the interval
between the outage date/time and the restoration date/time. Duration is an elapsed time
value.
•
Outages vs Circuit Length
This assessment does not normalize performance due to exposure (length) of a
circuit because there was no evidence of a strong empirical relationship. Figures A1-A3
illustrate the statistical relationship between outages as a function of line length for all
SW lines, and a the sample of SW lines with Circuit Mean Outages Less than or Equal to
1.5 per Year, both including and excluding the Mead-Phoenix Line.
The key statistic on Figures A1-A3 is “R-Square”, the coefficient of
determination, which explains the proportion of variation in outages as a function of
circuit length. A “high” value of R-Square (much closer to 1.0 than to 0) indicates a
statistical relationship and correlation. An example of high correlation is summer high
temperature and peak system load; a considerably weaker relationship might be the
relationship of residential rate per KWH vs. peak load (i.e., an inelastic demand).
In this analysis there is no case where circuit length explains a majority of outage
behavior. The weakness of the relationship is confirmed by the “shotgun” pattern of
outages and length. The width of the prediction intervals about the regression lines also
indicate a weak relationship. The intervals essentially state that any value within the
bands is possible, given the variation present.
Indeed, if we examine figures A2 and A3, we see the influence on the regression
model with including and excluding the Mead-Phoenix Line. By the addition of a single
line to the model, the coefficient of determination jumps from 0.10 to 0.40; the change is
entirely due to a single influential line.
It has been SGS’s experience that while there is some relationship between
outages and circuit length, empirical evidence from a population representing 50% of US
transmission circuits indicate the relationship is always weak and often not present. This
challenges a paradigm of transmission engineering. Most often, subtransmission voltage
class lines will have higher R-Square values (but seldom above 0.40), while lines at 230
kV or the 345 – 500 kV voltage class have much weaker relationships. This holds true in
all geographic regions, both including and excluding terminal- and system protectionrelated events.
The evidence demonstrates that the majority of outage behavior is influenced by
factors other than length. Length, therefore, should not be considered a primary, driving
factor in reliability analysis no should it be used as a predictor of reliability.
•
Outage Causes
BPA has a long history of outage data and a disciplined root cause analysis
process. Additionally, it maintains a history of operational data. BPA outage and
operational data resources are amongst the best in the transmission industry. It is from
this basis that BPA is able to segregate types of outages with a high degree of confidence.
The BPA PRA for Kangley-Echo Lake separated line outage causes into categories
(independent line outages, terminal outages, human factors, breaker failures) and
assessed risk within each category. These were combined into a composite, “corrected”
estimate.
The outage data used in the SGS Study, one the other hand, comes from multiple
outage reporting systems which are generally not as mature as BPA’s. There is no reason
to believe that outage data is incomplete, but root cause description across several
transmission owners is known to be inconsistent. Unfortunately there are no US or
WECC standards for root cause description to insure the consistentcy enjoyed by BPA.
Within the SGS Study, outages are classified into ten broad categories. The
categories are: Equipment (terminals), System Protection, Lines, Weather (other than
lightning), Lightning, Unknown, Vegetation, External (off-system events), Other
(operating errors, vandalism, fire), Operational (manual operation for voltage, stability,
thermal limits, etc). The outage cause categories do not contain the detail of BPA’s
outage reporting and it is not possible to segregate within categories with a high degree of
confidence in the consistency of the root cause.
The sample pool of outage data used in this assessment is considerably larger (in
terms of circuit-years or mile-years of data) than the one used by BPA. The analytical
approach is based on a class of lines from multiple owners in similar operating
environments. All data and root causes are pooled to estimate the Weibull parameters.
Segregating different root cause categories into separate analyses would have reduced the
amount of data used to estimate the Weibull parameters and would have introduced
greater variability into the estimates (the standard error of the Weibull estimates),
because statistical precision of an estimate is driven by sample size (think of the “margin
of error” always stated in political polling). Segregating outage causes (and providing
“corrected” and “uncorrected” MTBF estimates similar to BPA’s) would have degraded
the analysis with more variability and less precision in the parameter estimates.
•
Fitting the Weibull Distribution
The key element of this analysis is insuring that the Weibull distribution fits the
data. If a reasonable fit is found, the estimated Weibull parameters for the distribution
will be used in the Monte Carlo simulation. SAS® software was used for the analysis.
The standard technique for fitting a reliability distribution is performed by
linearizing the CDF and plotting CDF vs. TBF. If a fit is “reasonable”, the observed
failures will appear approximately as a straight line on the TBF vs. CDF plot. Because
the true distribution of any failure process is unknown and unknowable, we estimate the
failure distribution and also may place specified levels of statistical confidence about the
TBF vs. CDF plot and associated distribution parameters. The confidence levels give a
level of assurance that the estimated distribution is appropriate.
Computation of the Weibull parameters (β and η) is performed using maximum
likelihood estimation (ML), a numerical analysis method which provides stable, robust
estimates. The maximum likelihood method is the default distribution fitting method
within SAS software. The parameters may also be computed using least-squares
regression, but parameters will differ slightly. The least-squares regression method
provides an R-Square value for the model fit; for this data it was 0.94, which is evidence
of an excellent model fit.
FIGURE 2 is the Weibull fit of TBF for the selected lines. An examination of the
plot indicates that the vast majority of observed failures are within the 99% confidence
bands of the estimated Weibull distribution (censored values are not on the plot, but are
used in parameter estimation). There are a few points outside of the bands and slight
curvature at the extremes of TBF values.
The curvature on the low end of the plot indicates that there are a handful of
instances of repeated failures occur within a 24 hour period after another outage. These
instances are perhaps cases of an instantaneous fault and automatic reclosure, followed
by a second fault and a lock-out. Curvature on the high side is the result of circuits
having very long intervals between failures, which may be unlikely to fail again (e.g., a
new line which has experienced early “infant mortality” and is now operating with a
near-zero failure rate).
FIGURE 4 is the PDF and Hazard Function of TBF for the outage data. The plot
is on a log-log scale for clarity. The hazard function is in black, the PDF appears in red.
This indicates that as time increases both the density and propensity of failures decrease.
Estimated Weibull Parameters for TBF are:
Weibull Parameter Estimates:
Parameter
Weibull Scale
Weibull Shape
Estimate
6503.1092
0.5259
Mean
Mode
Median
Standard Deviation
Time Between Failures (Hours)
Standard
Error
1359.1839
0.0518
99% Confidence Limits
Lower
Upper
3795.8888
0.4080
11141.1140
0.6779
11897.7072
0.0000
3239.4690
24815.4641
FIGURE 3 is the Weibull Fit of Outage Duration (seconds). The Weibull fit
appears reasonable, however there is some lack-of-fit for a few momentary outages. The
hypothesized distribution predicts slightly more momentary outages than were observed.
Estimated Weibull Parameters for Duration seconds are:
Weibull Parameter Estimates:
Parameter
Weibull Scale
Weibull Shape
Estimate
4985.5161
0.5260
Mean
Mode
Median
Standard Deviation
•
Outage Duration (Seconds)
Standard
99% Confidence Limits
Error
Lower
Upper
1010.8345
0.0399
2957.2999
0.4325
8404.7516
0.6396
9120.6963
0.0000
2483.5506
19022.4961
Monte Carlo Simulation
The outage data is considered to be representative of a class of SW EHV
transmission lines, characterized by the Weibull distribution fitted in the above section.
