Information Notice No. 91-50: A Review of Water Hammer Events after 1985

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

                               August 20, 1991


Information Notice No. 91-50:  A REVIEW OF WATER HAMMER EVENTS 
                                   AFTER 1985 


Addressees:  

All holders of operating licenses or construction permits for nuclear power 
reactors. 

Purpose: 

This information notice is intended to alert addressees to a U.S. Nuclear 
Regulatory Commission (NRC) evaluation of water hammer events between 
January 1986 and March 1990.  It is expected that recipients will review the 
information for applicability to their facilities and consider actions, as 
appropriate, to avoid similar problems.  However, suggestions contained in 
this information notice do not constitute NRC requirements; therefore, no 
specific action or written response is required. 

Background: 

The NRC originally addressed water hammer in Unresolved Safety Issue (USI) 
A-1, reviewing 148 reported events from 1969 to 1980.  The NRC considered 
this USI resolved with the issuance of "Evaluation of Water Hammer 
Occurrence in Nuclear Power Plants," NUREG-0927, Revision 1, in March 1984.  
The NRC concluded that cost-benefit considerations did not support new 
requirements to reduce the number of water hammer events.  However, the NRC 
included guidelines on measures to prevent and reduce water hammer.  

After the event at the San Onofre Nuclear Generating Station, Unit 1, in 
November 1985, the NRC reassessed the occurrence of water hammer, reviewing 
40 events from 1981 to 1985.  In the reassessment, the NRC confirmed the 
original conclusions that new or additional requirements to reduce the 
number of water hammer events were not cost-effective.  The frequency of 
water hammer events had decreased significantly since the initial review.  
The NRC identified no new causal mechanisms for water hammer.  

Description of Circumstances: 

The NRC evaluated water hammer events that have occurred since January 1, 
1986.  The staff searched NRC databases from January 1986 through March 1990 
and found about a dozen reports of water hammer events or events related to 
the water hammer phenomenon.  In February 1991, the staff documented its 
findings in "A Review of Water Hammer Events After 1985," AEOD/E91-01.  A 
copy of this 

9108140097 
.

                                                            IN 91-50
                                                            August 20, 1991
                                                            Page 2 of 6


report is available in the NRC Public Document Room, 2120 L Street N.W., 
Washington, D.C.  The staff reviewed each of these reports to identify new 
physical phenomena, common mode aspects, and ways to prevent situations that 
could result in water hammer.  The following events at Dresden Units 2 and 
3, South Texas Unit 1, Trojan, and Susquehanna Unit 2 exhibit 
characteristics not emphasized in previous studies and offer lessons beyond 
implementing the guidance in NUREG-0927, Revision 1.  In addition, a recent 
event at Big Rock Point 1 involved damage to a gate valve.  Such damage has 
not been previously identified in a nuclear power plant.

Big Rock Point 1:  On May 28, 1991, while the reactor was operating at 97% 
of full power, the licensee was performing a routine surveillance test of 
the emergency core spray (ECS) injection valves.  After successfully testing 
both injection valves in the primary ECS system and the upstream injection 
valve in the backup ECS system, a signal to open the downstream injection 
valve was generated.  A few seconds after the valve started to open, a 
reactor operator at the valve heard a water hammer and observed movement of 
the 4-inch piping that continued for several seconds.  The water hammer bent 
one pipe hanger, partially pulled a bolt for another from the wall, and 
misaligned or damaged switches attached to the valve operators.

The backup ECS system is intended to deliver water from the fire suppression 
system at 150 psi to the reactor vessel after it has been depressurized 
following a loss of coolant accident.  Fire water would be delivered to a 
nozzle located within the reactor vessel.  The elevation of this nozzle is 
slightly higher than the elevation of the nozzles for the piping which 
connects the reactor vessel to the steam drum.  The length of backup ECS 
piping from where it penetrates the head of the reactor vessel to the 
nearest injection valve is approximately 60 feet.  This piping is routed 
horizontally and vertically upwards.  An 18-foot section of horizontal 
piping has a rise in excess of 4 inches in the downstream direction and thus 
acts as a water trap.  The upstream side of the injection valve nearest to 
the reactor vessel connects to a short pipe spool, a check valve, another 
short pipe spool, and the upstream injection valve.  A short length of 
1-inch pipe connects to the bottom of the upstream spool and terminates at a 
blind flange with an eighth-inch hole.  The pipe serves to demonstrate that 
the check valve is not leaking.

