Palo Verde to Westwing Double Line Outage Probability Analysis SRP

Tatyana Len Dhaliwal [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) An MTBF estimated by a traditional statistical reliability analysis method is on average once in 2824 years. 2) 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. 3) Both Westwing and Palo Verde switchyard use breaker and a half arrangement. 4) 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. 5) 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. 6) 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) 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. 8) 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. 9) The risk of earthquakes in Maricopa County is the lowest in the Western US. 10) The risks of flood, snow, and fire are negligible. 11) The PV-WW foundations are over designed in the range of 137 to 199%. 12) The lattice tower design is conservative for weather related loads. 13) Lines are designed with state of the art spacer dampers to control conductor motion. 14) The insulation level exceeds EPRI’s guidelines. 15) 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. 16) 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, north- northeast 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.

Current Current Jun/2003 Generators Total Equivalized Equivalized Palo Verde Nuclear(existing) 3861MW 3861MW 3861MW Duke Arlington Valley(existing) 600MW 600MW 600MW Pinnacle West Red Hawk(existing) 1000MW 1000MW 1000MW Sempra Mesquite(2003) 525MW 525MW 1050MW PGE National Harquahala(2003) 1148MW 1148MW 1148MW TECO Panda Gila River(2003)(network) 1560MW 780MW* 1040MW* Total Generation Value 8694MW 7914MW 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)

APPROXIMATE POST- RAS 500KV LINE IOS PRE-RAS RAS VOLTAGE DROP GEN TRIPPED (WORST CASE) (MW)

HASSAYAMPA/NORTH GILA CASE DIVERGED 1550MW 22% (AFTER 3 SEC RUN)

KYRENE/JOJOBA CASE DIVERGED 1500MW 24% (AFTER 3 SEC RUN)

PALO VERDE/RUDD CASE DIVERGED 1070MW 26% (AFTER 3 SEC RUN)

PALO VERDE/DEVERS CASE DIVERGED 600MW 19% (AFTER 3 SEC RUN)

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. 2003 SUMMER 2. PV/HAA GEN NET=8192MW W/ PV = -436MVAR, HAA = -364MVAR (NET –800MVAR). 3. SLG FAULT ON PV 500KV BUS; WITH TWO PV-WWG CKTS OUT. 4. 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 twice- yearly 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 Mead- Phoenix 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 Verde- Westwing 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:

β− β β 1   µ σ =  t  −  t   f(t : , )   exp    η  η    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  ≤ η β = − −  t   > Pr(T t : , ) 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:

β− β 1 µ σ =  t  > h(t : , )   , t 0 η  η 

For the three functions,

t is a random var iable of time µ is the expected value (mean) of time σ is the standard 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 Mead- Phoenix line was included in the sample. The sample consisted of:

Number Of 17 Circuits Circuit Years 134.06 Circuit Mile-years 10,409 Sum Miles 1303 Total Outages 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 pre- existing 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 protection- related 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: Time Between Failures (Hours)

Standard 99% Confidence Limits Parameter Estimate Error Lower Upper

Weibull Scale 6503.1092 1359.1839 3795.8888 11141.1140 Weibull Shape 0.5259 0.0518 0.4080 0.6779

Mean 11897.7072 Mode 0.0000 Median 3239.4690 Standard Deviation 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: Outage Duration (Seconds) Standard 99% Confidence Limits Parameter Estimate Error Lower Upper

Weibull Scale 4985.5161 1010.8345 2957.2999 8404.7516 Weibull Shape 0.5260 0.0399 0.4325 0.6396

Mean 9120.6963 Mode 0.0000 Median 2483.5506 Standard Deviation 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 1: Line X Coincidence 2: Line Y and Line Y Outages outage is within 30” of the end of a Line X Outage

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: X-Y Circuit Double Contingency TBF Years Standard 99% Confidence Limits Parameter Estimate Error Lower Upper

Weibull Scale 2585.3309 144.0207 2239.7451 2984.2395 Weibull Shape 0.8446 0.0310 0.7685 0.9282 Mean 2824.1149 Mode 0.0000 Median 1675.1157 Standard Deviation 3359.4214

