ML20199B234

Forwards Final Rept for C-310 Power Losses Which Resulted in Disabling of High Voltage UF6 Release Detection Sys. Supplemental Info Designated by Rev Bar in Right Margin of Rept.Attachment 2 List Commitments Made in Rept

ML20199B234
Person / Time
Site:	Paducah Gaseous Diffusion Plant
Issue date:	01/16/1998
From:	Polston S UNITED STATES ENRICHMENT CORP. (USEC)
To:	NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
GDP-98-1003, NUDOCS 9801280210
Download: ML20199B234 (7)

Text

_ _ _ _ _ _

USEC A Global Enwgy company January 16,1998 2

United States Nuclear Regulatory Commission SERI Al.: GDP 981003 Attention: Document Control Desk Washington, D.C. 20555 0001 Paducah Gaseous Diffusion Plant (PGDP)- Docket No. 70-7001 - Event Report ER-97-18, Rev.1 l

Pursuant to 10 CFR 76.120(d)(2), attached is the final report for the C-310 powe: losses which resulted l

In the disabling of the high voltage UF. releass detection system. This report provides supplemental l

information for Event Report 97-18 which was issued to the Nucicar Regulatory Commission (NRC) on November 14,1997. The supplemental information is designated by a revision bar in the right margin of the report. Attachment 2 is a list of commitments made in this report.

Should you require further information on this subject, please contact Bill Sykes at (502) 441-6796.

Sincerely, I

l Ik Steve olston General Manager Paducah Gaseous Diffusion Plant I

}

SP:WES:.INil:mel Attachments

}A cc:

NRC Region ill NRC Senior Resident inspector, PGDP

\\

l 9001280210 980116

'~

PDR ADOCK 07007001 C

PDR P.O. Box 1410, Paduuh, KY 4200t Telephone 502-441-5803 Fax 502-44!-5801 http://www.usec.com Offices in Livermore, CA Paducah, KY ltrismouth, OH Washington, DC

Docket,No. 70 7001 Page1of5 Event Report ER 97-18 Rev. I DESCRIPTION OF EVENT On October 17,1997, at approximately 2315, the C-310 East Normetex product withdrawal pump tripped during the startup of the pump. A subsequent lamp test indicated that power had been lost to the alarm annunciator panel in the Area Control Room (ACR). This panel includes the alarm for the high voltage process gas leak detection (PGLD) system. The heads for the C-310A hi@ voltage PGLD system were fired and no alarm was received in the ACR which confirmed that the high voltage PGLD system had been disabled. This safety system provides UF. detection above the UF condensers, accumulators, and heated housings in C-310A and 6

when activated will sound alarms in the ACR. 'Ihis safety system is required by technical safety requirement (TSR) limiting conditions for operation (LCO) 2.3.4.4 to be operable when the afTected process equipment is in Mode 2 or 3 and UF, pressure is above atmospheric pressure.

Immediately foW xing the power loss, the PGLD safety system was declared inoper*Sle, and a smoke watch was established as requi ed by LCO 2.3.4.4 action step A. The DC power was manually thrown over to the battery room by the building operators; however, this action did not restore the alarm power, as expected. At approximately 2340, the breaker supplying DC power to the No. I annunciator cabinet was found to be tripped. The breaker was reset and power was restored. This restored the alarm function of the PGLD system. Alarm panel capability was lost for approximately 25 minutes. The PGLD heads were tested and alarms were received, as required. Following the restoration of power, DC alarm power was switched from the battery room back to " normal" power by closing a breakt.r to tne rectifier.

On October 18,1997, at approximately 1400, the C-310 East Normetex pump tripped while running on-stream. It was determined that alarm power related to the high voltage PGLD system had been lost again. The breaker in the No. I annunciator cabinet was reset again at approximately 1402 which restored power to the alanns on the PGLD system. Alarm capability was lost for approximately 2 minutes. While resetting the breaker in the annunciator cabinet, operations personnel observed that the cabinet was excessively hot and found that the cabinet ventilation fans were not running. Upon checking the fans, a fan breaker was found to be tripped. The fans were restarted at approximately 1442 by resetting the fan bmaker.

Both instances of power loss to the alarms of the PGLD system w e reported to the plant shift superintendent (PSS) and documented on problem reports i

+i

< manner, but were initially

. incorrectly determined to be non-reportable. The losses of ala.

+r on the PGLD system which occurred on October 17 and October 18,1997, were repo.

NRC on October 21 at 0341 and 1238, respectively. By 10 CFR 76.120(c), verbal notification should be made to NRC within 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of discovery of disabled equipment which is required by TSR to be available and operable. (Reference NRC Event Notification Worksheet No. 33124 and No. 33126.)

