ML20126F630

Special Study Pressure Locking & Thermal Binding of Gate Valves

ML20126F630
Person / Time
Issue date:	12/31/1992
From:	Caroline Hsu NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD)
To:
Shared Package
ML20126F122	List: ML20126F119 ML20126F630
References
TASK-AE, TASK-S92-07, TASK-S92-7 AEOD-S92-07, AEOD-S92-7, NUDOCS 9212310028
Download: ML20126F630 (40)

Text

.

o AEOD/S92 07 SPECIAL STUDY PRESSUltE LOCKING AND TIIERMAL lilNDING OF GATE VALVES DECEMilER 1992 Prepared by:

Chuck lisu t

I i Reactor Operations Analysis Branch Omce for Analysis and Evaluation -

of Operational Data l U.S. Nuclear Regulatory Commission Y$$#$0828gggos l: PDR-

. - . . - - . ~ . . -.

d CONTENTS A illlllEVI ATI ON S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v I NTR O D U C rl O N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1

DISCUSSION.................................................... 1 A. P u rp o s e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 II. Il a c kgr o u n d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1 C. Thermal Binding Phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . . ._ . .

3 D. Pressure la)cking Phenomenon ............................. 3 E. Conseque nces of 1;>cking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 P. Pr eve ntive M e t hod s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 10 Methods to Prevent Pressure 12>cking . . . . . . . . . . . . . . . . . . . . . . . ._ 10 4 Methods to Prevent Thermal Hinding . . . . . . . . . . . . . . . . . . . . . . . . 10 G. S u rvey Fi nd i ngs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

11. Syn ops i s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 S A FETY SI G N I FI CANCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 FI N D I N G S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 CO N C LU S I O N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 R ECO M M E N D ATI O N S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 R E FE R E N C ES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 .

A PP E N D I X A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 PR ESSUR E LOCKING EVENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 A. laiw Pressure Coolant injection and 12)w Pressure Core Spray System Injection Valves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 H. - Safety Injection System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 ,

C. Containment Spray System . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 D. Residual lleat Removal Shutdown Cooling Suction Isolation a V a l v e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - . . . . . . . 23 - -,

E. Residual lleat Removal llot Leg Cross Over Isolation Valve . . 24 F. Residual Heat Removal Containment Sump Suction -

Isolation Valve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 O. Residual lleat Removal Suppression Pool Suetion Valve .. . . . . 25

11. Residual IIcat Removal IIcat Exchanger Outlet Valve ...... 26

1. liigh Pressure Coolant Injection Steam Admission Valve . . . . . _26 J. Emergency Feedwater Isolation Valve . . . . . . . . . . . . . . . . . . . 26 .

lii -- -

4 CONTENTS (cont.)

A P PEN D I X H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28  :

TilERMAL BINDING EVEN'I3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 A. Reactor Depressurization System Isolation Valve . . . . . . . . . . . 28 B. Residual lleat Removal Inboard Suction Isolation Valve . . . . . 28-C. lili ' Pressure Coolant injection Steam Admission Valve . . . . . -28 D. Fower Operated Relief Valve Block Valve . . . . . . . . . . . . . . . -- 29 '

E. Reactor Coolant System Letdown Cooler Isolation Valve . . . . . 29 F. Residual IIcat Removal Suppression Pool Suction Valve . . . . . 29

-G. Containment isolation Valves . . . . . . . . . . . . . . . . . . . . . . . . . 30

11. Condensate Discharge Valves . . . . . . . . . . . . . . . . . . . . . . . . . 30 APPENDIXC.................................................... 31 >

SUMMARY

OF AEOD PLANT SURVEY . . . . . . . . . . . . . . . . . . . . . . . . . 31 J ame s A. Fi tzPa trick: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 31 ,

Ginna: ............................................... -33 Nine Mile Point: ........................................ 34-S al e m : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 -

H ope Cr e e k: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Susquehanna: ..........................................36 FIGURES f Figure 1 Flexible. Wedge Gate Valve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 5 Figure 2 Double Disc Parallel. Seat Gate Valve. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Figure 3 Pressure 12)cking Flexible Wedge Gate Valve . . . . . . . . . . . . . . . . . . . . . J6 b

l l

l l

(

. lv _ _ . _ _ _ ._ __

l

AllllREVIATIONS I

ASP Accident Sequence Precursor j

BWR boiling water reactor CCDP conditional core damage probability CS containment spray DBA design basis accident Ap differential pressure r

ECCS emergency core cooling system .