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 Estimate (YRS) 0.1 0.72545382 0.2 1.64931322 0.5 4.88944322 1 11.1426631 2 25.4699836 5 76.7639898 10 180.016976 20 437.728146 30 762.750464 40 1167.06507 50 1675.11567 60 2331.10311 70 3220.8513 80 4541.81759 90 6940.63586 95 9478.10828 99 15770.1269 99.9 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 BPA- type 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) An MTBF estimated by a traditional statistical reliability analysis method is on average once in 2824 years. 2) 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. 3) Both Westwing and Palo Verde switchyard use breaker and a half arrangement. 4) 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. 5) 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. 6) 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) 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. 8) 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. 9) The risk of earthquakes in Maricopa County is the lowest in the Western US. 10) The risks of flood, snow, and fire are negligible. 11) The PV-WW foundations are over designed in the range of 137 to 199%. 12) The lattice tower design is conservative for weather related loads. 13) Lines are designed with state of the art spacer dampers to control conductor motion. 14) The insulation level exceeds EPRI’s guidelines. 15) 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. 16) 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,Springerville- Greenlee, Springerville –Vail were removed due to unfavorable weather conditions.

Circuit Name Owner Circuit Circuit Circuit -Miles Years Mile- Years CHOLLA - FOUR CORNERS #1 APS 159 8 1273 CHOLLA - FOUR CORNERS #2 APS 159 8 1273 CHOLLA - PINNACLE PEAK APS 130 8 1041 ELDORADO (SCE) - MOENKOPI APS 217 8 1737 FOUR CORNERS - MOENKOPI APS 179 8 1436 MOENKOPI - NAVAJO APS 76 8 608 MOENKOPI - YAVAPAI APS 101 4.6 466 NAVAJO - WESTWING APS 256 8 2049 NORTH GILA - PALO VERDE (SRP) SWYD. APS 114 8 913 WESTWING - YAVAPAI APS 79 4.6 364 BA - BLACKWATER PNM 223 12 2674 NORTON - BA PNM 42 12 506 IMPERIAL VALLEY-MIGUEL SDGE 84 12 1003 IMPERIAL VALLEY-NORTH GILA SDGE 81 12 968 NORTH GILA-PALO VERDE SDGE 114 12 1374 BROWNING SILVERKG SRP 39 1 39 BROWNING KYRENE SRP 19 1 19 CHOLLA CORONADO SRP 73 11 808 CORONADO SILVERKG SRP 180 11 1980 KYRENE PALOVERDE SRP 75 11 826 KYRENE SILVERKG SRP 58 10 580 MEAD - PHOENIX SRP 242 5.7 1393 PALOVERDE WWG 1 SRP 45 11 496 PALOVERDE WWG 2 SRP 45 11 496 AEPCO_GREENLEE TEP . 7 . AEPCO_VAIL_BICKNELL_345KV TEP . 7 . CORONADO_SPRINGERVILLE_345KV_ LINE_ 311 TEP 22 7 154 EPE_GREENLEE_HIDALGO TEP . 7 . EPE_SPRINGERVILLE_LUNA_345_KV_LINE TEP . 7 . GREENLEE_VAIL _345KV LINE_303 TEP 128 7 897 SAGUARO_TORTOLITA_500KV_LINE_501 TEP 1 7 7 SAN JUAN_MCKINLEY_#1_345KV_LINE_301 TEP 90 7 630 SAN JUAN_MCKINLEY_#2_345KV_LINE_306 TEP 90 7 630 VAIL_SOUTH_345KV_LINE_304 TEP 14 7 98 WESTWING_SOUTH_345KV_LINE_305 TEP 178 7 1247 TOLK STA TO EDDY CO INTG XCEL 158 5 790 TUCO INTG TO OKLAUNION (PSO) XCEL 162 5 810

Table 2: SGS Study Southwest US Outage Data for SRP PV-WW Probabilistic Risk Assessment

