ML073540262

Afpb Staff Response to NFPA 805 Transition Pilot Plant FAQ 07-0035, Revision 0

ML073540262
Person / Time
Site:	Harris
Issue date:	12/20/2007
From:	Funk D, Js Hyslop, Joglar F, Kazarians M, Najafi B, Nowlen S, Wyant F Electric Power Research Institute, NRC/RES/DRA/FRB
To:
References
EPRI TR-1011989, FAQ 07-0035, Rev 0, NUREG/CR-6850
Download: ML073540262 (12)

Text

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT RES & EPRI Team Response to FAQ 07-0035 NUREG/CR-6850, EPRI TR-1011989 Bus Duct Counting Guidance for High Energy Arcing Faults Final, Revision 2.1 - 12/18/2007 S. Nowlen, D. Funk, F. Wyant, M. Kazarians, F. Joglar, B. Najafi, J.S. Hyslop

Background

FAQ 07-0035 requests clarification regarding the treatment of high energy arc faults specific to bus duct failures. Appendix M of the consensus methodology document (NUREG/CR-6850, EPRI TR-1011989) provides guidance for the treatment of high energy arc faults in switchgear and load centers, but does not cover the treatment of bus duct fires. Further, while the document mentions bus ducts as a potential source of high energy arc fault events, no fire event frequency or treatment guidance is provided. The original FAQ 07-0035, dated June 1, 2007, as provided to the team for response is attached.

It should be noted that, in developing guidance for fire ignition sources, the absence of guidance on the treatment of bus duct arc faults and fires was an unintended oversight on the part of the report authors and should not be taken to indicate that such fires need not be treated. This FAQ response corrects this inadvertent omission and provides the required guidance for the treatment of bus duct arc fault fires.

Resolution Approach The team has reviewed the fire event database and identified events involving bus ducts.

The event set was supplemented by a set of fire events identified by the NRC staff as potentially relevant to the bus duct arc fault question. In addition, a public meeting was held (August 2007, ADAMS ML072560081) to discuss the teams preliminary insights, and to gain additional input from stakeholders. The proposed resolution is based on, and consistent with, all of the input received from these resources.

Technical Resolution Classification of bus ducts by type A review of bus duct physical configurations as typically used in power plant applications has resulted in a recommended practice that would first classify each bus duct as falling into one of four general categories:

• Non-segmented or continuous bus ducts: A bus duct where the bus bar associated with each power phase is comprised of single length of metal bar connecting two end-devices (e.g., terminating within a cabinet or at a specific piece of electrical equipment) with no intermediate junctions, transitions, or DRAFT 1 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT terminations along the length of the bus bars. Typically, the bus bars are wholly contained within a grounded metal enclosure that runs the full length of the distance between the termination points.

o Non-segmented bus ducts tend to be comparatively short (on the order of no more than 12 feet in length) due to practical limits on the length of a single segment of bus bar.

o Examples:

A bus duct connecting a station service transformer to the associated switchgear cabinets where the transformer and switchgear are located in close proximity within a common fire area.

A bus duct connecting two separate banks of load cabinets located in close proximity (e.g., across an intermediate access walkway) and fed from a common power source.

• Segmented bus ducts: A bus duct where the bus bars are made up of multiple sections bolted together at regular intervals (transition points). Here, the bus bars are contained within open ended sections of metal covers that are bolted together to form a continuous grounded enclosure running the full distance between termination points. Segmented bus ducts are able to accommodate tap connections to supply multiple equipment termination points.

o Segmented bus ducts tend to be longer in comparison to the non-segmented bus ducts. Segmented bus ducts are used in cases where the required lengths and/or geometries make the use of non-segmented bus ducts impractical.

o The length of each segment may vary depending on supplier and installation details.

o Segmented bus ducts tend to connect end devices that are remote from each other.

o Example: A segmented bus duct might be used to connect an oil-filled transformer located in an outdoor area to equipment (e.g., switchgear) located inside the plant buildings.

• Cable ducts: A power conductor configuration that provides a function like a bus duct but uses a length of insulated electrical cable in lieu of metal bus bars. Cable ducts may be routed in a variety of ways, not necessarily within continuous runs of metal enclosures.

o Cable ducts can be as long as, or longer than, a segmented bus duct because there is no practical limit to the length of cable that can be obtained and installed.

o Cable ducts may be used in application conditions similar to either a segmented or non-segmented bus duct.