EFS emergency feedwater system GL Generic Letter IIELF,A high energy line break accident IIPCI high pressure coolant injection IST inservice testing LER licensee event report .

LLRT local leak rate test i

LOCA loss of coolant accident LPCI low pressure coolant injection LPCS low pressure core spray.

MOV motor-operated valve NPRDS Nuclear Plant Reliability Data System PORV power-operated relief valve >

- PWR- - pressurized water reactor-RHR residual heat removal SCSS Sequence Coding and Search System SDC- shutdown cooling SI ' safety injection TOL thermal overload V-

INTRODUCTION This study was initiated based on a report of an event at the James A. FitzPatrick plant, described in licensee event report (LER) 333/91-014, which described pressure locking of flexible wedge gate valves. Although the specific problem was revealed during a hydrostatic test which pressurized the bonnet area of the flexible wedge gate valve, it was evident that these valves could also become pressure locked during normal plant operation and thus, may not function during an accident. Subsequent investigation of the industry operating experience showed that similar events, caused by the phenomenon, continue to occur and challenge the operability of safety systems.

Tlic operating experience shows that double-disc and flexible wedge gate valves in many safety applications have not been operable due to pressure locking or thermal binding.

These valves, as a result of their design, can become pressure locked by being placed in operating configurations which subject them to high pressure fluids in the bonnet. Such forces oppose moving the valve dise from the seat and these forces were not ec,nsidered ,

when sizing the valve's motor operator. These valves generally have their thermal overloads (TOLs) bypassed and could fall as a resuit of experiencing this phenomenon.

Thermal binding occurs because the flexible wedge gate body contracts a greater amount during cooldown than the valve disc and pinches the disc in the valve seat.

Consequently, when the valve is closed hot and allowed to cool, the difference in thermal contraction can cause the seats to bind the disc so tightly that reopening is either extremely difficult or impossible until the valve is reheated. The operating experience has also shown that valve surveillance testing is usually conducted under conditions that would not detect or identify valve susceptibility to pressure locking and, in some cases, thermal binding. As a result, safety related valves subjected to the phenomenon would stay undetected in the locked state.

De potential for valve inoperability caused by pressure locking and thermal binding has been known for many years in the nuclear industry, la spite of numerous generic communications issued in the past by the Nuclear Regulatory Commission (NRC) and industry, pressure locking and thermal binding continues to occur to gate valves installed in safety related systems of both boiling water reactors (BWRs) and pressurized water reactors (PWRs). It appears that the generic communications to date have not led to effective industry action to fully identify, evaluate, and correct the problem.

DISCUSSION A. Purpose The purpose of this study is to provide a review of operating experience to: (1) identify conditions under which the phenomenon of pressure locking or thermal binding has occurred, (2) identify the spectrum of safety systems that have been subjected to pressure 1

locking or thermal binding, and (3) determine what conditions may introduce the failure mechanism under both normal and accident conditions. This study also provides an assessment of the safety impact due to valve operator motor or valve internal damage as a result of the associated valve being locked in the closed position due to pressure locking or thermal binding. Based on findings from this study, several recommendations are developed.

B. Background De problem has been addressed by the NRC and industry since 1977 (Reference 1).

Particularly, throughout the 1980's, the industry issued a number of event reports concerning safety related gate valve failures due to disc-binding. These failures were attributed to either pressure locking or thermal binding. Binding of gate valves in the closed position is of safety concern because gate valves have a variety of applications in safety related systems and may be required to open during or immediately following a postulated design basis accident (DBA). During such events, valve performance is severely challenged by the rapid cooldown and depressurization rates and the valves are exposed to the largest differential pressures (a p) across their discs. Valve operators are generally not sized to open a valve against binding forces generated by a high pressure fluid trapped hi the bonnet cavity (pressure locking) or when the disc was seated hot and subsequently cooled due to differential thermal movement (thermal binding). Pressure locking or thermal binding of gate valves representt a nonrevealing common-mode valve failure mechanism since normal surveillance tests may not detect or identify them.