ANONYMOUS OUTAGE DATE / RESTORE DATE / TBF SGS STUDY CAUSE CODE CENSOR CIRCUIT ID TIME TIME HOURS 500-0100 01JAN02:00:00 . 96432 1 500-0400 16MAR92:19:21 17MAR92:04:52 10579 EXTERNAL 0 500-0400 17MAR92:04:53 17MAR92:04:53 10 EXTERNAL 0 500-0400 29DEC92:14:20 29DEC92:14:25 6897 WEATHER 0 500-0400 29DEC92:14:26 29DEC92:14:26 0 WEATHER 0 500-0400 25JUL93:10:31 25JUL93:10:39 4988 OTHER 0 500-0400 25JAN94:10:54 25JAN94:14:27 4416 WEATHER 0 500-0400 25JAN94:14:28 25JAN94:14:39 4 WEATHER 0 500-0400 12OCT97:06:38 12OCT97:07:45 32536 EXTERNAL 0 500-0400 11NOV97:11:12 11NOV97:11:47 725 OTHER 0 500-0400 31JUL00:00:24 31JUL00:00:26 23821 SYSTEM PROTECTION 0 500-0400 01JAN01:00:00 . 3696 1 500-0410 01MAY01:11:10 01MAY01:12:23 2891 SYSTEM PROTECTION 0 500-0410 01JAN02:00:00 . 5869 1 500-0420 01MAY01:11:10 01MAY01:12:21 2891 SYSTEM PROTECTION 0 500-0420 01JAN02:00:00 . 5869 1 MEAD-PHX 03JUL96:00:00 03JUL96:00:06 2136 LIGHTNING 0 MEAD-PHX 28OCT96:00:00 28OCT96:01:51 2808 SYSTEM PROTECTION 0 MEAD-PHX 07JAN97:00:00 07JAN97:00:04 1704 OTHER 0 MEAD-PHX 27JAN97:00:00 27JAN97:00:09 480 SYSTEM PROTECTION 0 MEAD-PHX 05APR97:00:00 05APR97:00:09 1632 OTHER 0 MEAD-PHX 15JUL97:00:00 15JUL97:00:39 2424 OTHER 0 MEAD-PHX 29AUG97:00:00 29AUG97:00:07 1080 LIGHTNING 0 MEAD-PHX 07JAN98:00:00 07JAN98:00:08 3144 OTHER 0 MEAD-PHX 11SEP98:00:00 11SEP98:03:07 5928 LIGHTNING 0 MEAD-PHX 30NOV98:00:00 30NOV98:00:06 1920 UNKNOWN 0 MEAD-PHX 19JAN99:00:00 19JAN99:01:08 1200 OTHER 0 MEAD-PHX 20JAN99:00:00 20JAN99:00:52 24 OTHER 0 MEAD-PHX 21JUL99:00:00 21JUL99:00:06 4368 EQUIPMENT 0 MEAD-PHX 01JAN02:00:00 . 21480 1 PV-WW1 12JAN01:04:03 12JAN01:11:50 87940 EXTERNAL 0 PV-WW1 01JAN02:00:00 . 8492 1 PV-WW2 01JAN02:00:00 . 96432 1 XXX 0001 21OCT98:07:33 21OCT98:16:05 21152 LIGHTNING 0 XXX 0001 01JAN01:00:00 . 19264 1 XXX 0005 31MAY96:12:05 31MAY96:12:09 204 EQUIPMENT 0 XXX 0005 21OCT98:08:46 21OCT98:11:09 20949 EQUIPMENT 0 XXX 0005 02NOV98:13:03 02NOV98:16:33 292 EQUIPMENT 0 XXX 0005 05NOV98:16:03 05NOV98:17:03 75 EQUIPMENT 0 XXX 0005 20APR99:07:33 20APR99:07:55 3976 OTHER 0 XXX 0005 01JAN01:00:00 . 14920 1 XXX 0008 26MAY93:21:51 26MAY93:21:53 3502 WEATHER 0 XXX 0008 03JUN93:07:28 03JUN93:07:42 178 EQUIPMENT 0 XXX 0008 15FEB94:06:58 15FEB94:18:40 6168 EQUIPMENT 0 XXX 0008 24JUN94:17:35 24JUN94:20:41 3107 VEGETATION 0 XXX 0008 12SEP94:12:06 13SEP94:11:07 1915 LINES 0 XXX 0008 29FEB96:08:36 29FEB96:08:46 12837 EQUIPMENT 0 XXX 0008 18MAR96:07:43 18MAR96:07:59 431 UNKNOWN 0 XXX 0008 20AUG96:16:12 20AUG96:16:13 3728 WEATHER 0 XXX 0008 04NOV96:14:45 04NOV96:14:47 1823 UNKNOWN 0 XXX 0008 15MAR99:08:36 15MAR99:08:47 20658 LINES 0 XXX 0008 23NOV99:09:06 24NOV99:16:05 6073 OTHER 0 XXX 0008 24AUG00:17:32 24AUG00:17:34 6608 WEATHER 0 XXX 0008 01JAN01:00:00 . 