ML20064K923
| ML20064K923 | |
| Person / Time | |
|---|---|
| Site: | 05000000, Arkansas Nuclear |
| Issue date: | 11/03/1981 |
| From: | Ellen Brown, Lanning W NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | |
| Shared Package | |
| ML082180533 | List:
|
| References | |
| FOIA-82-261 NUDOCS 8111040084 | |
| Download: ML20064K923 (6) | |
Text
-
s.
a AE0D/E127 This.js anJaternal, pre-decisio,nal document not
,. necessarily representing a-
~
's position of AE00 or NRC.
~
PRESSURE BOUNDARY DEGRADATION t-
\\*
DUE TO PUMP SEAL FAILURE AT ARKANSAS NUCLEAR,0NE '
Wayne. D. Lanning'and Earl J. Brown Office for Analysis and Evaluation '
'.of Operational Data
~
U.'S. Nuclear _ Regulatory Commission
"~
~~
~
Washington, D.C.
20555 l:
.fQ This report was pr prre for presentation at the Ninth Meeting of the Committee on Safety of Nuclear Installations, Paris, France on November 3-4, 1981.
Alu
.ag g
PRESSURE BOUNDARY DEGRADATION DUE-TO
. PUMP SEAL EAILU.8E_AL ARKANSAS NUCLEAR ONE ABSTRACT Arkansas Nuclear One, Unit 1 (ANO-1) was operating at 86% of full power on May 10,1980, wh4n a reactor coolant pump shaft seal failed. The reactor coolant pressure boundary was breached and apprdximately 227,000 liters (60,000 gallons) of coolant was leaked to the containment. The leak rate varied from 0.32 to 19 liters /second (5-300 gallons / minute).
The reactor was rapidly shut down in an orderly manner and no abnormal offsite radiologicaI releases resulted.
The ANO-1 event was significant because it revealed that the failure of one seal stage. can lead.to 'a total loss of seal integrity..The leakage rate from the failed se'al was larger than previously, predicted, although well within the installed system capabilities for reactor coolant inventory
, recovery.
A subsequent analysis of various seal failure events has led to the preliminary determination that because of the frequency of. seal failures in operating reactors, the probability.of small break loss-of-coolant events is larger than previous est'imations.
4 l
1;
m f
8
.1. 0 DESCRIPTION OF THE EVENT Arkansas Nuclear.One, Unit 1 (ANO-1) is a pressurized water reactor located in Rus'sellville, Arkansas.,_ The2u.cl_ ear steam supply system was designed by the Babcock and Wilcox Company and licensed by the Nuclear Regulatory Commission to operate at a power level of 2565aMWt (850'MWe).
The reactor-coolant system (RCS) employs two recirculation coolant loops.
Each loop utilizes a single line (hot leg) to direct flow to a once-through steam generator and two lines (cold legs) with a reac. tor coolant pump (RCP) in each line to return coolant to the reactor pressure vessel.
The RCP was manufactured by Byron Jackson Pump Division, Borg-Warner Corporation.
Prior to the event on May 10, 1980, the unit was operating at approx-
~
imately 86% of full power without any major operational a'tivities in c
progress.
The first indication that an RCS pressure boundary degradation an'd inventory loss occurred was when the operators observed a rapid 'de-crease (a step change) in the RCS makeup tank level.
The reactor ~
coolant pump (RCP) seal instrumentation confirmed that a problem existed with the shaft seal or associated cooling water piping.
Since the leak rate exceeded the plant technical specifications, a power reduction was initiated in preparation for reactor shutdown. ~ The initial ~-
, rate of power reduction was 5% per minute, and increased to 20'to 30%
per. minute,. in response to the leak rate increasing from 0.32 to 1.3 liters /second (5-20 gallons / minute). Subsequently, the RCP was stopped; the leak rate then immediately increased to an estimated 16 to 19 liters /second (250-300 gallons / minute).
The operators started and stopped the RCP bearing lift pumps four times in succession and the leak rate decreased.
The reactor was manually tripped from approxi-mately 10% power.
The safety injection.sysiem was manually actuated
~
l (bmfare may Amtrematic initiation setpoint was reached) during the e' vent
~
m y -.