The licensees of the plants with recent pressure locking and thermal binding events indicated that they did not address all of the potential operating conditions in their evaluations. One licensee has recently discovered that the root cause of a failed motor operator previously attributed to a broken torque switch was due to pressure locking.

Several licensees have indicated either to the author or stated in an LER that they planned to reevaluate gate valve susceptibility to pressure locking or thermal binding under all postulated system tests and operating conJitions.

Descriptions of pressure locking events involving gate valves installed in safety-related systems are listed in Appendix A of this report, he safety-related systems in which valves have become pressure locked include the high pressure coolant injection (HPCI) system, low pressure coolant injection (LPCI) system, low pressure core spray (LPCS) system, safety injection (SI) system, containment spray (CS) system, emergency feedwater system (EFS), and the residual heat removal (RHR) system. These events occurred under different operational modes in the time period from 1969 to 1992. The safety-related gate valves involved in these pressure locking events were LPCI and LPCS

, system injection valves, CS valves, RHR shutdown cooling (SDC) isolation valves, RHR hot leg cross over isolation valves, RHR contaimnent sump and suppression pool suction valves, HPCI steam admission valve, and emergency feedwater isolation valve, A review of these events shows that there were two potential causes of pressure locking; liquid entrapment in the bonnet and Ap across the disc while in the closed position.

Most of the events occurred during infrequent plant evolutions such as heat-up, 2

coo'. twn, and testing. The phenomenon can also occur during rapid depressurization.

Several events resulted in motor operator failures. All pressure lockings which occurred -

in these events have adversely affected the operation of motor-operated valves (MOVs),

and rendered the associated safety train unavailable.

A search of the nuclear plant reliability data system (NPRDS) database was conducted for thermal binding on gate valves. The search, which covered the period from 1983 to the present, identified a number of events involving thermal binding problems that posed a potential safety problem to plant operation. The desenydon of thermal binding events are listed in Appendix B. The safety related valves involved in these events were reactor depressurization system isolation valves, RHR inboard suction isolation valves, HPCI steam admission valves, pressurizer power-operated relief valve (PORV) block valves, reactor coolant system letdown isolation valves, RHR suppression pool suction valves, containment isolation valves (sample line, letdown heat exchanger inlet header),

condensate discharge valves, and reactor feedwater pump discharge valves. Thermal binding occurred when valves were closed while the associated systems were hot and allowed to cool.

In order to understand the extent and adequacy of the past and recent evaluations and corrective actions taken by licensees in solving the pressure locking problem, AEOD recently conducted a st'rvey of si selected plants. The primary areas addressed in the survey were: (1) methods of analyzing pressure locking potential, (2) methods of testings for leakage and operability, and (3) training programs provided for plant operating staff to understand the failure mechanism. The complete survey results are documented in the AEOD report," Survey of Licensee Actions Taken to Address Pressure Locking of Double Disk and Flexible Wedge Gate Valves" (Reference 2). A summary is provided in Appendix C and the survey findings are described.in the text (Section G) of this report.

C. Thermal Binding Phenomenon If a wedge gate valve is closed while the system is hot, thermal binding can occur as the system cools. The valve body and discs mechanically interfere because of the different thermal expansion and contraction characteristics of the valve body and the disc.. The difference in thermal contraction can cause the seats to bind the disc so tightly that:

reopening is either extremely difficult or impossible until the valve is reheated. This is particularly true for valves with internals which have reduced clearances _due to improper maintenance or alterations. Excessive closing force can contribute to thermal binding because excessive closing force causes the disc to be driven into the seat more tightly.

and, on cooling, the thermal binding effect is increased. Several potential remedies have been suggested to alleviate this situation, including slightly opening and reclosing a valve-

~

during _cooldown, limiting valve actuator closing forces, and using compensating spring packs to reduce valve initial closing forces. In general, neither ac nor de valve motor L operator sizing analyses account for the extra force needed to unseat a valve when it is thermally bound.