• Iso-phase bus ducts: A bus duct where the bus bars for each phase are separately enclosed in their own protective housing. The use of iso-phase buses is generally limited to the bus work connecting the main generator to the main transformer.

DRAFT 2 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT Selection of bus duct fire scenarios A review of the experience base for all types of bus ducts revealed one common characteristic; namely, that all of the identified bus duct arc fault events occurred either at the termination point of the duct or at a transition point along the length of a segmented bus duct.

With the exception of the iso-phase bus ducts, in those events occurring at the termination point, all had been included in the arc fault (switchgear and load centers) or catastrophic failure (transformers) event sets for the end devices. Hence, these events are already accounted for in the methodology and are treated as originating in the end device.

Because non-segmented bus ducts and cable ducts have no transition points other than the terminations at the end device, no treatment of bus duct faults/fires independent from the treatment of fires for the end devices is required. That is, arc faults for these two categories of bus ducts are inherently included in the treatment of the end device, and no further treatment is needed.

A review of available data indicates that events associated with iso-phase bus ducts also manifest themselves at the termination points (i.e., the main generator or main transformer) but these events had not been included in the associated end device frequencies. The potential effects of the iso-phase faults also appear unique in comparison to the end device (transformer or exciter) fires as recommended in the existing guidance. Hence, some additional treatment for iso-phase bus duct faults occurring at the termination points is needed.

For segmented bus ducts, a number of the identified fire events were manifested at bus transition points (a point where two segments of the bus duct are bolted together) rather than at the bus termination points. These events were generally attributed to loose bolted connections, to failed insulators, or to the accumulation of dirt/debris/contaminants in the bus duct. The key, however, is that the effects of the fault are manifested at identifiable transition points along the bus duct length. Fire scenarios for segmented bus ducts should, therefore, be postulated to occur at duct transition points (i.e., bolted connections).

Counting guidance The recommended practice for counting segmented bus ducts as a fire ignition source is to count the number of transition points. Based on input during the public meeting (referred to above) these transition points should be clearly identifiable based on visual observation or review of design drawings; the transition points for the bus bars will in most cases correspond to junctions in the outer ducting that surrounds the bus bars.

Hence, counting can be based on external observation and the counting of duct sections; it is not intended that the protective duct be removed to identify transition points. The plant-wide fire frequency is then partitioned to a specific location based on the number of DRAFT 3 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT transition points in the location of interest divided by the total number of transition points for the entire plant.

For iso-phase bus ducts, there should generally be one iso-phase bus per unit (an iso-phase bus includes all three phases). If there is more than one iso-phase bus, simply count the total number of iso-phase buses per unit.

Fire frequency and frequency partitioning The team has reviewed all of the identified candidate fire events. Tables 1 and 2 list the events used to estimate the frequencies for segmented bus ducts and for iso-phase ducts, respectively. The events listed were identified based on a review of the events in EPRIs Fire Event Data Base (FEDB) and on a similar review of events identified in various NRC documents (information notices, inspection reports, LERs, and staff reports) as provided by the NRR staff. Each event that included a bus duct was reviewed to determine (1) if the event meets the general criteria associated with a potentially challenging event consistent with the treatment of other fire ignition sources and (2) if the event was indeed uniquely associated with a bus duct arc fault and consequential fire rather than some other fire ignition source bin.

Table 1: Segmented Bus Duct Fire Events Used in Frequency Calculations FEDB Incident Event Date Description No:

195 4/15/80 Fire involved a supply bus located in a switchgear room.

575 3/19/87 A fault in a 6.9 kV feeder line to the in house buses of unit Aux transformer resulted in a fire and explosion outside the building.

678 3/2/88 A section of a bus bar was badly damaged due to installation failure and a subsequent fault. In addition to damage to the bus, insulation of several non-safety related cables located in a cable tray adjacent to the bus were damaged.

922 7/10/87 Insulation on a 4160 V bus bar failed. This condition resulted in a phase to ground fault which caused extensive damage to the bus bar and a fire.

994 10/14/91 Fault on bus bar caused the transformer to fail catastrophically leading to smoke and flames.

2426 5/15/00 A fault occurred on the 12 kV bus duct between the auxiliary transformer and two 12 kV buses. The main generator continued to feed the fault for 4-8 seconds. The sustained fault resulted in arcing in the 12 kV bus duct that jumped to and damaged a nearby 4.16 kV bus duct causing loss of three vital buses.

732 8/20/80 A short occurred on the bus work damaging the bus.