Docket,No. 70 7001 Attechment 1 Page 2 0f 5 Subsequent to these notifications, it was discovered that the event notification worksheets incorrectly stated the " event times and dates." These worksheets were correct 4 and faxed to the NRC llQ Operations Duty Oflicer. Additionally, the description sedion of Notification Worksheet No. 33124 indicated that the loss of power affected the PGLD for the West Normetex pump under LCO 2.3.4.3. The correct description should have been that the loss of power affected the high vohage PGLD under LCO 2.3.4.4. Problem reports were issued to evaluate these concems.

The breaker which tripped was located in the C-310 No. I alarm annunciator cabinet. This cabinet provides alarm contact relaying for over one hundred alarm points in the C-310 building. The cabinet is located in the C-310 basement adjacent to the No. 2 annunciator cabinet and includes a free standing cabinet, terminal strips, alarm annunciator racks, ventilation fans, and a 10-amp two-pole DC breaker. The original design of the cabinet had nine alarm annunciator racks mounted in the cabinet. Each one of the racks was fed 125 VDC from the breaker mounted in the cabinet. At a later date, six additional units were mounted on the wall behind the cabinet and connected to the nine original units. The original racks each contained 12 alarm points, and the six newer units each contained six points. This gives the cabinet a total alarm point count of 144 alatm points. Historically, problems with the alarm annunciators have been failed relays, resistors, and bad connections. Given the large number of alarm points in each cabinet, failure of a component alarm point is not unexpected. Since an alarm point component failure could, on occasion, trip the breaker, further testing will be done to determine potential failure points in the No, I annunciator cabinet. This testing will consist ofinfrared thennography and breaker current measurement.

Following the problems on October 17 and October 18, Electrical Maintenance (EM) was called in to troubleshoot the system and determine the cause of the main breaker trip. Since there was a concern that the tripping of the East Normetex pump was associated with or caused the tripping of the No. I annunciator cabinet main power breaker, testing was completed to simulate the increase in current caused by the general alarm that came in as a result of the Normetex pump trip. It was first su:.pected that the effect of this Normetex alarm coming in exceeded the rating of the main breaker and caused the trip. EM attempted to recreate the alarm condition to see if the breaker would trip again. The breaker did not trip and ammeters on the load side of the bresker lugs indicated a peak load of 6.7 amps, well under the breaker rating of 10 amps. EM went on to simulate some alarm points that had recently been disconnected because they we.e no longer required far the East Normetex pump. Again the peak load was well below the breaker's rating.

The possibility of high cabinet te nperature due to failed ventilation fans affecting the breaker trip point was also investigated. The data sheet from the breaker manufacturer indicates that this breaker utilizes a hydraulic-magnetic solenoid to trip the breaker. This type of current sensing provides precision over-current protection without regard to thermal influences. Field.sts performed on a replacement breaker identical to the breaker feeding the annunciator cabinet

DocketNo. 70,-7001 Page 3 of 5 yielded trip values within the stated ranges. At this time, EM has not found the direct cause of the breaker trip. The No. I alarm annunciator cabinet continues to function normally. The breaker which tripped will be removed for further testing At this time, the cause of the breaker trip is not known. To verify operability, a surveillance is performed which test fires the high voltage UF. detection system twice per shift when the equipment it serves is in mode 2 or 3 and the UF. pressure is above atmospheric pressure. This surveillance is required by TSR SR 2.3.4.4 1. Additionally, a lamp test is perfonned once per shift on the annunciator panel where the high voltage UF alarm is displayed to ensure the alarm displays are functional. The annunciator panel is located in the ACR which is staffed by operators on a continuous basis.

Since the cause of the breaker trip is not known at this time, a supplemental report will be issued when troubleshooting is complete and the root cause and associated corrective actions have been determined. The target date for the supplemental tr. port is January 19,1993.

. The Heinemann Model 2263 molded-case circuit breaker, device ACB-1, was removed from l

l annunciator cabinet No. I and was field tested for tripping characteristics (Completed Actions l

No. I and 4). The breaker performed under tes ting in accardance with the manufacturer's l

specifications. The replacement breaker which has the same continuous current rating of 10 l

amperes has a longer over-current time delay (medium delay). This is preferred to accommodate l

the high in-rush load for the numerous electromechanical relays and alarm lamps. Configuration l

control of this breaker will be established which will ensure future aplacements will require the i

longer over-current time delay feature (Corrective Action No.1). Troubleshooting included the l

use ofinfrared thermography to determine if there were any potentially overheating components I

within the No I annunciater cabinet; none were identified. Current measurements were taken I

while terminal strips were jumpered to simulate activated alarms and during the performance of l

the lamp test (Completed Action No. 2). The current measurements revealed that the bmaker is l

overloaded during the lamp test (in excess of 13 amps for a 10-amp breaker). During I

troubleshooting and lamp testing, the subject breaker tripped four times.

l l

1 CAUS!TOF EVENT l

A.