3

D. Pressure Locking Phenomenon Pressure locking in flexible-wedge (Figure 1) and double-disc gate valves (Figure L generally develops because of the nature of the design in combination with '

characteristics of the bonnet and specific local conditions at the valve. The esseru,s

' feature to develop pressure locking is the presence of fluid in the bonnet cavity including the area between the discs. The fluid may enter the bonnet cavity during normal open

, and close valve cycling at whatever line pressure exits at the time. Also, fluid may enter the bonnet cavity of a closed valve which has a Ap across the disc. The pressure differential causes the disc to move slightly away from the seat creating a path so that

the bonnet cavity becomes filled with high pressure fluid. Whether these situations lead -

! to a valve pressure locking scenario depends upon the fluid pressure when the bonnet 1 cmity was filled, temperature changes from when the fluid entered the bonnet cavity, and local line pressure compared with bonnet cavity pressure at the time the MOV is called upon to operate.

i Various plant operational sequences could introduce conditions conducive to pressure

locking. Irrespective of initial bonnet cavity fluid pressure (low o; high) and temperature, it is clear that a subsequent temperature increase of tb.

fluid will cause an i

increase in bonnet cavity pressure due to thermal expansion of the auid. The temperature increase can occur as fluid on either side of a disc heats up during various i modes of plant operation or possible changes in ambient air temperature caused by plant

operation, leaking pumps or valves, or in the event of a high-energy pipe break. In these i situations, the rate of temperature increase, which may be relatively slow to very high, controls the bonnet cavity pressure and valve susceptibility to pressure locking.

! Conversely, a bonnet cavity filled with high pressure fluid, such as leakage from the t

primary reactor coolant system, becomes a pressure locking candidate should a loss of-coolant accident (LOCA) or other transient cause pipe line depressurization.

l Additional consideration of local valve conditions illustrates the effect of a relatively small temperature increase when a gate valve is closed with fluid trapped in the bonnet' l cavity or area between the discs. As the system temperatures increase, the valve is

subsequently heated, the trapped fluid expands causing pressure to increase in the valve bonnet and between the wedges of the valve discs (Figure 3). The pressure increase j inhibits opening of the valve by causing the discs to press tightly against the valve seats, resulting in binding of the valve. The valve can be heated up and the trapped fluid expands as a result of heating either during normal plant startup or should a high-energy line break accident (HELBA) occur. The valve does not hwe to be in a high temperature system but only in close proximity where heu conduction thrt r.gh the pipe or via the surrounding air will heat the trapped fluid in the bonn

t cavity. The rate of pressure rise can be as high as 100 psi per 1 'F temperature increase in a solid filled bonnet for system temperature abave 450 *F. For lower system temperature (approximately 100 *F), the rate is reduced to 33 psi per 1 'F temperature rise. This type of pressure locking could cause failure of the valve pressure boundary parts under zero-leakage valve conditions.

4

h

/[

/ - -

/

!!!!(( l

\ (((([N J

Flexible wedge p

l _

.{_

Seat

, A r -

e m HHf y ,g/m/m> ft 6 y

Figure 1 Flexible-Wedge Gate Valve

-y ___---

UD/ na r- ///ZD

~

stem-au -

___ l g

Double f

/ /-

/ "" _

/y' 7 --

Parallel -

seat rings y /

b _

cf7///[/H/&g,}7' #/////7n Figure 2 - Double-Disc Parallel-Scat Gate Valve 5

t t 1 ,

a a I

% s- -

EXCESS \

\ P,RESSURE A SW/NPfWE4 / INTERCONNECTED

~~

> ,/ / ~ (EQUALPRES FLOW -

[/

- ' p/

. ~ /

}

/,////// , //////

z 2 Figure ; Pressure Locking Flexible-Wedge Gate Valve 6

t-

For the situation where the bonnet cavity is pressurized when the valve is closed and the valve is leak tight, the pressure will be retained, even when the rest of the system is depressurized, and then force the discs against the valve seats, causing double-disc drag forces. The resultant drag force can be excessively high such that the valve can not be opened when actuated. This type of pressure locking does not require that the bonnet cavity be filled solid with fluid. In fact, this condition can be' enhanced if a gas or steam bubble exists in'the top of the bonnet cavity because it will act as a pressurizer as the fluid eventually leaks past the seat. This action would prolong the effect of the pressure locking.

The LPCI and LPCS injection valves are susceptible to potential pressure locking during IDCA actuation. However, regular IST can not detect the potential problem because the inservice tests are normally performed during refueling outages when the trapped '

pressure in the valve has time to decay, no longer causing double-disc drag forces during valve testing. For the IST conducted during normal plant operation at some plants, high pressure nitrogen gas is introduced to the upstream side of the valve to reduce the Ap across the valve when performing u test. This is because the downstream side is at reactor pressure due to the check valve leakage. Introduction of high pressure nitrogen gas to the upstream side reduces the Ap to the design basis magnitude for the valvesc Testing under this condition will not detect pressure locking.