The two lines in question, PV-WW have reliability considerably better than any other
lines in the sample (or the entire SW region for that matter). For purposes of the
simulation, however, we are not characterizing particular transmission lines rather
generic lines characterized by the class of lines contained in the outage data.
Using the Weibull shape and scale parameters for TBF, 1,000,000 Weibull
random variates of TBF are generated for three generic transmission lines, X, Y and Z.
For each TBF value, a corresponding Weibull random variate is also generated to
characterize the duration of the outage. TBF values are accumulated for lines X, Y and
Z.
The simulated data series for lines X and Y are characterized by four figures. All
parameters from the simulated data are very close to the actual sample values contained
in the above section of Figures 2 and 3:
FIGURE 5: X Circuit TBF Random Weibull Distribution
FIGURE 6: Y Circuit TBF Random Weibull Distribution
FIGURE 7: X Circuit DURATION Random Weibull Distribution
FIGURE 8: Y Circuit DURATION Random Weibull Distribution
•
Coincident Events: Quantifying N-2 and N-3 MTBF Contingencies
For lines X and Y, each outage is evaluated for coincidence. This is identification
of the N-2 condition. Coincidence is defined if the current outage time on a given line (X
or Y) either coincides with or is within 30 minutes of a prior outage plus the prior
outage’s duration on the other line (Y or X). The coincident event is then flagged and a
TBF for coincidence is calculated.
Illustration of Coincident, Independent-Mode Line
Line
Line
Coincidence 2: Line Y
outage is within 30” of the
end of a Line X Outage
Coincidence 1: Line X
and Line Y Outages
FIGURE 9: X and Y Circuit TBF Independent and Coincident Events is a
“snapshot” which graphically illustrates how the data appears and highlights coincident
event. This plot contains approximately 26,000 of the 1,000,000 Line X and 1,000,000
Line Y random variates (one are blue dots, the other green dots, coincident events in
large red dots). There are five instances of coincident outages amongst the 26,000 plotted
outages; the actual data series of all 2,000,000 line X and Y values would occupy 80
pages of such a diagram.
A more precise characterization of N-2 is illustrated in FIGURE 10: X and Y
Circuit TBF Years of Coincident Events. The following table contains the associated
estimated Weibull parameters:
Weibull Parameter Estimates:
Parameter
Weibull Scale
Estimate
2585.3309
X-Y Circuit Double Contingency TBF Years
Standard
99% Confidence Limits
Error
Lower
Upper
144.0207
2239.7451
2984.2395
Weibull Shape
Mean
Mode
Median
Standard Deviation
0.8446
0.0310
2824.1149
0.0000
1675.1157
3359.4214
0.7685
0.9282
N-2 Occurrence MTBF: Thus the simulation estimates an N-2 occurrence, on
the average, once in 2824 years; this is the MTBF estimate. It must be recognized that
the average is the expected value and may be considered the best estimate of the N-2,
while the median occurrence of once in 1675 years. Because the simulation generated
486 occurrences of the N-2, there is a distribution about the mean value. Weibull
estimated percentile values for the N-2 are contained in the following table:
Percentile
0.1
0.2
0.5
1
2
5
10
20
30
40
50
60
70
80
90
95
99
99.9
Estimate (YRS)
0.72545382
1.64931322
4.88944322
11.1426631
25.4699836
76.7639898
180.016976
437.728146
762.750464
1167.06507
1675.11567
2331.10311
3220.8513
4541.81759
6940.63586
9478.10828
15770.1269
25488.0635
N-3 Occurrence MTBF: For the N-3 condition, the occurrence of N-2 events is
then combined with the simulated event history of line Z and evaluated in the same
manner as the N-2. There were no occurrences of N-3 observed in 1,000,000 simulated
events. Based on the simulation, the N-3 MTBF cannot be calculated, but may be
assumed to be much greater than the N-2 MTBF.
•
Analytical Caveats and Cautions
As with the BPA PRA, this method relies on historical outage information to
predict a specific type of future event. In both analyses, the estimates of N-2 are only as
good as the data which underlie them. There is no reason to suspect under-reporting of
outage by any transmission owner, but data elements such as the precise duration or
cause description are occasionally problematic.
Both the BPA and this Weibull analyses do not incorporate the occurrence of
major catastrophic events such as earthquakes, tornadoes or system collapse. In the case
of major natural disasters, none are represented in the outage data of either analysis. In
the case of system-related events. The August 10, 1996 event is contained in the SW data
and two of the sample lines were affected on that date. No other catastrophes are
contained in the data. Further such catastrophic events are exceedingly difficult to
predict.
Neither this Weibull nor the BPA analyses have factored in the aging of lines and
their associated materials. There is a finite life for individual line components and as
these components age one might expect an elevated outage rate, assuming there are not
reliability-centered maintenance procedures in place. Anecdotally it has been suggested
that there is a 100-year life for the PV-WW lines and that “around” year 80 there might
be an increasing failure rate.
Because the BPA and this Weibull analyses are based on a limited amount of data
and one may reasonable expect lines to age, it is both reasonable and prudent to re-visit
the analyses in the future and validate the models used by both.
The BPA analysis had five N-2 events on 3 pairs of lines with shared corridors.
Four of the five events were initiated by lightning, the fifth event cause is coded as
“weather”. Four of the five N-2 events appear to be common-mode independent outages,
while the fifth appears to be a common-mode dependent event. As previously stated,
there are no comparable lines with shared corridors in the SW. Consequently the BPAtype analysis was impossible. This analysis assumed independent unit-events for the
outage history. The N-2 contingency estimates in this analysis are independent, common
mode or independent coincident outages, as apparently are 4 of the five BPA N-2 events.
This analysis, by using the Weibull distribution, has assumed outage are
independent unit-events. Ideally, we would have preferred using the Mean Cumulative
Functioniv (MCF), which has both a parametric and non-parametric form. Unfortunately,
while the MCF is the best way of characterizing the reliability of a repairable system,
there is no way to generate random values from the MCF within SAS software, nor is this
approach described in statistical reference texts. It would appear that using the MCF in
this manner is a subject of academic research and software development.
The Weibull analysis attaches levels of statistical confidence to its estimates.
This is a recognition that the limited amount of outage history and resulting simulations
have variability. As such, the mean estimate of an N-2 once in 2824 years is the best
estimate, while the minimum estimate was 0.000001763925 years or less than one minute
apart!. Events with seemingly infinitesimal probabilities do happen in real life (e.g.,
people actually do win the lottery, even though an individual’s chance is near-zero). It
should be noted that the maximum value from the simulation was 18,937 years and the
maximum Weibull estimate was 25,488 years.
Analysis of the Palo Verde to Westwing Lines 1 and 2 double contingency (N-2)
qualifies to be moved to Category D based on the following statistical analysis and
mitigating factors:
1)
2)
3)
4)
5)
6)
7)
8)
9)
10)
11)
12)
13)
14)
15)
16)
An MTBF estimated by a traditional statistical reliability analysis method is
on average once in 2824 years.
In the 11 years of accurately recorded outage history in electronic format,
there has never been a double contingency outage of the Palo Verde to
Westwing lines. Evidence suggests that since both lines were in service, this
outage has not ever occurred.
Both Westwing and Palo Verde switchyard use breaker and a half
arrangement.