Not in 5/18/83 Startup bus failed because of a phase B to phase C fault, which FEDB propagated to ground. Further investigation revealed several degraded areas in the bus insulation at the support blocks.

DRAFT 4 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT Table 2: Iso-Phase Bus Duct Fire Events Used in Frequency Calculations FEDB Incident Event Date Description No:

792 7/15/1988 Arcing and fire was observed at 22kV iso-phase bus duct due to damaged ground straps and a deteriorated gasket between the cover and the duct.

929 10/9/1989 Multiple ground faults caused by aluminum debris in an iso-phase bus duct started a chain of events that led to three separate fires: (1) an oil fire in the 'B' main power transformer, (2) a hydrogen fire under the main generator, and (3) a small oil fire in the generator housing. In addition to site fire brigade, off-site fire departments were contacted to assist.

962 11/24/1992 A ground fault occurred between the main generator and the main unit transformers when an upper inspection window on the "A" phase bus duct came loose and fell inside. The gasket around the glass window contacted the high voltage bus disconnect link assembly inside the duct causing a phase-to-ground short circuit.

Not in 6/18/2004 An electrical fault in the one phase of the iso-phase bus lead to a fire FEDB near the main transformer at Vermont Yankee.1 The fire also involved oil leaking from a flange on the main transformer itself and resulted in severe damage to the low voltage bushing box on top of the Main Transformer, to the Generator PT Cabinet in the Turbine Building, and to the iso-phase bus duct itself.

The resulting set of relevant and potentially challenging events includes 8 events for segmented bus ducts and 4 events for iso-phase bus ducts. Note that, because of its significance, the June 18, 2004 Vermont Yankee iso-phase fire event is included in the frequency calculations. All other frequencies were based on events prior to December 31, 2000. The iso-phase event set, therefore, extends beyond the period covered by the FEDB; hence, the number of plant reactor years was adjusted to reflect plant operations through mid June 2004 for the iso-phase bus duct case only. Also, it was verified that no other iso-phase bus fires have been reported between January 1, 2001 and mid June 2004.

The resulting plant-wide fire event frequency for segmented bus duct arc fault failures and iso-phase bus duct fires is characterized by the frequency distributions presented in Table 3.

1

Reference:

Licensee Event Report (LER) 50-064-2004-003-01, Revision 1, June 14, 2005.

DRAFT 5 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT Table 3: Iso-Phase and Segmented Bus Duct Fire Frequency Distributions Total

1. of 5th 50th 95th Standard Frequency Bin Reactor Mean Events percentile percentile percentile Deviation Years Segmented Bus Duct 8 2486 9.7E-05 1.6E-03 1.1E-02 3.7E-03 1.1E-01 Iso-Phase Bus Duct 4 2164 4.5E-05 8.9E-04 6.5E-03 2.2E-03 6.4E-01 Partitioning of these fire frequencies to specific locations for the segmented bus ducts should be performed in accordance with the counting guidance provided above. That is, each transition point will be assumed to have an equal fraction of the total plant-wide fire frequency.

For the iso-phase bus, partitioning should assume that the likelihood of a fire is equal for each end of the bus (i.e., half of the frequency goes to the transformer end and half to the generator end) and that the initial fault is equally likely to occur in any one of the three phases (i.e., the fault initiates in one of the three phases, not all three concurrently).

Zone of influence for segmented (non-iso-phase) bus duct fires The zone of influence for a segmented bus duct arc fault fire is considered unique from that assumed for electrical cabinets. The experience base illustrates that the bus duct events generally involved a pool of molten conductor that forms within and then ejects itself outward. This material is expelled out of the bus duct, and may form a molten pool of metal on the floor or objects below, may splatter onto other nearby surfaces, and may ignite any combustible or flammable materials contacted. The recommended zone of influence is intended to reflect this experience base.

For reference, one well documented event considered prototypical of a bus duct fire occurred at Diablo Canyon on May 15, 20002. Figures 1 and 2 provide photographs taken after this event (as provided in the cited inspection report - see footnote). Note the breach of the bus duct at the fault point, and apparent surface damage of the neighboring duct. Damage is visible to the face panels on cabinets located below and to the side of the fault point. The photos show clear evidence (soot traces) that the surface-mounted components (e.g., the dial indicators, labels, switches, etc.) ignited and burned.

Individual points of charring on the panel face are taken as indicative of impinging molten metal droplets. The exterior surfaces of cabinets on both sides of the access walkway directly below the primary faulting point were damaged. The fire did not extend to the interior of these cabinets. Surface damage also occurred on cabinets on both sides of the aisle-way directly below the fault point, and extended to three adjacent panels on each side of the aisle.