3102 1 XXX 0011 11JAN90:00:00 11JAN90:00:01 240 EXTERNAL 0 XXX 0011 28DEC92:00:00 28DEC92:00:01 25968 EXTERNAL 0 XXX 0011 27MAR96:00:00 27MAR96:01:01 28440 OTHER 0 XXX 0011 30JAN98:00:00 30JAN98:01:00 16176 EQUIPMENT 0 XXX 0011 02APR99:00:00 02APR99:00:46 10248 OTHER 0 XXX 0011 12MAY99:00:00 12MAY99:00:06 960 SYSTEM PROTECTION 0 XXX 0011 29JUL00:22:07 29JUL00:22:10 10678 WEATHER 0 XXX 0011 01JAN02:00:00 . 12482 1 XXX 0015 06SEP95:16:51 06SEP95:16:51 5969 WEATHER 0 XXX 0015 01JAN02:00:00 . 55399 1 XXX 0027 05FEB99:05:17 05FEB99:05:22 79733 EQUIPMENT 0 XXX 0027 05FEB99:11:26 05FEB99:20:14 6 EQUIPMENT 0 XXX 0027 01JAN02:00:00 . 25453 1 XXX 0040 06NOV97:14:29 06NOV97:17:14 24974 SYSTEM PROTECTION 0 XXX 0040 26DEC99:04:23 26DEC99:04:24 18710 SYSTEM PROTECTION 0 XXX 0040 04MAY00:07:56 04MAY00:09:49 3124 EQUIPMENT 0 XXX 0040 13JUL01:00:58 13JUL01:05:36 10433 OTHER 0 XXX 0040 01JAN02:00:00 . 4127 1 XXX 0041 14AUG90:07:04 14AUG90:07:11 5407 LIGHTNING 0 XXX 0041 15SEP90:04:08 15SEP90:21:34 765 EXTERNAL 0 XXX 0041 01SEP91:08:00 01SEP91:08:06 8428 WEATHER 0 XXX 0041 21OCT91:14:27 21OCT91:14:32 1206 LINES 0 XXX 0041 02DEC95:04:29 02DEC95:04:37 36062 UNKNOWN 0 XXX 0041 02DEC95:04:44 02DEC95:10:03 0 UNKNOWN 0 XXX 0041 10AUG96:15:48 10AUG96:18:27 6059 EXTERNAL 0 XXX 0041 04NOV98:14:05 04NOV98:14:30 19582 OTHER 0 XXX 0041 01MAR99:22:14 01MAR99:22:21 2816 SYSTEM PROTECTION 0 XXX 0041 03OCT00:22:03 03OCT00:22:11 13968 LINES 0 XXX 0041 27NOV00:09:31 27NOV00:10:28 1307 WEATHER 0 XXX 0041 03JUN01:04:21 03JUN01:04:33 4507 WEATHER 0 XXX 0041 03JUN01:11:40 03JUN01:14:58 7 WEATHER 0 XXX 0041 04NOV01:03:47 04NOV01:04:04 3688 UNKNOWN 0 XXX 0041 11NOV01:13:37 11NOV01:13:48 178 UNKNOWN 0 XXX 0041 01JAN02:00:00 . 1210 1 XXX 0044 05AUG93:19:26 05AUG93:19:36 5203 WEATHER 0 XXX 0044 04MAY95:13:48 04MAY95:14:09 15282 EXTERNAL 0 XXX 0044 05MAY95:02:01 05MAY95:02:13 12 EXTERNAL 0 XXX 0044 15APR96:04:32 15APR96:07:09 8307 EQUIPMENT 0 XXX 0044 10AUG96:15:48 10AUG96:17:03 2819 EXTERNAL 0 XXX 0044 07APR98:14:22 07APR98:14:24 14519 LIGHTNING 0 XXX 0044 16NOV99:19:44 17NOV99:03:53 14117 EQUIPMENT 0 XXX 0044 01JAN01:00:00 . 9868 1 XXX 0049 16MAY95:05:00 16MAY95:05:09 20765 EQUIPMENT 0 XXX 0049 05NOV97:21:49 05NOV97:21:54 21713 UNKNOWN 0 XXX 0049 10NOV98:20:00 10NOV98:20:20 8878 UNKNOWN 0 XXX 0049 05FEB99:06:17 05FEB99:06:22 2074 EQUIPMENT 0 XXX 0049 05FEB99:12:26 05FEB99:21:14 6 EQUIPMENT 0 XXX 0049 02NOV99:06:12 02NOV99:08:40 6474 LINES 0 XXX 0049 25JUL00:09:35 25JUL00:09:38 6387 UNKNOWN 0 XXX 0049 31JUL00:00:24 31JUL00:14:01 135 EXTERNAL 0 XXX 0049 01JAN01:00:00 . 3696 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 Engineering Review of Lines by SRP