The backup ECS piping normally contains noncondensible gases and saturated 
steam and water at 1350 psi.  The relative concentrations of these 
constituents depend on the temperature distribution in the piping, the 
length of time that the reactor has been at power, and the number of 
surveillance tests that have been performed since the last startup.  On May 
28, 1991, conditions in the pipe led to acceleration of a slug of water that 
struck the gate of the downstream injection valve as it was opening during 
the surveillance test.  The impact was great enough to leave imprints of the 
valve seats on both sides of the gate and to cause some cracking of the 
gate.  Leakage of the valve prior to the event and consequent steam cutting 
of the upstream side of the gate and seat may have contributed to the event.  
The licensee's corrective actions included changing the slope of the 
horizontal section of piping, so that it will drain back to the reactor, and 
repairing the valves, valve operators, and pipe hangers.
.

                                                            IN 91-50
                                                            August 20, 1991
                                                            Page 3 of 6


Dresden Unit 2:  A series of events occurred in the high pressure coolant 
injection (HPCI) system at Dresden Units 2 and 3 over several months.  They 
are discussed in order of occurrence to enhance the understanding of changes 
in system valve positions.  This event is also discussed, but in less 
detail, in Information Notice No. 89-8O, "Potential for Water Hammer, 
Thermal Stratification, and Steam Binding in High-Pressure Coolant Injection 
Piping," December 1, 1989.  On October 31, 1989, while the plant was 
operating at 100 percent full power, the licensee declared an unusual event 
and began an orderly shutdown because time had expired for the limiting 
condition of operation (LCO) with the HPCI system inoperable.  During the 5 
months before taking this action, the licensee had observed high 
temperatures in the piping at the HPCI pump discharge and between two 
motor-operated valves (MOVs) (Points A and B on Figure 1 showing the normal 
configuration) near the interface between the HPCI system and the feedwater 
system.  During this time, the licensee concluded that feedwater was leaking 
back through feedwater isolation check valve No. 7 and the normally closed 
HPCI injection MOV No. 8, found deficiencies in about one-half of the pipe 
supports (16/34), and concluded that steam voids could form.  On October 23, 
the licensee declared the HPCI system inoperable.  To correct the problem 
temporarily, the licensee realigned the HPCI system valves to open MOV No. 8 
and close MOV No. 9 to serve as the normally closed HPCI injection valve.  
In a later inspection, the licensee found a bent stem and erosion of the 
disc and the seat in MOV No. 8.  In the realignment, the normally closed MOV 
No. 10 becomes subject to feedwater pressure.  If MOV No. 10 is not fully 
closed, then the open MOV No. 15 in the HPCI test return line to the 
condensate storage tank (CST) allows a condition conducive to water hammer 
when check valve No. 7 is leaking.  This alignment permits hot pressurized 
feedwater to flow in a cold low pressure test system (LER 50-237/89-29-01). 

Dresden Unit 3:  On October 31, 1989, while the plant was operating at 100 
percent full power, the licensee declared the HPCI system inoperable, having 
found conditions similar to those at Unit 2 described above:  similarly 
elevated temperatures, deficiencies in about one-half of the pipe supports 
(21/40), and valves that could be leaking.  The licensee found the discharge 
piping in the steam tunnel to be insulated, contrary to the original con-
struction documentation.  Initially, the licensee revised the alignment of 
the HPCI system as it had at Unit 2, to open MOV No. 8 and close MOV No. 9.  
The licensee later determined that feedwater was leaking back through MOV 
No. 10.  Accordingly, the licensee closed MOV No. 15 to return the HPCI 
system to operable status.  In a later inspection, the licensee identified 
damage in the seating surfaces in MOVs No. 8 and 10 and check valve No. 7 to 
confirm feedwater back leakage.  Closure of MOV No. 15 offers protection 
from a potential water hammer condition that could develop if MOV No. 10 
does not fully close when check valve No. 7 is leaking (also LER 
50-237/89-29-01).  

Dresden Unit 2:  On March 19, 1990, while the plant was operating at 96 
percent full power, the licensee was conducting HPCI system surveillance.  
Before testing, the valve alignment corresponded to that described above for 
the short-term correction to the October 23, 1989 event, i.e., MOV No. 8 was 
open and MOV No. 9 was closed.  To conduct certain tests, the licensee 
temporarily closed MOV No. 8 to isolate the HPCI system from the feedwater 
system, even though both MOV No. 8 and check valve No. 7 were still leaking.  
After complet-ing the routine surveillance tests, including a test of MOV No. 
10, the 
.