2

References:

(1) U.S. NRC Information Notice 2000-14, Non-Vital Bus Fault Leads to Fire and Loss of Offsite Power, 9/27/2000 and (2) U.S. NRC Diablo Canyon Inspection Report No. 50-275/00-09; 50-323/00-09, July 31, 2000.

DRAFT 6 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT Based on the observed behavior, the recommended zone of influence for a segmented bus duct fire is as follows:

• Assume that the effects of the bus duct fault will be manifested at a transition point (the fault point). Recall that failures at end point terminations are captured under the end point equipment.

• Assume that molten metal material will be ejected from the bottom of the bus duct below the fault point and will spread downward encompassing the shape and volume of a right circular cone whose sides are at an angle of 15° from the vertical axis (a total enclosed solid angle of 30°).

• Assume that molten metal material will also be ejected outward and will spread within a sphere of 1.5 foot radius from the fault point.

• Assume that any exposed combustible or flammable materials within this cone-shaped and spherical zone of influence will be ignited. Combustible/flammable materials will not be considered exposed if they are protected by a fire-rated raceway wrap, conduit, or solid steel panels. Specific examples of the recommended treatment of exposed versus non-exposed materials are as follows:

o The solid metal side panels of a cabinet will prevent ignition of the combustible/flammable materials inside the cabinet.

o For cabinets with a solid steel top where all cable or conduit penetrations are sealed (e.g., consistent with the guidance provided in NUREG/CR-6850, EPRI TR 1011989, Chapter 11, with respect to the propagation of fires out of an electrical panel), molten material deposited on top of the cabinet will not burn through the panel top3.

o For cabinets with a ventilated top or unsealed cable or conduit penetrations, molten material deposited on top of the panel will penetrate into the panel and ignite the contents.

o Open ventilation sections on cabinet side panels that are made up of an open mesh or screen section will allow the penetration of molten material into the cabinet if the openings are within the zone of influence.

o For cabinet side panels or doors that include louvered ventilation openings where the louvers point downwards to the outside of the panel, molten material deposited on the surface of such panels will not penetrate into the cabinet.

o Cables in open-top cable trays will be ignited if they are within the zone of influence.

o Cables in conduit will not be ignited by molten materials deposited on the outer conduit surface if the open ends of the conduit are located outside the zone of influence.

o Cables in trays that are equipped with un-ventilated steel covers will not be ignited by molten metals falling from above (see footnote 2).

3 Note that, relative to this particular point of guidance, it is the judgment of the authors that even the minimum thickness of a typical steel cabinet top panel as employed in practice by manufacturers will be sufficient to prevent burn-through of the molten material ejected from a bus duct. The guidance specifically excludes credit for aluminum panels.

DRAFT 7 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT o Cables in trays that are equipped with aluminum covers of any kind, or with ventilated steel covers, will be ignited by molten metals falling from above.

• Damage and ignition within the initial zone of influence occurs at time zero (concurrent with the initial fault).

• Subsequent analysis of fire development, fire detection, and fire suppression response follow the same practices as applied to high energy arc faults for switchgear and load centers. In particular, the manual fire brigade response curve applicable to high energy arc faults for switchgear also applies to bus duct faults.

Figure 1: Photograph of the point on the bus duct where the arcing fault at Diablo Canyon was manifested (the fault point). Note that the tops of the cabinets to the left and right (the cabinets to the left are shown in figure 2) are visible in the lower corners of this photograph.

DRAFT 8 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT Figure 2: Photograph of the surface damage and burning observed for the more heavily impacted cabinets on one side of the aisle-way below the fault point (the cabinets to the left as seen in Figure 1).

Zone of influence for iso-phase duct fires For the iso-phase bus duct fires, it is recommended that the zone of influence should assume damage to any component or cable that would normally be considered vulnerable to fire damage (i.e., excluding items such as water-filled piping that would not normally be considered vulnerable to fire damage) located within a sphere centered on the fault point and measuring 5 feet in radius. Any flammable or combustible material within this same zone of influence should be assumed to ignite. The recommended zone of influence is intended to cover both the initial fault effect and the potential burning of hydrogen gas4 that may be released at low pressure from the bus casing upon rupture.

Also, note that the voltage/current of an iso-phase bus is higher than for a segmented bus.