As a result of the Rudd line installation, the Palo Verde to Westwing lines 1
and 2 outage is no longer the most critical outage. For a single-line-to-ground
fault at the Palo Verde 500kV bus and subsequent loss of the two Palo Verde
to Westwing circuits, no significant power flow nor stability problems were
found with the net generation of 9,595 MW.
According to UFSAR, the failure of this line at the crossing over the PV-WW
1 and 2 is no longer postulated under the revised 10CFR50.59 rules.
Therefore grid studies need not address this scenario.
The Robust design features are overhead ground wires, lines are built 130 feet
apart (centerline to centerline) with towers designed to fail in the middle. The
failure and fall of one tower does not jeopardize the continued safe operation
of the other tower.
Palo Verde to Westwing 500kV lines are located outside the areas of
consideration for air traffic. The elevation of the lines in beyond and beneath
the criteria FAA defines for consideration as an obstacle or hazard.
The isokeraunic level near Palo Verde and Westwing is one of the lowest in
the Western US, ranging from 1.0 strikes per square mile per year near Palo
Verde to 2.5 strikes per square mile per year near Westwing switchyard.
The risk of earthquakes in Maricopa County is the lowest in the Western US.
The risks of flood, snow, and fire are negligible.
The PV-WW foundations are over designed in the range of 137 to 199%.
The lattice tower design is conservative for weather related loads.
Lines are designed with state of the art spacer dampers to control conductor
motion.
The insulation level exceeds EPRI’s guidelines.
Electronic protection is provided by redundant microprocessor based
technology with communication via fiber optics and digital microwave
systems on independent paths. A third microprocessor based relay system
operating in current differential scheme is provided for backup protection.
SRP aggressively maintains the lines with twice yearly patrols, bird guard
systems in place, an insulator-washing program, and a spacer damper
replacement program.
In summary, based on an MTBF estimated by traditional statistical reliability analysis of
2824 years and excellent design and maintenance practices, it is recommended that this
N-2 outage be moved to Category D (Extreme Events) with no other conditions or
requirements.
References
1
Statistical Methods for Reliability Data, William Meeker and Luis Escobar, John Wiley &
Sons, New York, 1998, page 87
2
Ibid., p. 28
3
Ibid.
4
Ibid, pages 393-419
Other References:
SAS/QC User’s Guide, Version 8, SAS Institute, Cary, NC, 1999, 1994pp.
Reliability in Engineering Design, K.C. Kapur and L.R. Lamberson, John Wiley & Sons,
New York, 1977
i
Statistical Methods for Reliability Data, William Meeker and Luis Escobar, John Wiley &
Sons, New York, 1998, page 87
ii
Ibid., p. 28
iii
Ibid.
iv
Ibid, pages 393-419
Other References:
SAS/QC User’s Guide, Version 8, SAS Institute, Cary, NC, 1999, 1994pp.
Reliability in Engineering Design, K.C. Kapur and L.R. Lamberson, John Wiley & Sons, New
York, 1977
Appendix I
Figure 1 and Figure 2 Caption: Flash density in Arizona for all years and months
combined from 1984-1993 (1985 and 1986 omitted). Density in flashes per square mile
per year shown by scale. Values in gray exceed 8. 500 kV lines in red; 345 kV lines in
black. Salt River Project service areas outlined in black. (a) Complete “Figure 2” from
NSSL 1995; (b) zoom in on area of concern with key sites labeled.
A-1
Figure 3. Tallest tree underneath the Palo Verde to Westwing 500kV Lines.
Figure 4. Tallest cactus underneath the Palo Verde to Westwing 500kV Lines.
Figure 5. Typical tower footing.
A-2
Figure 6. A hawk.
Figures 7 and 8. Protective bird guards on the Palo Verde to Westwing 500kV lines.
A-3
Figure 9.
A typical spacer damper.
Figure 10. RPEWG Recommended Analysis Steps.
Seven step Process for PBRC adjustment:
1. Provide Complete Project Description, and why it is being considered for PBRC
adjusted rating, including supportive data:
a. Overview of terminations
b. Physical Layout and Transmission Construction
c. Substation Configurations
d. Protective Relaying
e. Isochronic Level
f. Aircraft Hazard
g. Fire Hazard
2. Identify the Statistical Base to be used:
a. Historical
b. Similar Lines
c. Mileage
d. Terrain
e. Climate
3. Determine Uncorrected of Mean Time Between Failure (MTBF)
•
All events should be counted and considered, and then select events and
circumstances can be removed on a case-by-case basis.
4. Provide a Corrected Estimate of MTBF (based on Project Robustness Features)
A-4
•
A partial list of events that may be justified out is included in section 3.6 of
the PBRC process.
•
Consider various robustness features introduced to reduce the risk of outage.
For examples see reference [1].
5. Complete Exposure Analysis. (Refer to example)
6. Illustrate the Consequences of Outage. (Refer to example)
7. Conclude the how the adjustment meets the PBRC criteria. (Refer to example)
Reference
[1] Robust Line Design Features, RPEWG working paper, 5/28/02.
A-5
Appendix II
Palo Verde/Hassayampa Station Site and 500kV Transmission Lines.
A-6
Table 1: SGS Study Southwest US EHV Transmission Lines
Some EHV Transmission lines were removed from the “first cut” database due to
incompatibility with Southwest desert conditions database or due to frame (wood)
structures. These are: Cholla-Preacher Canyon – Pinnacle Peak (an APS line) – traverses
rim area, Cholla-Saguaro (APS), San Juan –Four Corners (PNM) –in common corridor
The following lines were on frame structures and thus removed: San Juan-Four Corners,
Four Corners –West Me, San Juan –Ojo, San Juan-Shiprock, West Mesa – BA, West Mesa –
Sandia, San Juan – BA. Additionally, Mckinley – Springerville #1 and #2 – ice,SpringervilleGreenlee, Springerville –Vail were removed due to unfavorable weather conditions.
Circuit Name
CHOLLA - FOUR CORNERS #1
CHOLLA - FOUR CORNERS #2
CHOLLA - PINNACLE PEAK
ELDORADO (SCE) - MOENKOPI
FOUR CORNERS - MOENKOPI
MOENKOPI - NAVAJO
MOENKOPI - YAVAPAI
NAVAJO - WESTWING
NORTH GILA - PALO VERDE (SRP) SWYD.
WESTWING - YAVAPAI
BA - BLACKWATER
NORTON - BA
IMPERIAL VALLEY-MIGUEL
IMPERIAL VALLEY-NORTH GILA
NORTH GILA-PALO VERDE
BROWNING SILVERKG
BROWNING KYRENE
CHOLLA CORONADO
CORONADO SILVERKG
KYRENE PALOVERDE
KYRENE SILVERKG
MEAD - PHOENIX
PALOVERDE WWG 1
PALOVERDE WWG 2
AEPCO_GREENLEE
AEPCO_VAIL_BICKNELL_345KV
CORONADO_SPRINGERVILLE_345KV_ LINE_ 311
EPE_GREENLEE_HIDALGO
EPE_SPRINGERVILLE_LUNA_345_KV_LINE
GREENLEE_VAIL _345KV LINE_303
SAGUARO_TORTOLITA_500KV_LINE_501
SAN JUAN_MCKINLEY_#1_345KV_LINE_301
SAN JUAN_MCKINLEY_#2_345KV_LINE_306
VAIL_SOUTH_345KV_LINE_304
WESTWING_SOUTH_345KV_LINE_305
TOLK STA TO EDDY CO INTG
TUCO INTG TO OKLAUNION (PSO)
Owner Circuit
-Miles
APS
APS
APS
APS
APS
APS
APS
APS
APS
APS
PNM
PNM
SDGE
SDGE
SDGE
SRP
SRP
SRP
SRP
SRP
SRP
SRP
SRP
SRP
TEP
TEP
TEP
TEP
TEP
TEP
TEP
TEP
TEP
TEP
TEP
XCEL
XCEL
159
159
130
217
179
76
101
256
114
79
223
42
84
81
114
39
19
73
180
75
58
242
45
45
.