E. Consequences of Locking These phenomena can delay the valve stroke time or cause the valve motor actuator to stall. The events at FitzPatrick and Susquehanna indicate that the RHR/LPCI and LPCS injection valves of a BWR are susceptible to the cavity pressurization condition of pressure locking. In these two systems, the motor-operated injection valves are normally shut and are required to automatically open upon an actuation signal. Because the -

testable check valve between the reactor (or the recirculation line) and the injection -

valve is not a leak-tight valve, leakage past the check valve over time can pressurize the piping between the valves and the injection valve cavity to reactor pressure. Near leak '

tight seating surfaces of the injection valve may allow the valve cavity to remain pressurized and become subject to pressure locking when injection is needed during a LOCA. Under this condition, the bonnet pressure is more than 1200 psi, while the--

downstream pipe suddenly depressurizes to between 400 and 500 psi. The upstream pipe has a pressure around 300 psi. This high internal-to-external Ap across both seating surfaces would result in double-dise drag forces, which if they exceed the available thrust of the actuator, will produce pressure locking. Over time, bonnet pressure will decay, due to leakage past the seating surfaces or packing, at a rate dependent on the leak tightness of the valve. The valve will not stroke until the cavity pressure decays to a -

level less than the maximum allowable bonnet pressure. If the depressurization time for the valve is longer than the system response time to initiate valve actuation during a LOCA, the valve will lock up and the valve actuator motor will stall.

When a valve disc becomes locked'in a closed position due to pressure locking or thermal binding, actuation of the motor will result in locked-rotor current which will rapidly increase the temperature of the motor internals. Within 10 to 15 seconds, the 7

' heat buildup can degrade the motor's capability to deliver a specified torque, damage the motor, or both. These characteristic responses illustrate that MOV susceptibility to pressure locking needs to be assessed under all plant operating conditions. This can be appreciated by comidering anticipated extremes for an MOV that may have one safety function to open but is subject to pressure locking prior to its one time operation and another MOV that could be subjected to intermittent pressure locking followed by valve stroking (perhaps inservice testing [IST)).

The primary issues for the MOV subjected to pressure locking for one time operation are the force needed to overcome the effects of bonnet pressure, bonnet pressure changes with time, and the MOV opening time. The usual stroke speed of a motor operator is 12 inches per minute. Thus, MOV stroke times would generally vary from 15 seconds to 2 minutes for valves ranging from 3 inches to 24 inches in diameter. With this information it is evident that the 10 to 15 second heat buildup time becomes the critical element. If valve bonnet pressure is sufficiently high, pressure locking will result in operator motor burnout. The physical configuration of the bonnet and valve discs offers limited opportunity for the bonnet pressure to decay. When considered with the 10 second time for motor heat buildup at locked-rotor current, this suggests that depending on pressure decay to permit MOV opening is not a viable option. Also, the heat buildup could degrade the motor torque capacity so that the motor operator would not open the valve even under design basis loads. Thus, the concern relates to misinterpretation of potential beneficial effects from bonnet cavity pressure decay. For example, a calculation may show that the sum of the time delay to initiate the stroke (due to pressure decay) and the MOV stroke time may still. meet the minimum time required to inject fluid in the accident analysis. However, if the MOV received a signal-to open and drew locked rotor current for greater than 10 seconds while the bonnet pressure was decaying, then the valve would most likely fail to ever open because of either motor burnout or degradation of motor torque capacity.

One aspect previously discussed was valve operator motor damage due to locked rotor -

currents during pressure locking. TOL devices are frequently used to protect the motors of MOVs in safety systems from the effects of overheating. Regulatory Guide 1.106 (References 3 and 4) provides staff guidance for the use of such devices. The intent of the guide was to balance protection of the motor in comparison with assurance of system function. A review of operating experience, AEOD report S503," Evaluation of Recent .

