Information Notice No. 91-43: Recent Incidents Involving Rapid Increases in Primary-to-Secondary Leak Rate

                             UNITED STATES 
                      NUCLEAR REGULATORY COMMISSION
                  OFFICE OF NUCLEAR REACTOR REGULATION
                         WASHINGTON, D.C.  20555

                              July 5, 1991


Information Notice No. 91-43:  RECENT INCIDENTS INVOLVING RAPID 
                                   INCREASES IN PRIMARY-TO-SECONDARY LEAK 
                                   RATE


Addressees:

All holders of operating licenses or construction permits for 
pressurized-water reactors (PWRs).

Purpose: 

This information notice is intended to inform addressees of recent 
incidents involving very rapid increases in the primary-to-secondary leak 
rate.  One of these incidents was followed by a steam generator tube 
rupture (SGTR).  The  leakage during these incidents increased at rates 
that were significantly higher than would be predicted on the basis of 
Figure 1 of Bulletin 88-02, "Rapidly Propagating Fatigue Cracks in Steam 
Generator Tubes."  It is expected that recipients will review the 
information for applicability to their facilities and consider actions, 
as appropriate, to minimize the probability of SGTR events.  However, 
suggestions contained in this information notice do not constitute NRC 
requirements; therefore, no specific action or written response is 
required.

Description of Circumstances: 

Mihama Unit 2 (Japan)

Mihama Unit 2 is a 19-year old PWR built by Mitsubishi and based on a 
Westinghouse Electric Corporation two-loop design.  The Mihama Unit 2 
steam generators (SGs) are based on the Westinghouse Model 44 design.  At 
12:24 hours on February 9, 1991, plant personnel received an "attention" 
signal from the SG blowdown monitor (R-19).  The attention signal 
setpoint was at 60 counts per minute (cpm), compared to the normal 
reading of 35 cpm.  At 12:33 hours plant personnel received an attention 
signal from the air ejector monitor (R-15).  The attention signal 
setpoint for the R-15 was at 900 cpm, compared to a normal reading of 800 
cpm.  At 13:00 hours, plant personnel sampled the blowdown from SGs A and 
B.  Results were obtained at 13:20 hours indicating a radioactivity 
concentration only slightly higher than normal in SG A, and no detectable 
concentration in SG B.

At 13:40 hours, an R-15 "counting rate alarm" (alarm setpoint: 2000 cpm) 
was sounded.  At 13:45 hours, the R-19 counting rate alarm (alarm 
setpoint: 400 cpm) was sounded.  At this time, plant personnel manually 
started a third charging pump because of decreased pressure and water 
level in the pressurizer.  At 13:48 hours, personnel began to manually 
reduce reactor power at a rate of 4.2 percent per minute.  At 13:50 
hours, the R-15 "counting rate high" alarm (alarm 

9106280018 
.

                                                          IN 91-43
                                                          July 5, 1991
                                                          Page 2 of 4


setpoint: 1 X 106 cpm) was sounded, followed by reactor trip on "low 
pressurizer water level," turbine trip, generator trip, and actuation of 
safety injection on low pressure and low water level in the pressurizer.  
Leakage from the primary to the secondary was essentially terminated at 
14:48 hours.  Plant personnel brought the plant to cold shutdown at 02:30 
hours on February 10, 1991.

Following this SGTR event*, the utility investigated the rupture and 
found that it was a complete circumferential failure of tube R14C45 in SG 
A, at the upper-most support plate.  The utility found that the failure 
mechanism was high cycle fatigue caused by fluid-elastic vibration.  By 
design, all tubes in rows 11 and greater are supposed to be supported by 
anti-vibration bars (AVBs).  However, the subject tube was not found to 
be so supported because of a reported "incorrect insertion" of the 
adjacent AVBs.