.
22
.
.
128
1
90
90
14
178
158
162
Circuit
Years
8
8
8
8
8
8
4.6
8
8
4.6
12
12
12
12
12
1
1
11
11
11
10
5.7
11
11
7.
7.
7
7.
7.
7
7
7
7
7
7
5
5
Circuit
MileYears
1273
1273
1041
1737
1436
608
466
2049
913
364
2674
506
1003
968
1374
39
19
808
1980
826
580
1393
496
496
154
897
7
630
630
98
1247
790
810
Table 2: SGS Study Southwest US Outage Data for SRP PV-WW Probabilistic Risk
Assessment
ANONYMOUS
CIRCUIT ID
500-0100
500-0400
500-0400
500-0400
500-0400
500-0400
500-0400
500-0400
500-0400
500-0400
500-0400
500-0400
500-0410
500-0410
500-0420
500-0420
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
MEAD-PHX
PV-WW1
PV-WW1
PV-WW2
XXX 0001
XXX 0001
XXX 0005
XXX 0005
XXX 0005
XXX 0005
XXX 0005
XXX 0005
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0008
XXX 0011
XXX 0011
XXX 0011
OUTAGE DATE / RESTORE DATE /
TBF
SGS STUDY CAUSE CODE
TIME
TIME
HOURS
01JAN02:00:00
.
96432
16MAR92:19:21
17MAR92:04:52
10579 EXTERNAL
17MAR92:04:53
17MAR92:04:53
10 EXTERNAL
29DEC92:14:20
29DEC92:14:25
6897 WEATHER
29DEC92:14:26
29DEC92:14:26
0 WEATHER
25JUL93:10:31
25JUL93:10:39
4988 OTHER
25JAN94:10:54
25JAN94:14:27
4416 WEATHER
25JAN94:14:28
25JAN94:14:39
4 WEATHER
12OCT97:06:38
12OCT97:07:45
32536 EXTERNAL
11NOV97:11:12
11NOV97:11:47
725 OTHER
31JUL00:00:24
31JUL00:00:26
23821 SYSTEM PROTECTION
01JAN01:00:00
.
3696
01MAY01:11:10
01MAY01:12:23
2891 SYSTEM PROTECTION
01JAN02:00:00
.
5869
01MAY01:11:10
01MAY01:12:21
2891 SYSTEM PROTECTION
01JAN02:00:00
.
5869
03JUL96:00:00
03JUL96:00:06
2136 LIGHTNING
28OCT96:00:00
28OCT96:01:51
2808 SYSTEM PROTECTION
07JAN97:00:00
07JAN97:00:04
1704 OTHER
27JAN97:00:00
27JAN97:00:09
480 SYSTEM PROTECTION
05APR97:00:00
05APR97:00:09
1632 OTHER
15JUL97:00:00
15JUL97:00:39
2424 OTHER
29AUG97:00:00
29AUG97:00:07
1080 LIGHTNING
07JAN98:00:00
07JAN98:00:08
3144 OTHER
11SEP98:00:00
11SEP98:03:07
5928 LIGHTNING
30NOV98:00:00
30NOV98:00:06
1920 UNKNOWN
19JAN99:00:00
19JAN99:01:08
1200 OTHER
20JAN99:00:00
20JAN99:00:52
24 OTHER
21JUL99:00:00
21JUL99:00:06
4368 EQUIPMENT
01JAN02:00:00
.
21480
12JAN01:04:03
12JAN01:11:50
87940 EXTERNAL
01JAN02:00:00
.
8492
01JAN02:00:00
.
96432
21OCT98:07:33
21OCT98:16:05
21152 LIGHTNING
01JAN01:00:00
.
19264
31MAY96:12:05
31MAY96:12:09
204 EQUIPMENT
21OCT98:08:46
21OCT98:11:09
20949 EQUIPMENT
02NOV98:13:03
02NOV98:16:33
292 EQUIPMENT
05NOV98:16:03
05NOV98:17:03
75 EQUIPMENT
20APR99:07:33
20APR99:07:55
3976 OTHER
01JAN01:00:00
.
14920
26MAY93:21:51
26MAY93:21:53
3502 WEATHER
03JUN93:07:28
03JUN93:07:42
178 EQUIPMENT
15FEB94:06:58
15FEB94:18:40
6168 EQUIPMENT
24JUN94:17:35
24JUN94:20:41
3107 VEGETATION
12SEP94:12:06
13SEP94:11:07
1915 LINES
29FEB96:08:36
29FEB96:08:46
12837 EQUIPMENT
18MAR96:07:43
18MAR96:07:59
431 UNKNOWN
20AUG96:16:12
20AUG96:16:13
3728 WEATHER
04NOV96:14:45
04NOV96:14:47
1823 UNKNOWN
15MAR99:08:36
15MAR99:08:47
20658 LINES
23NOV99:09:06
24NOV99:16:05
6073 OTHER
24AUG00:17:32
24AUG00:17:34
6608 WEATHER
01JAN01:00:00
.
3102
11JAN90:00:00
11JAN90:00:01
240 EXTERNAL
28DEC92:00:00
28DEC92:00:01
25968 EXTERNAL
27MAR96:00:00
27MAR96:01:01
28440 OTHER
CENSOR
1
0
0
0
0
0
0
0
0
0
0
1
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
1
0
1
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
XXX 0011
XXX 0011
XXX 0011
XXX 0011
XXX 0011
XXX 0015
XXX 0015
XXX 0027
XXX 0027
XXX 0027
XXX 0040
XXX 0040
XXX 0040
XXX 0040
XXX 0040
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0041
XXX 0044
XXX 0044
XXX 0044
XXX 0044
XXX 0044
XXX 0044
XXX 0044
XXX 0044
XXX 0049
XXX 0049
XXX 0049
XXX 0049
XXX 0049
XXX 0049
XXX 0049
XXX 0049
XXX 0049
30JAN98:00:00
02APR99:00:00
12MAY99:00:00
29JUL00:22:07
01JAN02:00:00
06SEP95:16:51
01JAN02:00:00
05FEB99:05:17
05FEB99:11:26
01JAN02:00:00
06NOV97:14:29
26DEC99:04:23
04MAY00:07:56
13JUL01:00:58
01JAN02:00:00
14AUG90:07:04
15SEP90:04:08
01SEP91:08:00
21OCT91:14:27
02DEC95:04:29
02DEC95:04:44
10AUG96:15:48
04NOV98:14:05
01MAR99:22:14
03OCT00:22:03
27NOV00:09:31
03JUN01:04:21
03JUN01:11:40
04NOV01:03:47
11NOV01:13:37
01JAN02:00:00
05AUG93:19:26
04MAY95:13:48
05MAY95:02:01
15APR96:04:32
10AUG96:15:48
07APR98:14:22
16NOV99:19:44
01JAN01:00:00
16MAY95:05:00
05NOV97:21:49
10NOV98:20:00
05FEB99:06:17
05FEB99:12:26
02NOV99:06:12
25JUL00:09:35
31JUL00:00:24
01JAN01:00:00
30JAN98:01:00
02APR99:00:46
12MAY99:00:06
29JUL00:22:10
.