Valve Operator Motor Burnout Events," (Reference 5) found that most TOL devices were either permanently bypassed or were significantly oversized. Thus, the' practical impact was that TOL devices would not protect the valve actuator motors from locked-rotor current (several events have been identified where the TOL device did not trip until the motor burned out). It is critical that the valve opt. rations,which require excessive loads be analyzed'to avoid the sustained locked-rotor conditions. Motor operation at locked rotor current will most likely result in motor damage with little chance that bonnet cavity pressure decoy could lead to successful MOV operation.

For the FitzPatrick event, based on leak rate testing results, two of the four low pressure emergency core cooling system (ECCS) valves could fail to open if the reactor vessel was rapidly depressurized as it would be following a large-break LOCA. The conditional 8

core damage probability (CCDP) for this event was estimated (Reference 6) to be 9.5 x 10's. The core damage probability was generated by utilizing the Accident Sequence Precursor (ASP) program event analysis. The event was modeled as an unavailability of two of the four LPCI and IECS injection valves for a 1-year time int:rval. Both IECI valves were assumed to be unavailable. Conditional failure probabilities of 0.3 and 0.5 were assigned to the two potential operable LPCS injection trains. The impact of the valve failure on large-break LOCA sequences was included in the analysis. A sensitivity study was then performed assuming all four ECCS valves were failed and the core 4

damage probability was estimated to be 3.9 x 10 . This is a factor of 4 higher than the nominal conditional probability estimated for this event cited above. The sensitivity study difference is primarily a result of the conditional probabilities assumed for the two LPCS trains, given the failed ISCI trains.

The potential for pressure locking also exists for the ECCS low head injection valves of PWR plants except for those valves which are normally open (this is the case for most Westinghouse plants). Operating experience has shown that interfacing isolation check valves between the reactor vessel and the injection valve are subject to leakage. Similar to the scenario described in the preceding paragraph for the RHR/LPCI injection valve and the LPCS injection valve, leakage past the isolation check valve can pressurize the injection valve to reactor pressure. If this pressure is trapped in the valve bonnet, then the injection valve may be subject to pressure locking when attempting to open during sudden reactor depressurization, such as for injection during a DBA LOCA and probably during SDC.

Three events (Vogtle Unit 1, Ginna, and Grand Gulf) have demonstrated that pressure locking can occur at a relatively low temperature increase, The lowest temperature at which this type of failure was reported in these events was slightly below 200 *F. This indicates that thermal conduction can heat up fluid trapped in the bonnet and cause pressure locking failures in valves which would not normally be censidered as having temperatures exceeding 200 F. Also, the event at Vogtle Unit 1 indicates that evaluation for pressure locking potential should be extended to valves which are normally open and to valves which would not be subject to thermal binding or pressure locking under ordinary circumstances to avoid the possibility that some safety-related valves having the potential for failures are overlooked.

F. Preventive Methods There are several preventative and corrective measures for pressure locking and thermal binding. Each method has limitations with respect to applicability, safety, effectiveness and cost. All of these measures have been atilized in the past. The following lists summarize these methods (many of them have been described in the previous generic communications listed in Reference 1).

I 9

_ _ . , _ _ _ _ _ _ . _ _ . _ _ . _ . _ _ - _ _ .__ _ . _ _ _ _ m _. __

Methods to Prevent Pressure Locking

1. - Drill a small hole on the upstream side of the valve disc (or upstream disc for a double disc valve) to relieve pressure buildup in the bonnet and between the discs. This method makes the valve unidirectional in sealing against high pressure.

The drilled side of the disc should always be towards the high pressure. It takes _

away the option sometimes used to prolong useful seat life by reversing the disc after a period of time; An alternative is to drill a hole in the bridge between the seat ring and the valve bonnet on the upstream side of the valve. This method also makes the valve unidirectional and should be installed accordingly. This is the simplest and a very effective method requiring no operator actions, and the cost is minimal.

2. bstal1 a pressure relief or vent valve in the bonnet to automatically relieve the bonnet pressure. This method requires the use of external components. If a manual vent valve instead of automatic relief valve is used to release pressure when the system heats up, operator action would be needed to position it.

3. Install an external bypass line with a manual valve from the bonnet to the upstream side of the valve. Manually open the bypass valve during heatup to relieve pressure from the bonnet. A relief valve or a relieving check valve also has been used. This method provides an alternate to method No.1 above when isolation in both directions is required.