Select “Similar” Lines Figure 1 and <1.5 Outages Table 1, 2 per Year

Fit Weibull Distribution for TBF Hours and Duration Figures 2, 3 using Outage Data

Using Weibull Parameters, Generate 1,000,000 Random Events (TBF and Figures 4-7 Duration) for Circuits X, Y and Z

Evaluate Circuit X and Y Figure 8 (sample), for Coincidence. Figure 9 TBF (all) 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 Report Body Circuits, N-1, N-2 and N-3

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 1994 1995 1996 1997 1998 1999 2000 2001 2002 CENSOR LEFT FAILED 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 Scale 6503 95 Shape 0.526 90 Uncensored 73 80 Right Censored 15 70 Left Censored 15 60 Conf. Coeff. 99% 50 Fit ML 40 30

20 Percent 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 EXPONENTIAL DISTRIBUTION

99.9 Scale 6921 99.9 Scale 3138 Shape 0.423 Uncensored 73 Uncensored 73 Right Censored 15 Right Censored 15 Left Censored 15 99 Left Censored 15 99 Fit LSXY R Squared 0.942 95 Fit LSXY 95

80 80 60 60

40 40 30 30 20 20 Percent Percent 10 10

5 5

2 2

1 1

.5 .5 .01 .1 1 10 100 1000 10000 100000 .01 .1 1 10 100 1000 10000 100000 Weibull Plot For (t1, t2) Exponential Plot For (t1, t2)

LOGNORMAL DISTRIBUTION EXTREME VALUE DISTRIBUTION

99.9 Location 7.601 99.9 Location 10233 Scale 2.814 Scale 5599 Uncensored 73 Uncensored 73 Right Censored 15 Right Censored 15 Left Censored 15 99 Left Censored 15 R Squared 0.827 Interval Censored 2 Fit LSXY R Squared 0.615 99 95 Fit LSXY 80 95 60 90 40 80 30 70 20 60

Percent 50 Percent 40 10 30 20 5 10 2 5 2 1 1 .5 .5 .01 .1 1 10 100 1000 10000 100000 0 10000 20000 30000 40000 Lognormal Plot For (t1, t2) 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 Weibull Fit for Duration 99 Scale 4986 Shape 0.526 95 Uncensored 99 80 Conf. Coeff. 99% 70 Fit ML

50 40 30 20

Percent 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 Scale 6643 90 Shape 0.527 60 Uncensored 10000 40 Conf. Coeff. 99% 20 Fit ML 10 5 2 1 .5

Percent .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 Scale 6501 90 Shape 0.532 60 Uncensored 10000 40 Conf. Coeff. 99% 20 Fit ML 10 5 2 1 .5

Percent .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 Scale 5066 90 Shape 0.527 60 Uncensored 10000 40 Conf. Coeff. 99% 20 Fit ML 10 5 2 1 .5

Percent .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 Scale 5012 90 Shape 0.531 60 Uncensored 10000 40 Conf. Coeff. 99% 20 Fit ML 10 5 2 1 .5

Percent .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 Scale 2585 95 Shape 0.845 Uncensored 487 70 Conf. Coeff. 99% 50 Fit ML 30 20

10

5

Percent 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

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 PV- WW 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 2 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.

-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 Risk of Evaluation R1 Fire affecting both lines Very low risk R2 One tower falling into another Low risk (suggest expert review of joint line detail in addition to finite element analysis to strengthen justification) R3 Conductor from one line being See R5 dragged into another line R4 Lightening strikes tripping both Very low risk lines R5 Aircraft flying into both lines Very low risk R6 Station related problems resulting Very low risk in loss of tow lines for a single event R7 Snow or earth slides Very low risk R8 Loss of two lines due to an Very low risk overhead crossing Other Information provided R9 Earthquakes Very low risk R10 Flood Very low risk R11 Protective relaying Low risk (suggest model line testing of relays for very low risk to further reduce risk of sympathetic tripping R12 Faults caused by birds Mitigated – very low risk R13 Maintenance 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-