06SEP95:16:51
.
05FEB99:05:22
05FEB99:20:14
.
06NOV97:17:14
26DEC99:04:24
04MAY00:09:49
13JUL01:05:36
.
14AUG90:07:11
15SEP90:21:34
01SEP91:08:06
21OCT91:14:32
02DEC95:04:37
02DEC95:10:03
10AUG96:18:27
04NOV98:14:30
01MAR99:22:21
03OCT00:22:11
27NOV00:10:28
03JUN01:04:33
03JUN01:14:58
04NOV01:04:04
11NOV01:13:48
.
05AUG93:19:36
04MAY95:14:09
05MAY95:02:13
15APR96:07:09
10AUG96:17:03
07APR98:14:24
17NOV99:03:53
.
16MAY95:05:09
05NOV97:21:54
10NOV98:20:20
05FEB99:06:22
05FEB99:21:14
02NOV99:08:40
25JUL00:09:38
31JUL00:14:01
.
16176
10248
960
10678
12482
5969
55399
79733
6
25453
24974
18710
3124
10433
4127
5407
765
8428
1206
36062
0
6059
19582
2816
13968
1307
4507
7
3688
178
1210
5203
15282
12
8307
2819
14519
14117
9868
20765
21713
8878
2074
6
6474
6387
135
3696
EQUIPMENT
OTHER
SYSTEM PROTECTION
WEATHER
WEATHER
EQUIPMENT
EQUIPMENT
SYSTEM PROTECTION
SYSTEM PROTECTION
EQUIPMENT
OTHER
LIGHTNING
EXTERNAL
WEATHER
LINES
UNKNOWN
UNKNOWN
EXTERNAL
OTHER
SYSTEM PROTECTION
LINES
WEATHER
WEATHER
WEATHER
UNKNOWN
UNKNOWN
WEATHER
EXTERNAL
EXTERNAL
EQUIPMENT
EXTERNAL
LIGHTNING
EQUIPMENT
EQUIPMENT
UNKNOWN
UNKNOWN
EQUIPMENT
EQUIPMENT
LINES
UNKNOWN
EXTERNAL
0
0
0
0
1
0
1
0
0
1
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
NOTE: Anchor date is not given. SRP Lines are explicitly identifed with their internal IDs. Other
lines are anonymously identified.
Multiple Contingency Analysis
Using Weibull Distribution and Monte Carlo Simulation
Start
Figure 1
Table 1, 2
Figure 8 (sample),
Figure 9 TBF (all)
Engineering Review
of Lines by SRP
Select “Similar” Lines
and <1.5 Outages
per Year
Fit Weibull Distribution for
TBF Hours and Duration
using Outage Data
Figures 2, 3
Using Weibull Parameters,
Generate 1,000,000
Random Events (TBF and
Duration) for Circuits X, Y
and Z
Figures 4-7
Evaluate Circuit X and Y
for Coincidence.
Calculate TBF for N-2
For Coincident X and Y
Events, Evaluate for
Coincidence with Circuit Z.
Calculate TBF for N-3
Compute Summary
Statistics for X, Y and Z
Circuits, N-1, N-2 and N-3
Report Body
END
SGS Statistical Services: 24FEB03
Figure A.1: Regression of Average Outages vs. Length for All SW 345-500 kV Circuit in SGS Study
outage_mean_yr
12
11
10
9
8
7
6
5
4
3
2
1
0
0.00
100.00
200.00
300.00
Line Miles_Max
Regression Equation:
outage_mean_yr = 0.116768 + 0.016275*length_Max
Model R-Square: 0.30
SGS Statistical Services: 23FEB03
Figure A.2: Regression of Average Outages vs. Length for OUTAGES LE 1.5 and Mead-Phoenix Line
outage_mean_yr
3
2
1
0
0.00
100.00
200.00
300.00
Line Miles_Max
Regression Equation:
outage_mean_yr = 0.261143 + 0.006508*length_Max
Model R-Square: 0.40
SGS Statistical Services: 23FEB03
Figure A.3: Regression of Average Outages vs. Length for OUTAGES LE 1.5
outage_mean_yr
1.5
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.00
20.00
40.00
60.00
80.00
100.00
120.00
140.00
160.00
Line Miles_Max
Regression Equation:
outage_mean_yr = 0.421961 + 0.003682*length_Max
Model R-Square: 0.10
SGS Statistical Services: 23FEB03
FIGURE 1: Graphical Depiction of Outage Data Used in Analysis
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
XXX 0049
XXX 0044
XXX 0041
XXX 0040
XXX 0027
XXX 0015
XXX 0011
XXX 0008
XXX 0005
XXX 0001
PV-WW2
PV-WW1
MEAD-PHX
500-0420
500-0410
500-0400
500-0100
1990
1991
1992
1993
CENSOR
1994
1995
LEFT
1996
1997
FAILED
1998
1999
2000
2001
2002
RIGHT
SGS Statistical Services: 24FEB03
FIGURE 2: Weibull Fit of Time Between Failures
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
99.9
Weibull Fit for TBF
99
95
90
80
70
60
50
40
Scale
Shape
Uncensored
Right Censored
Left Censored
Conf. Coeff.
Fit
6503
0.526
73
15
15
99%
ML
Percent
30
20
10
5
2
1
.5
.01
.1
1
10
100
1000
10000
100000
Weibull Plot For (t1, t2)
SGS Statistical Services: 24FEB03
FIGURE 2-A: Comparison of Weibull, Exponential, Lognormal and Extreme Value Distribution
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
WEIBULL DISTRIBUTION
99.9
99
EXPONENTIAL DISTRIBUTION
99.9
99
6921
0.423
73
15
15
0.942
LSXY
Scale
Uncensored
Right Censored
Left Censored
Fit
3138
73
15
15
LSXY
95
80
80
60
60
40
30
40
30
Percent
Percent
95
Scale
Shape
Uncensored
Right Censored
Left Censored
R Squared
Fit
20
20
10
10
5
5
2
2
1
1
.5
.5
.1
.01
1
10
100
1000
10000
100000
.1
.01
1
Weibull Plot For (t1, t2)
LOGNORMAL DISTRIBUTION
99.9
Location
Scale
Uncensored
Right Censored
Left Censored
R Squared
Fit
10
100
1000
10000
100000
Exponential Plot For (t1, t2)
EXTREME VALUE DISTRIBUTION
99.9
99
95
7.601
2.814
73
15
15
0.827
LSXY
99
Location
Scale
Uncensored
Right Censored
Left Censored
Interval Censored
R Squared
Fit
10233
5599
73
15
15
2
0.615
LSXY
80
95
60
80
70
60
50
40
30
20
Percent
Percent
90
40
30
20
10
5
10
2
5
1
2
1
.5
.5
.01
.1
1
10
100
Lognormal Plot For (t1, t2)
1000
10000
100000
0
10000
20000
30000
40000
Extreme Value Plot For (t1, t2)
SGS Statistical Services: 27FEB03
FIGURE 3: Weibull Fit of Outage Duration (seconds)
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
99.99
99.9
99
95
80
70
Weibull Fit for Duration
Scale
Shape
Uncensored
Conf. Coeff.