4. For valves not required to provide complete isolation, stopping the valve disc travel by position limit switches rather than motor torque can keep the valve from going completely closed and thereby, prevent high pressure fluid from being trapped in the bonnet.

Methods to Prevent Thermal Illnding

1. Double-disc, parallel-seat valves are less susceptible to thermal binding than flexible-wedge gate valves. Replacing existing flexible-wedge gate valves involves additional cost. -It is best suited for plants in the design or construction phase.

2. _ While cooling a system, periodically open the valve slightly and then reclose it several times to allow uniform cooling and contraction of discs and bodies. This -

will involve changes of operating procedures and operator actions.

3. Ensure that the valve actuation or actuating medium is properly adjusted to prevent excessive closing forces on the valve disc. This may not be an effective -

means to prevent thermal binding if the temperature transient is large.

4. Installation of compensating spring packs on motor operators to absorb inertial closing forces after the motor has de-energized will avoid excessive closing forces on fast acting valves.

10

e i G. - Survey Findings AEOD recently conducted a survey of six selected plants (four BWRs and two PWRs).

The purpose of these six site visits was to: (1) understand the past and recent licensee evaluations and corrective actions concerning gate valve pressure locking and (2) identify -

any nonconservative or incomplete aspects of these licensee assessments. Appendix C-describes a summary of the survey report (Reference 2), Based on this summary, the following findings are provided:

1. Pressure locking is a potential generic problem to both flexible-wedge and double-disc, parallel-seat gate valves. Pressure locking has occurred to various sizes of both gate valve types at the plants surveyed as well as others. The valves involved were manufactured by Anchor-Darling, Alloyco, Crane, Lunkenheimer, Pacific,.

Powell, Rockwell, Velan, Westinghouse, and Copes Vulcan. Gate valves are also susceptible to thermal binding. Different types of gate valves vary in terms of susceptibility, the amount of disc Dexibility being the primary determinant. - From most susceptible to least, they are solid wedge, flexible wedge, split wedge, and double-disc, parallel-seat. . Double-disc, parallel-seat gate valves are one solution to the thermal binding problem because they are the least susceptible to interference oue to thermal contraction. This method of prevention may be an optimal solution because it does not rely on either operator action or devices to prevent thermal binding.

2. Valve surveillance testing may not detect or identify valve susceptibility to the problem. This is because the surveillance tests are normally conducted during either refueling outages or normal operating conditions. Pressure Jocking or ._

. thermal binding phenomena generally do not exist in these conditions. Most of them occur during infrequent plant evolution such as heatup, cooldown, and rapid" depressurization.

3. The scope of licensee reviews varied widely and there were no uniform guidelines.-

. Some completed their review based on engineering judgement without analysis.

. Although others used engineering analysis, the methods may not have been comprehensive nor conservative.

Leakage rate assumption was not based on actual operating conditions, and valve configuration used in the' assumption was not -

for valves with new, rebuilt, or reworked valve discs and seats.

Interface heatup such as heat from thermal conduction through the adjoining pipe was not considered.

Environmental temperature and pressure changes induced such as during a HELBA were not included in the analysis.

11

- Motor-operator thrust capability assumptions were based on vendor-supplied data rather than actual diagnostic testing results. For conservatism, the results should have considered the effects of diagnostic errors, live load packing, stem and stem nut drag, stem rejection load, etc. _ Also, in some cases, degraded voltage conditions should be considered.

4. No licensees surveyed had provided a training program on the subject for their-engineering staff. This may lead to incorrect conclusions during root cause findings.

5. Previous generic communications may not be specific enough to guide the licensees to identify valves susceptible to either pressure locking or thermal binding.

Although the phenomena which can cause pressure locking or thermal binding on gate valves were described in the generic communications, information was not -

provided based on system-specific conditions. Also, due to lack of system specific failure experience, the potential for the valve locking mechanism was not fully assessed on a case-by-case basis at some plants.

6. Two basic types of modification were used in the plants surveyed to prevent valve pressure locking. One modification was to drill a small hole in the upstream disc to relieve bonnet pressure. The second type was to install an external bypass line with a blocking valve between the valve bonnet and the upstream side of the valve to equalize bonnet pressure with the valve inlet pressure.