                                                            IN 91-50
                                                            August 20, 1991
                                                            Page 4 of 6


licensee began a quarterly valve timing test of the HPCI pump discharge 
valve MOV No. 9.  The licensee heard banging noises and observed HPCI pump 
discharge pipe movement.  The licensee terminated the timing test, restored 
system valves to the pretest configuration, and monitored the noises and 
movement until they ceased about one and a half hours later (also LER 
50-237/89-29-01).  

Subsequent valve manipulation and HPCI pump discharge pipe temperature 
measurement led the licensee to conclude that the root cause of the event 
was feedwater that had leaked back through the HPCI test return MOV No. 10.  
The licensee postulated that this valve did not fully close after one of the 
required manipulations.  This valve did not have a seal-in feature to 
complete the stroke after initiation, and the limit switch was set to bypass 
the torque switch in the open direction until the valve was 25 percent open.  
This same limit switch also controlled illumination of the "valve closed" 
indication light in the control room.  Consequently, the "valve closed" 
light would be illuminated over the part of the valve stroke for which the 
open torque switch was bypassed.  Thus, a control room operator who removed 
a closure signal when the light indicated the valve was closed would leave 
the valve approximately 25 percent open.  Accordingly, the licensee revised 
the appropriate procedures to maintain the closure signal for 30 seconds 
after the "valve closed" light illuminates.

In addition, the licensee revised the valve alignment of the HPCI system to 
protect against backleakage through MOV No. 10 by closing MOV No. 15.  

South Texas Unit 1:  On November 5, 1987, before the plant attained initial 
criticality, the licensee, Houston Lighting & Power, declared the A train of 
the auxiliary feedwater (AFW) system inoperable when a one-inch double valve 
vent line in the pump discharge piping was severed completely.  A second 
failure occurred three days later in a double valve instrument tap in the D 
pump discharge line.  In making the initial assessment, the licensee 
attributed the cause as water hammer resulting from improper venting of the 
system.  The licensee continued to note vibrations in the AFW system.  
During later testing, the licensee found that pressure pulsed when the flow 
control valves were in highly throttled positions.  The resulting 
combination of both hydraulic and structural resonances was sufficient to 
cause the damage.  The licensee made design changes to eliminate this 
problem (LER 498/87-016-01). 

Trojan:  On May 12, 1987, while the plant was in a refueling outage, the 
licensee, the Portland General Electric Company, was transferring water from 
the pressurized A accumulator (583 psig) to the depressurized D accumulator 
to prepare for maintenance on the A accumulator.  This event has also been 
discussed from a different perspective in Information Notice No. 88-13, 
"Water Hammer and Possible Piping Damage Caused by Misapplication of 
Kerotest Packless Metal Diaphragm Globe Valves," April 18, 1988.  The 
nozzle-to-pipe weld in the A accumulator one-inch fill line ruptured, 
spilling 2000 gallons of borated water.  The licensee repaired the line, 
satisfactorily hydrotested it, and again aligned the system for the 
transfer.  The licensee had not released the system for operations, but no 
controls were on the system because the clearance had been released.  This 
time, differential pressure was about 650 psig.  When the transfer was 
started, the licensee heard loud noises, stopped the operation, checked the 
arrangement of the valves and restarted the transfer.  After 
.

                                                            IN 91-50
                                                            August 20, 1991
                                                            Page 5 of 6


one more cycle of this sequence of events, the nozzle-to-pipe weld in the A 
accumulator fill line ruptured again.  

The licensee performed metallurgical analysis of the accumulator nozzle 
welds after each failure and found that the ruptures resulted from low 
cycle, high stress fatigue cracking.  This cracking resulted from excessive 
flow back through the packless diaphragm globe valve in the A accumulator 
fill line.  This backflow imposed a high differential pressure across the 
valve, causing the valve disc to vibrate.  The licensee concluded that 
operator error and insufficient procedures had contributed to this failure.  
The operators had failed to follow procedures and to determine the causes of 
the loud noises before proceeding.  The licensee had a procedure for 
transferring water through the sample lines but not through the fill lines.