Fit
4986
0.526
99
99%
ML
Percent
50
40
30
20
10
5
2
1
.5
.2
10
100
1000
10000
100000
1000000
outage duration
SGS Statistical Services: 24FEB03
FIGURE 4: PDF and Hazard Function of Time Between Failures
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
ht
0.1000000
0.0100000
0.0010000
0.0001000
0.0000100
0.0000010
0.0000001
0
1
10
100
1000
10000
100000
tbf_hours
SGS Statistical Services: 24FEB03
Red is PDF [f(t)] Black is Hazard [H(t)]
FIGURE 5: X Circuit TBF Random Weibull Distribution
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
99.9999
Weibull Fit for TBF
99
90
60
40
20
Scale
Shape
Uncensored
Conf. Coeff.
Fit
6643
0.527
10000
99%
ML
Percent
10
5
2
1
.5
.2
.1
.01
.001
.0001
.0001
.001
.01
.1
1
10
100
1000
10000
100000
1000000
Weibull Plot For x_hr
SGS Statistical Services: 24FEB03
10,000 of 1,000,000 Random Values Used for Computing/Ploting Clarity
FIGURE 6: Y Circuit TBF Random Weibull Distribution
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
99.9999
Weibull Fit for TBF
99
90
60
40
20
Scale
Shape
Uncensored
Conf. Coeff.
Fit
6501
0.532
10000
99%
ML
Percent
10
5
2
1
.5
.2
.1
.01
.001
.0001
.0001
.001
.01
.1
1
10
100
1000
10000
100000
1000000
Weibull Plot For y_hr
SGS Statistical Services: 24FEB03
10,000 of 1,000,000 Random Values Used for Computing/Ploting Clarity
FIGURE 7: X Circuit DURATION Random Weibull Distribution
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
99.9999
Weibull Fit for Duration
99
90
60
40
20
Scale
Shape
Uncensored
Conf. Coeff.
Fit
5066
0.527
10000
99%
ML
Percent
10
5
2
1
.5
.2
.1
.01
.001
.0001
1E-06
.00001
.0001
.001
.01
.1
1
10
100
1000
10000
100000 1000000
Weibull Plot For x_dur
SGS Statistical Services: 24FEB03
10,000 of 1,000,000 Random Values Used for Computing/Ploting Clarity
FIGURE 8: Y Circuit DURATION Random Weibull Distribution
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
99.9999
Weibull Fit for Duration
99
90
60
40
20
Scale
Shape
Uncensored
Conf. Coeff.
Fit
5012
0.531
10000
99%
ML
Percent
10
5
2
1
.5
.2
.1
.01
.001
.0001
.0001
.001
.01
.1
1
10
100
1000
10000
100000
1000000
Weibull Plot For y_dur
SGS Statistical Services: 24FEB03
10,000 of 1,000,000 Random Values Used for Computing/Ploting Clarity
FIGURE 9: X and Y Circuit TBF Independent and Coincident Events
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
60
55
50
45
40
35
30
25
20
15
10
5
0
0
50
100
150
200
250
300
YEARS
SGS Statistical Services: 24FEB03
60 Random 300 Year Series for Circuits X and Y
FIGURE 10: X and Y Circuit TBF Years of Coincident Events
Circuit Mean Outages Less than or Equal to 1.5 per Year and Mead-Phoenix Line
Salt River Project Palo Verde Westwing Probabilistic Risk Assessment
99.99
Weibull Fit for TBF N-2
99
95
70
50
Scale
Shape
Uncensored
Conf. Coeff.
Fit
2585
0.845
487
99%
ML
30
20
Percent
10
5
2
1
.5
.2
.1
.01
1E-06
.00001
.0001
.001
.01
.1
1
10
100
1000
10000
100000
X-Y Circuit Double Contingency TBF Years
SGS Statistical Services: 24FEB03
Attachment 3
6/3/2003
Attachment 4
RPEWG Evaluation
Palo Verde to Westwing Double 500 kV Line Outage
Seven Step Process for PBRC Adjustment
BACKGROUND
Salt River Project (“SRP”) requested a Probabilistic Based Reliability Criteria (“PBRC”)
Adjustment of the Palo Verde to Westwing (“PV-WW”) 500 kV double line outage from
the NERC/WECC Category C to Category D. The report submitted by SRP includes
significant detail describing the PV-WW lines and was augmented with consultant
reports for specific elements of the report. SRP followed the Seven Step Process For
PBRC Adjustment1 and provided additional detail beyond the defined steps. RPEWG’s
evaluation follows.
Evaluation of the Seven Step Process:
1. Provide Complete Project Description, and why it is being considered for PBRC
adjusted rating, including supportive data:
a. Overview of terminations
b. Physical Layout and Transmission Construction
c. Substation Configurations
d. Protective Relaying
e. Isochronic Level
f. Aircraft Hazard
g. Fire Hazard
This step was met. A complete project description, including the suggested a-g
supportive data, was addressed in the report. Many photographs accompanied the
discussion. The report states that the double PV-WW “... outage is considered of
such low probability of occurrence and recurrence, that it warrants submittal to the
WECC Phase I Probabilistic Based Reliability Criteria (PBRC) Performance
Category Evaluation (PCE) Process...”
2. Identify the Statistical Base to be used:
a. Historical
b. Similar Lines
c. Mileage
d. Terrain
e. Climate
The outage history of PV-WW circuits has been excellent. There has not been any
double PV-WW circuit outage in the history of the lines. One of the circuits has
had two outages and the other line has not had an outage. Given the non-existent
PV-WW double line outage data, SRP chose to meet this step by developing a
1
PCC Handbook, Revised October 2002, page IX-2.
-1-
6/3/2003
sample database from a larger population of EHV transmission lines within the
Southwest. An external statistical database of like lines in the Southwest was used
because the EHV lines that SRP owns or participated in is small. The sample was
made up of single EHV lines and contained only a few lines with two circuits on
the same corridor. SRP used SGS Statistical Services, LLC, to identify the
historical sample database from SGS’s Transmission Reliability Benchmarking
Study database. Participation in the SGS database is voluntarily and not all
transmission utilities participate. A comparison of the SGS circuit names to the
listing of “Multiple Circuit Corridors/Lines > 300 kV”2 identified one Southwest
utility included in the Multiple Circuit listing that was not also included in the SGS
data. The SGS derived sample database was useful for developing MTBF for
coincident independent outage events only. There appears to be no superior data
set of lines similar to the PV-WW line at the present time from which a valid
sample of similar lines with common mode outages for two circuits in a common
corridor can be developed.
3. Determine Uncorrected of Mean Time Between Failure (MTBF)
•
All events should be counted and considered, and then select events and
circumstances can be removed on a case-by-case basis.