\\
to restore ' pressurizer leve1.
After the trip, containment pressure increased
~
from 101 to 105 kilopascal (14.7 to 15.2 psig). Approximately 227,000 liters 1/
(60,600 gallons) of water accumulate.d inside containment during the event.
~
More than 28*C (50*F) subcooling was maintained during the event.
4 During the controlled system depressurization, the operators decided to avoid discharging the core flood tanks to the RCS.
But the R'CS pressure decreased to below 4.2 megapascal (600 psig) and. some water from the core flood tanks entered the RCS before containment entry could.be made to isolate the tanks.
No nitrogen cover gas entered the RCS, however, from the core flood tanks.
At this plant, the electrical breakers for the isolation valves between the core flood tanks and the RCS are 16cated inside primary containment and were not operable froni outside containment at ti e time of the event.
l Approximately seven hours after initiation of the event, the cooldown was completed with the resi' dual heat removal system in service and all four reactor coolant pumps secured..
~
Examination of the RCP seal revealed that it had experienced catastrophic destruction which resulted in an unexpected high leak rate.
It is believed that the upper (third) stage assembly failed first and that damage to the. _,
other stages was a direct result of a stationary carbon ring failure in the
- - - - e upper stage.
The cause of the failure could not be positively determined.
The postulated causes are:
one, exce'ssive wear of the carbon ring may have caused it to break apart; or two, possible excessive axial movement or improper seatin,g of the seal cartridge, caused the carbon ring to fail in compression.
Either cause of failure can result in the loss of seal axial integrity and a loss-of-coolant accident (LOCA).
g
\\
F, e.
- 2. 0 EVALUATION OF THE OCCURRENCE 2.1 Reactor Coolant System Response The operators received Earl'y iiidication of RCP seal problems because operation
~-
personnel were obtaining RCS leak rate data at the time of the event and 4
e s
observed a step decrease in the makeup tank level. Further investigation revealed that RCP seal cavity pressure, temperature and flow were indicating abnormal conditions.
The operators implemented the procedures for a small break loss-of-coolant event and initiated a poNr redur$1on to achieve cold shutdown.
The reactor coolant system cooldown rate was ac'celera,ted to 41.7C(75'F)/ hour.
The technical specification limit is SS.6*C(100*F)/ hour.
The main turbine-generator was taken off line approximate 1'y 62 minutes after power reduction was initiated and the affected RCP was tripped one minute l ater.' The leak rate' incre,ased.immediately to 16-19 liters /second (250-300 gallons / minute) at this time, which exceeded the flow rate of the makeup pump.
After the RCP bearing lift pumps were started and stopped four times by the
~
operators, the leak rate decreased.
The safety injection system was manually initiated in response to the decreaTing RCS level and pressure due to the leak-and reactor coolant siirinkage (volume decrease) from cooldown following the.
Normal reactor coolant drainage flow (letdown) and RCP seal.,_
coolant return flow were isolated by the operators.
The increasing primary containment building. pressure and radiation levels confirmed that.the leak was located in the containment area.
The RCS responded as expected and all safety systems performed satisfactorily.
However, the inability to isolate the core flood tanks from outside primary containment was identified as a design deficiency although it had no adverse effect during this event.
As a result of the fast cooldown rate and depres-
, surization, some inventory from the tanks enfered the RCS before containment
.D..
entry and breaker closure could-be achieved.
Failure to isolete the core flood tanks.during an.RCS depressurization.below 4.2 megapascal (600 psig) could become a problem if nitrogen was introduced into the RCS after the tanks
~
empty.
The licensee has subsequently relocated *the core' flood tank isolation valve breakers outside of containment.
- 2. 2 Analysis of RCP Seal Failure The cartridge-type shaft seal consists of an upp'er, middle, and lower stage.
These three stages are cooled by s.eal injection coolant provided by the normally operating R,CS makeup pump and by the integral heat exchanger which is cooled by the component cooling water system.
The stages are in series and each stage is designed to be capable of withstanding RCS operating pressure such that a' single stage failure could be detected and appropriate operator act' ion coinpleted in a timely inanner without incident or consequential failure of the re~maining two stage's.
