Information Notice No. 86-47: Erratic Behavior of Static "O" Ring Differential Pressure Switches

                                                            SSINS No.:  6835  
                                                            IN 86-47 

                                UNITED STATES
                            WASHINGTON, DC 20555

                                June 10, 1986

Information Notice No. 86-47:   ERRATIC BEHAVIOR OF STATIC "O" RING 
                                   DIFFERENTIAL PRESSURE SWITCHES 


All boiling water reactor (BWR) and pressurized water reactor (PWR) 
facilities holding an operating license (OL) or a construction permit (CP). 


This information notice is intended to advise licensees of erratic behavior 
of certain differential pressure switches supplied by SOR, Incorporated 
(formerly Static "O" Ring Pressure Switch Company) which apparently caused 
failure of the LaSalle 2 reactor to scram automatically when it was 
operating with water level below the low level setpoint. Similar switches 
are also installed in the high pressure core spray system and the residual 
heat removal system. 

It is expected that recipients will review this information for 
applicability to their reactor facilities and consider actions, if 
appropriate, to preclude the occurrence of a similar problem at their 
facility. Suggestions contained in this notice do not constitute NRC 
requirements. Therefore, no specific action or written response is required.

The NRC evaluation of this incident is continuing. If specific action is 
determined to be necessary, a separate notification will be issued. 

Summary of Circumstances 

On June 1, 1986, LaSalle 2 experienced a feedwater transient that resulted 
in a low reactor water level. One of the four low level trip channels 
actuated, resulting in a half scram. The operator recovered level and 
operation was continued. Subsequent reviews by licensee personnel raised 
concerns that the level had apparently gone below the scram setpoint and 
thus a malfunction of the reactor scram system may have occurred. Based on 
this concern, the licensee declared an "Alert" and shut the plant down. The 
NRC dispatched an augmented inspection team to the site. Subsequently, the 
licensee found that the "blind" switches which operate on differential 
pressure perform erratically. The licensee also found erratic operation for 
similar switches in the high pressure core spray system and the residual 
heat removal system which operate valves in the minimum flow recirculation 
lines. Based on these results, the licensee declared all emergency core 
cooling systems in LaSalle 1 and 2 to be inoperable. Both units are in cold 
shutdown pending further evaluation of the problem. 


                                                            IN 86-47 
                                                            June 10, 1986 
                                                            Page 2 of 4 

Description of Circumstances: 

The following description was constructed from a preliminary sequence  of 
events prepared by the augmented inspection team and from other input by the

At 4:20 A.M. on Sunday, June 1, 1986, LaSalle 2 was operating at 93 percent 
of full power. Both turbine-driven feedwater pumps were operating, with the 
"A" pump in manual control and the "B" pump in automatic control. The 
motor-driven feedwater pump was in standby. While a surveillance test was 
being conducted on feedwater pump "A", the turbine governor valve opened 
further and caused pump speed and reactor water level to start increasing. 
At about the same time, the automatic control systems for both 
turbine-driven pumps locked out. The reactor operator regained control of 
feedwater pump "A" and ranback feedwater pump speed in an attempt to restore 
water level to the nominal value (36 inches on the narrow range recorder). A 
few seconds later when the control system was reset, the "B" feedwater pump 
controller, automatically ranback the pump speed to zero for no apparent 
reason. Reactor water level started falling at about 2 inches/second. 

Subsequently, the reactor protection system responded via separate level 
switches to the falling reactor water level by reducing recirculation flow 
to reduce power, and the operator started the motor-driven feedwater pump to
increase level. The level continued to fall for a few more seconds before 
turning around. The minimum reactor scram setpoint required in the technical
specification is 11 inches. The level channels are formally set to trip at 
13.5 inches, and the operators are trained to expect reactor scram by the 
time that the water level reaches 12.5 inches. As the level was falling, one
of the four reactor scram level switches (the "D" switch) tripped at 
approximately 10 inches, causing a "half scram." As designed, this did not 
initiate control rod motion. None of the other three level switches tripped 
during this transient. No reactor scram occurred during this transient, 
either automatically or manually. 

In the BWR scram system logic, which is one-out-of-two-taken-twice, at least
one instrument channel in each scram system must trip to generate a scram 
demand signal and thereby initiate control rod motion.  Preliminary results 
of the investigation indicate that the reactor water level fell to a minimum
value of about 4.5 inches on the narrow range instrumentation, which is 
several inches below the specified scram setpoint but still 13 to 14 feet 
above the top of reactor fuel.  The period that the water level was below 
the specified scram setpoint value was approximately 2 seconds.  After 
feedwater flow turned the transient around, the plant stabilized at a power 
level of about 45 percent. The "B" scram system half scram was manually 
reset about 30 seconds later. The power level was increased to 60 percent 
about 3 hours later. 

Shortly after the subsequent shift change, the oncoming shift engineer's 
review was effective in indicating that the reactor water level appeared to 
have fallen below the scram setpoint and the level switches may not have 
performed properly. He then requested that an instrumentation technician 
check the calibration of the switches. The results were that the "A" and "C"
switches, which are in the "A" scram system, tripped at 10 and 13.5 inches 
respectively during the calibration check; the "B" and "D" switches, which 
are in the "B" scram system, tripped at 11 and 13.5 inches respectively. The
switches were readjusted to  

                                                            IN 86-47 
                                                            June 10, 1986 
                                                            Page 3 of 4 

trip at 13.5 inches. Based on these results, the operating staff believed 
that a malfunction of the scram system may have occurred. An orderly 
shutdown of the plant was initiated at 2:00 P.M. (CDT). At 2:30 P.M., the 
resident inspector was notified, and at 5:30 P.M., the NRC Operations Center 
was called via the emergency notification system and informed of this event 
by the licensee. 