This step was completed using the SGS database. SRP and SGS winnowed SGS’s
database by identifying lines with “matching terrain, weather conditions, and
structure type.” This created the uncorrected database of 39 circuits. The resulting
sample was used in a Weibull method of analysis. This analysis identified a MTBF
mean and median of 500 year and 299 years, respectively. This analysis yielded
MTBF results for coincident independent outage events and not an N-2 MTBF for
common mode dependent outage events. A MTBF analysis for a single event
causing a double line outage (i.e., dependent event) was not completed because a
database of common mode double line outages could not be established from the
SGS Statistical Services data. The coincident independent outage events MTBF
analysis in itself is not a sufficient justification for PBRC adjustment since WECC
does not require planning for two unrelated independent line outage events. The
PBRC adjustment assumes that there will be some dependency for lines sharing
common rights of way.
4. Provide a Corrected Estimate of MTBF (based on Project Robustness Features)
•
•
A partial list of events that may be justified out is included in section 3.6 of
the PBRC process.
Consider various robustness features introduced to reduce the risk of outage.
For examples see reference [1].
2
Table 2, Phase I: Event Probability, Development and Implementation Plan, WSCC Reliability
Subcommittee Probabilistic Methods Work Group report, June 25, 1998
-2-
6/3/2003
This step was completed after winnowing the uncorrected SGS database to 17
corridors by excluding lines with 1.5 single contingency outages per year per line
as a cutoff because of SRP’s low outage rates on the 500kV system. However, the
Mead-Pheonix 500kV line was included in the sample even though its outage rate
is greater than the cutoff. This developed the corrected database of 17 circuits.
Since the resulting sample contained only two instances of N-2 outages, an N-2
MTBF analysis using the Weibull method was conducted. This analysis identified a
MTBF mean and median of 2,824 year and 1,675 years respectively. This analysis
yielded MTBF results for coincident independent outage events and not an N-2
MTBF for common mode dependent outage events. A MTBF analysis for a single
event causing a double line outage (i.e., dependent event) was not completed
because a database of common mode double line outages could not be established
from the SGS Statistical Services data.
RPEWG raised a concern regarding the 1.5 outages per year per line for single
contingency outages cutoff. The report states that SRP chose 1.5 single
contingency outages per year per line as a cutoff for the similar lines or “like” lines
database because of SRP’s low outage rates on the 500 kV system.
RPEWG critically reviewed the PV-WW robustness features using RPEWG’s
Robust Line Design Features document. Of special interest were (1) the
opportunity for a tower of one line to fall into the second line and (2) the
opportunity of the Mead-Phoenix line falling into the PV-WW lines. The
robustness of the PV-WW lines to address these two interests is described below.
The conclusion from this review was that the “… analysis confirms that the PVWW lines meet or exceed the Robust Line Design Features required for PBRC
adjustment.” See the attached RPEWG Evaluation, Palo Verde to Westwing
Double 500 kV Lines, Robust Line Design Features document.
Robust line design minimizes the opportunity for initial transverse tower failure to
strike the second line. A detailed study of the transmission towers conducted by
Power Engineers, Inc. was included in the report. Based on extensive analysis it
was concluded that the initial transverse tower failure does not strike the adjacent
line’s tower, however any towers that are “hauled down” as secondary failures
could impact the tower of the other line. The timing between the sequences of
tower failures is estimated to be 1 to 3 seconds. . The Power Engineers, Inc.,
analysis used standard modeling assumptions for member performance, including
nonlinear P-delta affects. The failure scenarios were based on member
performance relative to the finite element modeling assumptions. A critical
parameter in the modeling assumptions is that the tower was originally detailed to
minimize member connection eccentricities. Therefore, it is recommended that the
structural details of the 5T2 and 5T3 towers be reviewed by a experienced Detailer
and/or Professional Tower Design Engineer to determine that these towers satisfy
standard detailing practices that will result in minimizing connection eccentricities
and validates member connection modeling assumptions.
-3-
6/3/2003
Robust line design minimizes the chance of the Mead-Phoenix 500 kV line falling
into the PV-WW lines. There is one line crossing over the PV-WW lines. The
Mead-Phoenix 500kV line crosses over the PV-WW lines a few miles from the
Westwing Receiving station. A great deal of engineering was placed on developing
design criterion that exceeded industry standards. The conductor design tensions
are significantly less than the N.E.S.C. requirements for the Mead-Phoenix line.
The phase conductor insulator and hardware assemblies have a built-in
redundancy. The assembly could have one or two insulator failures, continue in
service and still meet the N.E.S.C. mechanical strength requirements to 50%
Specified Mechanical Load. All hardware has a rated strength of at least two times
the final design tension of the phase wire positions. All hardware is redundant,
with the exception of on Yoke Plates (rated strength 150,000 lbs.) In the event of a
hardware failure, the remaining components have sufficient strength to hold and
still meet N.E.S.C. minimum requirements. Also, according to UFSAR, the failure
of this line at the crossing over the PV-WW lines is no longer postulated under the
revised 10CFR50.59 rules.
The justification for protective relying could be improved by model line testing to
reduce risk of sympathetic tripping.
5. Complete Exposure Analysis. (Refer to example)
About 2% of the time the double PV-WW outage will be the critical outage. As a
result of the recent Rudd line installation, the PV-WW double line outage is not the
most critical outage as was the case when SRP made its original PBRC Adjustment
application. SRP operations estimate that the combined scheduled and unscheduled
outage time when the PV-WW double would be the critical outage will be 150 to
200 hrs per year total for all lines combined. An existing safety net will remain in
place during the time when the PV-WW double outage is the critical outage.
6. Illustrate the Consequences of Outage (Refer to example)
An existing RAS will remain in service and will act as a safety net. There is an
existing RAS scheme that can respond to the double PV-WW outage. The situation
during which the PV-WW 500kV line outage becomes the most limiting
contingency, SRP proposes to maintain the RAS scheme to protect against this
outage. This would then be a system safety net.
WECC voltage dip criteria will not be violated. SRP’s operating studies show that
with any one of the circuits emanating from Palo Verde initially out of service, a
subsequent SLG fault on the Palo Verde 500kV Bus that would take both PV-WW
500kV lines out of service will not exceed the 30% voltage dip criterion.
-4-
6/3/2003
7. Conclude the how the adjustment meets the PBRC criteria (refer to example)
There are two points of reference used by RPEWG in evaluating whether or not a
PBRC Adjustment request should be approved:
1. A review of the outage data including a MTBF analysis.
2. A review of the Robust Line Design Features of the line.
The historical performance of the PV-WW line has been excellent. There has been
no double line outage in the history of the PV-WW lines. One of the PV-WW
circuits has had two outages and the second circuit has had zero outages. Thus, a
MTBF analysis using only PV-WW data was impossible.
There are a limited number of lines sharing the same corridor in the Southwest
that are similar to the PV-WW lines. A data sample was developed and a MTBF
analysis using a Weibull process was presented. This MTBF analysis provided
information for coincident independent line outages and not for common mode
dependent caused outage. A MTBF analysis for a common cause double line
outage was not presented because of a lack of similar line data.
RPEWG used its Robust Line Design Features working paper to evaluate the
robustness of the line design. The analysis compared thirteen factors in examining
the line robust features. The conclusion from this evaluation was that the PV-WW
lines meet or exceed the Robust Line Design Features required for PBRC
adjustment.
RECOMMENDATION: RPEWG members voted to approve and recommend
that PCC also approve SRP’s request for PBRC Adjustment for the double PV-WW
line outage from Category C to Category D. RPEWG cannot make a
recommendation regarding MTBF of greater than 300 years.