The licensee developed a model for dynamic analysis and found that the 
backflow through the fill line resulted in the load on the nozzle-to-pipe 
weld being far greater than the pipe's failure threshold.  During a backflow 
test using a similar packless globe valve, the licensee found that the pipe 
would fail at a flow of about 70 gpm.  The licensee revised operating 
procedures to prohibit water transfer between the accumulators and developed 
an operation improvement plan (LER 344/87-013-01). 

Susquehanna Unit 2:  On October 12, 1986, while the plant was shut down, a 
water hammer occurred.  This event is discussed in more detail in NRC 
Information Notice No. 87-10, "Potential for Water Hammer During Restart of 
Residual Heat Removal Pumps," February 11, 1987.  The licensee, the 
Pennsylvania Power and Light Company, had established a temporary pathway 
from the B recirculation loop to the condenser for control of reactor water 
level while the residual heat removal (RHR) system was in service.  With the 
D RHR pump running, the licensee started the B RHR pump and then stopped the 
D RHR pump.  However, at approximately the same time, the outboard isolation 
valve in the letdown line from the B recirculation loop to the suction of 
the B RHR pump closed automatically, tripping the pump.  To compensate for 
the resulting loss of shutdown cooling, the licensee established alternate 
cooling using the control rod drive cooling system and the reactor water 
cleanup system.  The licensee reset the logic and reopened the valve to the 
B RHR pump suction without filling and venting the system.  The system had 
partially drained to the condenser through the temporary pathway and a water 
hammer resulted when the suction valve was opened.  To prevent water hammer 
from occurring in the future, the licensee reviewed this event and two 
previous similar events and revised procedures for reestablishing RHR 
service (LER 388/86-015-01).  

Other Plants:  Other water hammer events are discussed in the following re-
ports.  These events resulted from causes similar to those discussed in the 
USI assessment and reassessment.

    PLANT               SYSTEM                         REPORT        

Palisades          Accumulator Injection        NRC Inspection Report 
                                                  255/90-14
Oconee 3           Main Steam                   LER 50-287/89-02
Waterford 3        Steam Generator Blowdown     LER 50-382/89-15
ANO 2              Steam Supply to AFW          LER 50-368/88-23
Indian Point 3     Feedwater                    LER 50-286/88-02
Oyster Creek       Isolation Condenser          LER 50-219/88-21
Shearon Harris 1   Steam Generator Blowdown     LER 50-400/87-29-01
.

                                                            IN 91-50
                                                            August 20, 1991
                                                            Page 6 of 6


Discussion: 

These water hammer events occurred at both boiling water reactor (BWR) and 
pressurized water reactor (PWR) plants.  In BWRs, the events occurred in the 
RHR (shutdown cooling mode), isolation condenser, and HPCI systems.  In 
PWRs, the events occurred in the feedwater, main steam, auxiliary feedwater 
(AFW), steam generator blowdown, and accumulator systems.  These systems 
have been associated with water hammer in previous studies.  

Some aspects differ between these events, such as the location of the water 
hammer.  For example, the accumulator fill lines and injection lines and the 
AFW vent lines were not previously recognized as typical sites of water 
hammer.  However, the physical phenomena involve the previously identified 
mechanisms of formation of steam voids and fluid transfer between high and 
low pressure systems.  The licensees have cited a number of causes for these 
events, including improper filling and venting, the overly rapid stroking of 
valves, a lack of guidance about system configuration, accumulating water at 
low points, depressurizing a system to cause local flashing, and bypassing 
steam traps.  Such causes were addressed in NUREG-0927, Revision 1.  

These events illustrate the complex nature of water hammer events and 
hydrodynamic interactions.  The events at Dresden Units 2 and 3 point out 
the care that is necessary in altering system alignment during operation or 
to perform testing.  Such alignments can increase the susceptibility to 
water hammer.  Details of component operability and control features may 
easily be overlooked when the immediate goal is to find a means to continue 
operation.  

This information notice requires no specific action or written response.  If 
you have any questions about the information in this notice, please contact 
one of the technical contacts listed below or the appropriate NRR project 
manager. 




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

Technical Contacts:  Earl J. Brown, AEOD 
                     (301) 492-4491 

                     C. Vernon Hodge, NRR
                     (301) 492-1861

Attachments:  
1.  Figure 1.  High Pressure Coolant Injection System Normal 
      Valve Configuration
2.  List of Recently Issued NRC Information Notices 
.