At 6:20 P.M., the licensee decided that the "A" scram system had failed to 
perform during the transient. The "A" scram system was manually tripped 
providing a half scram on the side that had apparently malfunctioned. The 
orderly shutdown was continued, and an "Alert" was declared. When all the 
control rods had been fully inserted at 9:22 the next morning, the Alert was

On Monday, June 2, the NRC determined that the incident warranted a thorough
investigation. The NRC Regional Administrator dispatched an augmented 
inspection team to the plant site. 

On Monday evening, June 2, the licensee checked the calibration of the 
reactor scram water level switches by varying the actual level in the 
vessel. The results were that the "A" and "C" switches tripped at indicated 
levels of 9.0 and 6.9 inches respectively and the "B" and "D" switches 
tripped at 3.9 and 10.2 inches respectively. These data were obtained about 
30 hours after the switches had been calibrated according to plant 
procedures and suggest a non-trivial difference. Additional data obtained 
over the next two days by varying reactor water level demonstrated continued 
erratic behavior of switch setpoints. 

On Saturday, June 7, after calibrating the Static "O" Ring flow switch which
actuates the minimum flow recirculation valve in the high pressure core 
spray system, the licensee performed a different test using actual system 
flow. The switch actuated when flow was at 530 gpm instead of 1000 gpm where 
it had been set to actuate. The licensee found similar performance of flow 
switches in the residual heat removal system. The licensee now suspects all 
Static "O" Ring differential pressure switches and has declared all 
emergency core cooling systems in both units to be inoperable. Both units 
remain in cold shutdown. 


It appears at present that the water level decreased below the scram 
setpoint for about two seconds and reached a minimum level of about 4.5 
inches. This is based on a recording from the narrow range water level 
instrument and records from the startup testing data acquisition system 
which recorded levels from the same transmitter. Had the reactor operator 
been aware of this fact before the water level had increased to a level 
above the setpoint, the reactor operator would have been expected to scram 
the reactor manually. 

The differential pressure switches which provide the water level trip input 
to the reactor scram system were provided by SOR, Incorporated. These level 
switches are not original equipment; but were installed during replacement 
of equipment in secondary containment. Affected licensees had determined 
that the original switches were not qualified to operate in the environment 
created by an accident. Operation of the SOR switches has been demonstrated 
to be erratic with little correlation between the setpoints established 
during atmospheric pressure  

                                                            IN 86-47 
                                                            June 10, 1986 
                                                            Page 4 of 4 

calibrations and switch actuations under system pressure conditions. 
Exercising the switches by applying successive differential pressure cycles 
appears to mask erratic setpoint behavior. Similar problems with SOR 
differential pressure switches have been reported at Oyster Creek. 

Per plant procedure, the switches for reactor water level had been exercised
prior to calibration following failure of the reactor to scram 
automatically. For this reason, performance of the level switches may have 
been different during calibration than during the event. Further, none of 
the level switches in the LaSalle 2 reactor scram system operate in 
conjunction with individual level transmitters. Therefore, the calibration 
and performance of the individual low level trip channels cannot easily be 
compared to each other. In effect, the operator is blind to switch 

The vendor has indicated that those plants identified in Attachment 1 have 
similar differential pressure switches. This list of plants includes 
pressurized water reactors as well as boiling water reactors. NRC intends to
meet with representatives of General Electric Company, SOR Incorporated, and
interested licensees at 10 A.M. on Thursday, June 12, 1986, in Bethesda, 
Maryland to discuss experience with the switches. 

It is suggested that licensees consider advising their reactor operators of 
the LaSalle incident and providing guidance to them as to how to promptly 
detect the occurrence of a similar problem at their plants and the proper 
remedial action to be taken. 

No specific action or written response is required by this notice. If you 
have any questions regarding this matter, please contact the Regional 
Administrator of the appropriate regional office or this office. 

                                   Edward L. Jordan, Director 
                                   Division of Emergency Preparedness 
                                     and Engineering Response 
                                   Office of Inspection and Enforcement 

Technical Contacts: J. T. Beard, NRR 
                    (301) 492-4415 

                    Roger W. Woodruff, IE 
                    (301) 492-7207 

1.   Plants with Similar Differential Pressure Switches 
2.   List of Recently Issued IE Information Notices

                                                            Attachment 1  
                                                            IN 86-47  
                                                            June 10, 1986 


     PLANT                                   SOR MODEL NUMBER 

Penn. Pwr. & Light/Susquehanna                    103/B202 

So. Cal. Edison/San Onofre                        103/B903 

TVA/Brown's Ferry                                 103/B212 

TVA/Sequoyah                                      103/BB212 

WPPS                                              103/BB203 

GPU/Oyster Creek                                  103/B905  

N.E. Nuc./Millstone                               103/B903 

South Texas Projects                              103/BB212  

Commonwealth Edison/LaSalle                       103/B202  

Page Last Reviewed/Updated Friday, May 22, 2015