-5-
Draft #2 5/30/2003
Attachment 5
RPEWG Evaluation
Palo Verde to Westwing Double 500 kV Lines
Robust Line Design Features
BACKGROUND
Salt River Project (“SRP”) requested that the RPEWG approve an adjustment of the Palo
Verde to Westwing (“PV-WW”) 500 kV double line outage from the NERC/WECC
Category C to Category D. The report submitted by SRP contains significant detail
describing the case presented for approval, which including a probability analysis. SRP’s
analysis followed the “Seven Step Process For PBRC Adjustment”1 to the extent
permitted by line characteristics and historical outage information.
The corrected MTBF of 2824 years represented in the report is the expected result for
overlapping outage of two independent events. A MTBF analysis for a single event
causing a double line outage was not calculated because a database of common mode
double line outages could not be established from the SGS Statistical Services data2. The
independent events MTBF analysis in itself is not a sufficient justification for PBRC
adjustment since WECC does not require planning for two unrelated independent line
outage events. The PBRC adjustment assumes that there will be some dependency for
lines sharing common rights of way.
Specifically, the PV-WW case presented two areas of particular concern: 1) A line
crossing issue (The Mead Phoenix line crosses over the PV-WW corridor) and 2) the
issue of whether the circuit separation is sufficient to prevent towers falling into each
other. Additional information regarding these dependent events was provided consisting
of construction details, failure analysis, and historical performance data.
As permitted in Step 4 of the Seven Step Process for PBRC Adjustment, RPEWG
approval can be evaluated in terms of the robustness of its line design. The RPEWG used
the Robust Line Design Features3 as the standard to compare the PV-WW for a robust
line design. This evaluation follows.
ANALYSIS
Compliance is examined in the context of the eight risk factors (R1-R8) outlined in the
Robust Line Design Features. Reference is made to the pertinent sections of the report
measures taken to achieve robustness.
1
PCC Handbook
In the history of the two PV-WW lines, one line has not had any outage and the other line has had two
outages.
3
See attached “Robust Line Design Features, RPEWG working paper 5/28/02” that is referenced in WECC
Planning Coordination Committee Handbook, Revised October 2002, page IX-2.
2
-1-
Draft #2 5/30/2003
R1 Risk of fire affecting both lines
•
•
•
•
Sparse low vegetation (photos provided) in low desert terrain (page M-3).
Twice yearly helicopter patrols to identify and take corrective action of vegetation
problems (page M-5).
SRP likely to de-energize line during a fire (page M-5).
Risk of fire extremely unlikely (page M-5)
R2 Risk of one tower falling into another line
•
•
•
•
•
130 foot line separation (page M-10)
Overdesign of tower foundations 137-199% (page M-4 and photo of footing)
Separate document providing tower failure mode analysis by Power Engineering.
This concludes that an initial failure of a tower is not likely to jeopardize the
parallel line.
Average tower capacity is 100 mph wind gust. Average return period of 90 mph
gust is 150 years (page M-5)
The Power Engineers, Inc., analysis used standard modeling assumptions for
member performance, including nonlinear P-delta affects. The failure scenarios
were based on member performance relative to the finite element modeling
assumptions. A critical parameter in the modeling assumptions is that the tower
was originally detailed to minimize member connection eccentricities. Therefore,
it is recommended that the structural details of the 5T2 and 5T3 towers be
reviewed by a experienced Detailer and/or Professional Tower Design Engineer to
determine that these towers satisfy standard detailing practices that will result in
minimizing connection eccentricities and validates member connection modeling
assumptions.
R3 Risk of a conductor from one line being dragged into another line
•
[see R5]
R4 Risk of lightning strikes tripping both lines
•
•
•
•
Low isokeraunic level one of lowest in Western US (citation?)
Estimated lightning flash densities of 1 to 2.5 strikes per square mile per year
13-40 days per year when thunder is heard (page M-2)
Both lines equipped with shield wires (page M-2)
R5 Risk of an aircraft flying into both lines
•
•
Lines do not qualify as a hazard to Public Use or Military as defined by FAA
(page M-1).
Closest public airport 3.2 miles (Buckeye, page M-1)
-2-
Draft #2 5/30/2003
•
•
Shield wire marker balls used to near Buckeye in area of known potential route
for small airplanes (page M-10).
Lines well below FAA minimum height for aircraft (page M-10)
R6 Risk of station related problems resulting in loss of two lines for a single event
•
Breaker and half design at both Palo Verde and Westwing (pages M-11,12)
R7 Risk of snow or earth slides
•
No threat from snow or ice accumulations (page M-3).
R8 Risk of loss of two lines due to an overhead crossing
• Low line tensions used to minimize risk (page M-6)
• Insulator and hardware assemblies have built in redundancy (page M-6)
• NRC determined not credible from standpoint of loss of cooling risk (page M-6)
ADDITIONAL FACTORS ADDRESSED IN REPORT
R9 Earthquakes
•
Damage due to earthquake highly unlikely. Lowest western US Category B (page
M-3)
R10 Flood
•
•
Detailed description of distance from river channels and flood reaches (page M-3)
Closest flood encroachment not less than 2 miles (page M-3)
R11 Protective Relaying
•
•
•
•
No high-speed reclosure due to avoid possible generator shaft impacts (page M-9)
Redundant microprocessor relay technology (page M-9)
Alarming allows quick identification and correction of problems (page (M-9)
The justification would be further improved by information on model line testing
to reduce risk of sympathetic tripping.
R12 Faults Caused by Birds
•
Largest birds capable in coming in contact with wires are small and incapable of
causing a phase to phase fault (page M-10 and picture provided)
-3-
Draft #2 5/30/2003
R13 Maintenance
•
•
•
•
•
Damage caused by vandalism (probably the biggest risk) is mitigated by spring
and fall aerial patrol and twice per year ground patrol (page M-10)
Areas of insulator contamination by birds are protected by protective devices
above the insulator (page M-10)
Insulator washing, with tools and training available to SRP, can be done with the
lines energized (page M-10)
Changing insulators, repair of gunshot conductor, replacement of other hardware
are done with the lines energized (page M-10)
Deteriorating items, such as spacer dampers, change as they deteriorate and fail
(page M-10)
SUMMARY of Robust Line Design
Element
R1
R2
Risk of
Fire affecting both lines
One tower falling into another
line
R3
Conductor from one line being
dragged into another line
R4
Lightening strikes tripping both
lines
R5
Aircraft flying into both lines
R6
Station related problems resulting
in loss of tow lines for a single
event
R7
Snow or earth slides
R8
Loss of two lines due to an
overhead crossing
Other Information provided
R9
Earthquakes
R10
Flood
R11
Protective relaying
R12
R13
Faults caused by birds
Maintenance
Evaluation
Very low risk
Low risk (suggest expert review of joint
detail in addition to finite element
analysis to strengthen justification)
See R5
Very low risk
Very low risk
Very low risk
Very low risk
Very low risk
Very low risk
Very low risk
Low risk (suggest model line testing of
relays for very low risk to further reduce
risk of sympathetic tripping
Mitigated – very low risk
Aggressive practice
RPEWG EVALUATION
The above analysis confirms that the PV-WW lines meet or exceed the Robust Line
Design Features required for PBRC adjustment.
-4-