Maine Yankee Atomic Power Station 

Maine Yankee is a PWR designed by Combustion Engineering, Incorporated, 
and was licensed in 1973.  Between 14:00 hours on December 12, 1990, and 
00:57 hours on December 17, 1990, the rate of primary-to-secondary 
leakage gradually increased from 0.0006 gallons per minute (gpm) to 0.008 
gpm as determined from grab samples from the condenser air ejector.  
During this period, the licensee took grab samples at approximately 
4-hour intervals.  The licensee analyzed a grab sample taken at 02:34 
hours on December 17, 1990, and found that leakage in SG 1 had jumped to 
0.017 gpm, with a corresponding reading of 75,000 cpm on the air ejector 
radiation monitor.  At 03:40 hours, the licensee began reducing power at 
a rate of 5 percent per hour.  At 04:50 hours, the radiation monitor 
reading increased from 75,000 to over 400,000 cpm in less than 1 minute.  
Using this and the previous leak rate, the licensee quickly estimated a 
leak rate of 0.11 gpm.  This estimate exceeded the licensee's 
administrative limit of 0.07 gpm, and the licensee increased the rate of 
power reduction to 50 percent per hour.  The licensee analyzed a grab 
sample taken at 05:21 hours and confirmed a leakage rate of 0.105 gpm.  
At 06:07 hours, the reading from the air ejector radiation monitor jumped 
to 600,000 cpm.  Based on a grab sample taken at 06:36 hours, the 
calculated leak rate was 1.4 gpm, which exceeded the technical 
specification leak rate limit of 0.15 gpm.  The plant reached hot standby 
status at 06:53 hours and cold shutdown status at 03:10 hours on December 
18, 1990.

Subsequent investigation established the source of the leak to be a 
4-inch long axial crack at the apex of the row 6, line 43 U-bend.  The 
licensee described this location as a "steam blanketed region" where the 
batwing supports restrict flow permitting a steam void to form and 
contaminants to be deposited on the tube surface.  On the basis of the 
size of the crack found, the staff believes that the early identification 
and response to the rapidly increasing leak rate was the key factor in 
averting an SGTR event before plant shutdown. 

                    
*An SGTR event is defined in this information notice as a 
 primary-to-secondary leak exceeding the normal charging pump capacity of 
 the primary system. 
.

                                                          IN 91-43
                                                          July 5, 1991
                                                          Page 3 of 4


Three Mile Island Unit 1 (TMI-1) 

Three Mile Island Unit 1 is a PWR designed by Babcock and Wilcox (B&W) 
with once-through type steam generators and was licensed in 1974.  On 
March 6, 1990, TMI-1 was operating at 75 percent steady power, in a 
30-hour hold for restart physics testing.  The licensee was first alerted 
to the onset of primary-to-secondary leakage by an alarm from the 
condenser air ejector radiation monitors at 08:23 hours.  A post-incident 
review of the radiation monitor data indicates that the activity actually 
began to increase above its normal steady value at 08:01 hours.  At 
around 08:50 hours, the radiation monitor reading had increased from its 
initial value of 50 cpm to about 50,000 cpm, 5 times the alarm setpoint.  
At 09:00 hours, the licensee's preliminary estimates of leak rate were 
0.5 gpm based on mass balance estimates for the reactor coolant system 
(RCS) and between 0.5 and 0.75 gpm based on the decreased level in the 
make-up tank.  These estimates were below the plant technical 
specification limit of 1.0 gpm.  At 09:12 hours, a plant shutdown was 
commenced at the rate of 2 percent per minute, and the radiation monitor 
readings began to decrease.  The plant reached hot shutdown status at 
10:42 hours.  On March 7, 1990, the plant reached cold shutdown status by 
07:30 hours. 

The licensee later determined from activity measurements that leakage had 
reached 1.1-1.8 gpm before plant shutdown was commenced.  However, 
approximately 2.5-3 hours are needed to obtain results from this method, 
and, thus, this information was not available to the operators prior to 
the decision to commence plant shutdown.