Direct Cause l

l The direct cause of this event is overloading on the breaker in the No. I annunciator l

cabinet. The troubleshooting did not conclusively define the source of the overload for this l

particular event, but did show that the breaker is overloaded during the lamp test. To l

reduce the load on the breaker, lamps from annunciatorlights that are not required were l

removed. Forty-three 4-watt bulbs were removed from non-essential slaved alarm lights i

located in the C-331 area control room. This reduced the lamp load on the No.1 l

1 1

Docket.No. 70,-7001 Page 4 of 5 annunciator cabinet breaker by 3 amps or nr proximately 25 percent (Completed Action l

No. 3). The loud on No. I annunciator cabi iet will be further reduad to a value below 9 J

amps by replacing the lamps in the alarms with lower wattage bulbs (Corrective Action l

No. 2).

l l

Corrective Actions No. 3 and 4 will relocate the high voltage UF detection alam. from the l

6 DC alarm annunciator system to the AC alarm annunciator system,"Ronan." Relocation to l

the "Ronan system will provide a configuratbn with other safety-related alarms and with l

less vulnerability.

I l

A review of problem reports for the past 24 months concetning breaker tripping due to l

overload revealed no safety-related components reported with fanure conditions similar to I

this : vent, except for the subject breaker. PGDP has approximately 1100 safety-related I

circuit breakers.

l l

l 1

B.

Root Cause l

I The soot cause of this event is that there were inadequate design controls in place at the l

time that circuits were added to the No. I annunciator cabinet. With a TSR-required alarm, l

like the high voltage PGLD alarm, in this cabinet, the design of the cabinet should be such i

that the potential loads would r.at exceed the breaker rating. Since the time when circuits I

were added to the No. I annunciator cabinet, many engineering controls have been added I

which would prevent this problem if the circuits had been installed under the current design l

control procedures. This includes a more stringent modification design control program l

and the development of both a configuration management program and a document control l

program. These controls are documented under the following procedures:

l UE2-TO-EG1031," Modification Design Control"; UE2-EG-CF1032," Configuration l

Management Program"; and UE2-TO-RM1031," Document Control." These controls l

address the root cause of this event.

I l

i CORRECTIVE ACTIONS A.

Completed Actions l

l 1.

On December 31,1997, the b:eaker in the No. I annunciator cabinet which tripped l

was replaced.

I I

2.

On December 31,1997, thermography testing was completed and cur ent I

measurements were taken on the No. I annunciator cabinet.

i I

- Docket.No; 70,7001

- Attachment 1 Page 5 of 5 -

3. __ On December 31,1997, the lighting load on the No I annunciator cabinet was l

reduced by approximately 25 percent to the continuous current rating of the breaker.

l l

4.

On January 6,1998, field testing was completed on the breaker which had been l.

removed.L.

l

-l S..

- On January 8,1998, ESO NO. Z99700_was issued to mquest a design to relocate the

-l' high voltage UF. detection alarm from the No. I annunciator cabinet. _

~

-l l

l

. B.

Cormcdve Actions To Be Taken l

4 l

'l 1..

By February 27,'1998, Configuration Management will place ACB-1 breaker on the -

l

-l "Q" list.

-l 1

2.-

By ApW129,1998, Operations will fmther reduce the load on the No, I annunciator-

l

- 4

cabinet to below 9 amps by replacing alarm lamps with lower _ wattage bulbs Signs l

v

will be installed at C-310 ACR panels containing alarm lights for DC annunciator to {

zl-

= require use of maximum 3 watt bulbs.

l g.-

l3 : : By September 12,1998, Engineering will develop and " certify for construction" a

-l.-

design to relocate the high voltage UF. detection alarm from the DC alarm system to.

-l o

. the AC alarm annunciator system' "Ronan."

.l 4.

By January 28,1999, Operations will implement the design and place in service the -

l:

_ high Voltage UF, detection alarm on the "Ronan" annunciator system.

l l-l' EXTENT OF EXPOSURE OF INDIVIDUALS TO RADIATION OR TO RADIOACTIVE MATERIAM

+

No exposure to radiation or radioactive materials occurred during the outage of the high voltage l

o

- PGLD system.

I g

_ EESSONS LEARNED

{f Effective design / configuration control programs are critical for maintaining the reliability of i

safety systems, i

4 4

n+ss w

+

w

-e,-

Docket.No. 70.-7001 JAttachment 2

- Page 1 of1 List of Commitments l

-- =

-l 1.

By February 27,1998, Configuration Management will place ACB 1 breaker on the "Q" list.

1

2. -

By April 29,1998, Opentions will further reduce the load on the No, I annunciator l

cabinet to below 9 amps by replacing idarm lamps with lower wattage bulbs. Signs will l

be installed at C-310 ACR panels containing alarm lights for DC annunciator to require l

use of maximum 3 watt bulbs.

l 1-i 3.

By September 12,1998, Engmeenng will develop and " certify for construction" a design l

to relocate the high voltage UF. detection alarm from the DC alarm system to the AC l

l alarm annunciato system,"Ronan."

l l

4.

By January 28,1999 Operations will implement the design and place in service the high l

voltage UF. detection alarm on th;"Ronan" amunciator system '

l-I f

l l

_