An enduring fire (i.e., lasting beyond the initial fault) should be assumed consistent with the nature of any flammable or combustible materials present within the zone of influence and potential fire spread beyond the zone of influence.

For the case of fires occurring as the main transformer termination points, the potential for involvement of the main transformer (and its oil) should be considered. In particular, the electrical lines will each penetrate the casing of the transformer and this could allow the fire to spread to the transformer itself. Failure of the electrical penetration seals (e.g.,

4 Iso-phase bus ducts are generally filled with hydrogen gas at low pressure to enhance both cooling and electrical isolation. Upon rupture the hydrogen gas will leak from the duct, but neither a jet fire nor an explosion are anticipated due to the initial rupture.

DRAFT 9 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT melting of a rubber boot) could also create a path for oil leakage outside the transformer as was observed in the Vermont Yankee event.

The analysis should also consider the potential for involvement of additional hydrogen gas beyond that which will leak from the casing as a result of the initial fault. That is, the configuration of, and potential failure in, the hydrogen purge/fill system should be evaluated to determine if additional leakage of hydrogen gas is plausible. This assessment will require consideration of case-specific storage, piping and valve arrangements.

DRAFT 10 DRAFT

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT FAQ Number 07-0035 FAQ Revision 0 FAQ Title Bus Duct Counting Guidance for High Energy Arching Faults Plant: Harris Date: June 1, 2007

Contact:

Dave Miskiewicz Phone: 919.546.7588 Email: David.Miskiewicz@pgnmail.com Distribution: (NEI Internal Use) 805 TF FPWG FRATF RIRWG BWROG PWROG Purpose of FAQ:

Clarification/enhancement of Ignition Source counting guidance for High Energy Arcing Faults (HEAF) in NUREG/CR-6850, supporting NFPA-805 Fire PRA application.

Is this Interpretation of guidance? Yes / No Proposed new guidance not in NEI 04-02? Yes / No Details:

NEI 04-02 guidance needing interpretation (include section, paragraph, and line numbers as applicable):

New attachment on interpretation issues Circumstances requiring guidance interpretation or new guidance:

Pilot discussions and benchmarking of NUREG/CR-6850 for Task 6, Fire Ignition Frequency, has shown inconsistency in the treatment of High Energy Arcing Faults (Bin 16).

There is a need to resolve these issues to prevent future rework and to reduce burden associated with uncertainty treatment. This topic has impact on the NFPA-805 pilots, non-pilots and other users of NUREG/CR-6850.

The guidance provided in NUREG/CR-6850 for Task 6, Fire Ignition Frequency (Section 6.5.6, Bin 16), states:

Bin 16 - High-Energy Arcing Faults (Plant-Wide Components): High-energy arcing faults are associated with switchgear and load centers. Switchyard transformers and isolation phase buses are not part of this bin. For this bin, similar to electrical cabinets, the vertical segments of the switchgear and load centers should be counted. Additionally, to cover potential explosive failure of oil filled transformers (those transformers that are associated with 4.16 or 6.9kV switchgear and lower voltage load centers) may be included in vertical segment counts of the switchgear.

DRAFT Page 1 of 2 FAQ 07-0035 DRAFT Rev. 0

DRAFT NRC Staff Comments on FAQ 07-0035, Revision 0 [RHG] {12/20/2007} DRAFT FAQ Number 07-0035 FAQ Revision 0 FAQ Title Bus Duct Counting Guidance for High Energy Arching Faults The current guidance is silent regarding the treatment of bus duct. Preliminary discussions between the user community and the NUREG authors indicate that some specific guidance is needed to assure more consistent treatment of bus duct.

Detail contentious points if licensee and NRC have not reached consensus on the facts and circumstances:

Potentially relevant existing FAQ numbers:

This guidance is specific to the characterization of bus duct for Bin 16 HEAF determination.

The characterization and counting of electrical cabinets for Bin 16 determination is addressed by FAQ 06-0017.

Response Section:

Proposed resolution of FAQ and the basis for the proposal:

Because bus duct terminates at electrical cabinets, the HEAF counted for the electrical cabinet would also include those bus duct events and no further counting is necessary.

Basis:

The response is consistent with the guidance currently provided in NUREG/CR-6850.

Without additional guidance provided by the authors of NUREG/CR-6850, there is no basis for when or how to count bus duct.

If appropriate, provide proposed rewording of guidance for inclusion in the next Revision:

DRAFT Page 2 of 2 FAQ 07-0035 DRAFT Rev. 0