After shutting down the plant, the licensee found the source of the leak 
to be a 360 degree circumferential crack in tube 1 of row 77 in the 
"lane-wedge" region at the lower face of the upper tubesheet.  The staff 
believes that the early identification and response to the rapidly 
increasing leak rate was key to preventing a much larger leak (if not an 
SGTR event) before plant shutdown.  Similar fatigue cracks in other B&W 
steam generators have caused larger leaks, but not SGTRs, because these 
cracks were confined within the tubesheets or support plates.

Discussion: 

Earlier incidents involving rapidly increasing primary-to-secondary leak 
rates were the subject of NRC Bulletin 88-02 and NRC Information Notice 
88-99, "Detection and Monitoring of Sudden and/or Rapidly Increasing 
Primary-to-Secondary Leakage."  Bulletin 88-02 was issued in response to 
the July 15, 1987, SGTR event at the North Anna Power Station Unit 1.  
The NRC staff requested, in item C.1 of Bulletin 88-02, that an enhanced 
primary-to-secondary leak rate monitoring program be implemented at 
certain PWRs (i.e., PWRs with Westinghouse steam generators, carbon steel 
support plates, and denting corrosion) until a potential fatigue problem 
at these plants could be resolved.  The staff requested this enhanced 
leak rate monitoring program to ensure that licensees could detect and 
respond to a rapidly increasing leak rate caused by high cycle fatigue 
before an SGTR occurs.  The effectiveness of this program was to be 
evaluated against the assumed time-dependent leak rate curve given in 
Figure 1 of the bulletin.  This curve was based on the estimated rate of 
increase in leakage before the SGTR 
.

                                                          IN 91-43
                                                          July 5, 1991
                                                          Page 4 of 4


event at North Anna Unit 1.  This curve yields an estimated 63 hours for 
leak rates to increase from 20 gallons per day (gpd) to 500 gpd.

During discussions with numerous industry representatives, the staff has 
found that Figure 1 of Bulletin 88-02 is widely used throughout the 
industry as a benchmark and/or performance measure for plant-specific 
leak rate monitoring programs.  This is true even at plants which were 
not subject to the actions requested by the bulletin. 

However, there have been several incidents since issuance of Bulletin 
88-02 where the rate of leakage increase occurred more rapidly than would 
be predicted on the basis of Figure 1 of the bulletin.  The leak rates at 
Mihama Unit 2, Maine Yankee, and Three Mile Island Unit 1 escalated from 
very low levels to more than 500 gpd over time periods ranging between 
one hour and six hours.  An earlier leakage incident at Indian Point Unit 
3, described in NRC Information Notice 88-99, developed over a similar 
time span.  Thus, these incidents are indicative of the limitations of 
Figure 1 of the bulletin as a benchmark and/or performance measure for 
plant-specific leak rate monitoring programs. 

Leak rate monitoring programs can provide for early detection and 
response to rapidly increasing leak rates and, thus, can be an effective 
approach for minimizing the frequency of steam generator tube ruptures.  
This can be achieved by having, as close as possible, real time 
information on leak rate and rate of increase of leak rate on which to 
act.  Data from the air ejector radiation monitors, for example, are 
displayed continuously in the control room and have been shown to provide 
a relatively good time response to rapidly increasing leakage.  Use of 
these data, in conjunction with appropriate alarm setpoints, can quickly 
alert the operators to a rapid increase in leak rate and the need for 
confirmatory leakage measurements and/or the need to shut down the plant.  

Nitrogen-16 (N-16) monitors on the steamlines are coming into increasing 
use in the U.S. industry as a supplemental method for monitoring 
primary-to-secondary leakage.  These monitors also exhibit good time 
response to changes in the leakage rate.  Data from the N-16 monitors can 
be continuously displayed in the control room directly in terms of 
leakage rate and can be alarmed.

No specific action or written response is required by this information 
notice.  If you have any questions about this matter, please contact the 
technical contact listed below or the appropriate NRR project manager. 




                            Charles E. Rossi 
                            Division of Operational Events Assessment
                            Office of Nuclear Reactor Regulation


Technical Contact:  E. Murphy, NRR 
                    (301) 492-0710

Attachment:  List of Recently Issued Information Notices
.