On examination of this failed seal, however, all
thiree stages were found t~o be. severely damaged. ' The up~per sta'ge ex~perienced the most damage. The stationary ca_r] hon ring had disintegrated; it appeared to l
have been ground into carbon particles and washed away.
It is believed that t
this carbon ring breakdown was the initial failure; the-loss of this ring probably resulted in the other two stages shifting upward causing subseqderit -,'
breakage of the carbon, ring in each of the other two stages.
The failure of the upper stage carbon ring was postulated to have occurred from either excessive wear or fatigue due to compression.
The mechanism or conditions leading to'the ultimate failure of the ring are not positively known.
The licensee has postulated that either excessive axial movement or improper seating of the seal cartridge lead to wear or failure by com-pression.
In general, RCP rotor vibration is a common cause of'RCP seal
- f
?.
-5 failures.
Prior.to the event, however, there were.no indications of unusua'l vibration or pending seal, degradation.
3.0 SAFETY SIGNIFICANCE OF RCP SEAL FAILURE
- r The reactor coolant pump seal provides for discrete pressure and temperature decreases from the high reactor coolant pressure and temperature conditions to near atmospheric conditions by means of controlled flow or leakage of coolant through the seal cartridge.
The principal safety issue is that the catastrophic failure of an RCP seal results in a loss of primari reactor pressure-boundary integrity, which leads to a small loss-of-coolant event and a challenge to the safety syste[ns.
Other areas of concern are the equivalent break size and the frequency of RCP seal leaks compared to. pre-
~
viously estimated sina11 break loss-of-coolant probabilities.
5ince aniquivalent break size for the event was not accurately known, the method for relating the consequences of the seal failure to previous LOCA analyses was through a comparison of leakage rates.
For example, in the 2/
Reactor Safety Study, the "s :ll-small" LOCA (S ) is defined as an RCS
~
2 rupture between 1.3 and S.1 c'entimeters (0.5 to 2 inches) equivalent diameter.
This break size opening corresponds to a -leak rate of 3 to 50 liter /second-(50-800 gallons / minute).
Therefore, the ANO-1 event is considered similar to the "small-small" LOCA category of the Reactor Safety Study since the average leak rate was not exceeded.
A review of RCP shaft seal failure events reported to the NRC was conducted to assess their frequency for comparison with the probbility estimated in the Reactor Safety Study.
The operational data on RCP seal leakage events
- Basis for repdrting to the NEA Incident Reporting System (IRS) as a significant event.
y
~6-for all RCP types were used.in determining the. event.. frequency and in esti-3/
mating the probability of small loss-of-coolant event.s.-~ The NRC study covered over 200 seal leakage events reported since 1967.
An initial conclusion of 4
the study indicated that the probability of sma}1 break 1,0CAs from RCP seal,
failures appears to be an order of magnitude higher than the "small-small" LOCA probability obt'ained in the Reactor Safety Study.
The RCP shaft seal configuration was designed bf Byron Jackson to limit leakage in the event of a seal failure.
However, the ANO-1 event seems to have exhibited leakage rates greater than usually observed or anticipated I'
after seal failure occurs.
The simultaneous failure.of all three stages was similarly an unexpected occurrence.
This raises a safety question relative to whether the design can reasonably be expected to limit the leak rate to a predetermined value.
Important factors affecting seal integrity such as normal wear, the number of RCP starts and stops, seal cooling, and operation with some seal degradation have not been quantified to assess their impact on the design basis of the seal.
Studies are in progress to assess the operational history of RCP seals and other factors in order to improve reliability of RCP s6als.
REFERENCES 1.
Licensee Event Report 80-015/01X-2, Arkansas Power and Light Company, Docket 50-313, April 13,1981.
2.
U.S. Nuclear Regulatory Commission, " Reactor Safe.ty Study - An Assessment
. of Accident Risks in U.S. Commercial Nuclear Power Plants," WASH-1400 (NUREG-75-014),Octob.er1975.
3.
Memorandum, T. E. Murley, NRC' to D. G. Eisenhut, NRC,
Subject:
Reactor Coolant Pump Seal Failure, dated March 27, 1981.
o y.,-
2-----
--