7. One licensee indicated they had surveyed several BWR plants and found that many plants have not modified the LPCI and LPCS injection valves to prevent pressure locking.

8. Most licensees surveyed did not initially consider the problem credible based on a limited assessment effort. Some even decided not to implement corrective actions after the potential problem had been identified.

H. Synopsis The operating experience illustrates that pressure locking and thermal binding have occurred under a wide range of plant operating conditions. Thus, MOVs could be-subject to conditions involving accumulation of damage to'the valve actuator motor due-to pressure locking or the load magnitudes could exceed the design basis loads of the -

operator. In addition, a LOCA could cause the consequential failure of a safety system

- installed to mitigate that accident (the LOCA creates the pressure locking in the MOV which renders it inoperable). De systems in which MOV pressure locking has occurred include HPCI, LPCI, LPCS, SI, CS, EFS, and RHR.'

12

Pressure locking occurs when the _ valve bonnet cavity pressure creates loads that exceed those which the valve operator was intended to overcome. High bonnet cavity pressure can result from valve open or close cycles where fluid fills the cavity and is subsequently heated by conduction from plant operation or an increase in ambient temperatures such as a postulated high energy line break. A differential pressure across the valve discs can also result in a high bonnet cavity pressure. Pressure locking that prevents movement of the valve discs off the seat causes the motor to experience locked-rotor conditions with .

very high motor current. Within 10 to 15 seconds, the motor internals heat up which may either cause motor burnout or degraded motor torque capacity so that the valve may not be able to perform during design basis conditions.

If pressure locking occurs, the mechanisms available to limit damage to the MOV motor appear to be TOL protective devices or bonnet cavity pressure decay due to leakage.

Previous AEOD studies have shown that TOL, devices frequently are either permanently bypassed or oversized. As a result, it is imperative that plant operating modes that potentially cause valve motor operator loads in excess of designed operator sizing be-analyzed. Although a maximum valve leakage rate is frequently specified, the actual leakage rate may vary with maintenance practices be: ween zero and a maximum level, but in effect it is indeterminate for a given condition. Further, if leakage cannot reduce the bonnet cavity pressure within about 10 seconds, then MOV performance is doubtful.

The potential beneficial effects of leakage to bonnet cavity pressure decay are diminished when pressure locking occurs because the increased cavity pressure will reduce the leakage through the valve discs.

Binding of the valve disc in the closed position due to thermal binding or pressure hcking appears to be a result of valve design characteristics. Thermal binding occurs occause of the w:.y a gate valve is designad, the seats move inward an amount that is proportionally greater than the disc shrinkage.when the valve cools after closing. This .

causes the seats to pinch the' discs tightly. Consequently, when the valve is closed hot i and allowed to cool, the difference in thermal contraction can bind the disc so tightly l that reopening is impossible until the valve is reheated. Pressure locking in flexible-l wedge and double-disc gate valves generally _ develops because of the nature of design in combination with characteristics of the bonnet and specific local condition at the valves.

l An operating characteristic of these valve designs is that fluid enters the bonnet cavity l during normal open and close cycling operations. = As system temperature increases, the bonnet fluid temperature eventually increases, resulting in potentially high pressure.

Another design characteristic is that when a valve has a~ differential pressure across the i disc in the closed position, the pressurized side of the disc can move away slightly from L its seat, allowing high pressure fluid to enter the bonnet cavity. If the pressure in the valve body is subsequently decreased, the bonnet pressure will result in double. disc drag l-forces against its seats.

L' Valve surveillance testing may not detect or identify the susceptibility of inadequately designed valves for these failure mechanisms. This is because the surveillance tests are l

normally conducted during either a refueling outage or normal operating conditions.

The phenomena which can cause the valve binding generally do not exist during these testing conditions. A comprehensive evaluation is needed to determine susceptibility to 13

the binding mechanism for a gate valve. This often requires various detailed analyses involving transient conditions and some unknown factors which may need conservative assumptions to demonstrate the analyses are adequate and complete.

On the basis of AEOD site visits and a review of recommended modification from valve suppliers, valve design modification is needed to effectively prevent occurrence of pressure locking. Two basic types of design modification are widely used in operating -

plants, one is to drill a weep hole in the upstream disc to allow trapped fluid to flow from the bonnet cavity. The second method is to install a vent line from the bonnet with a blocking valve to the upstream side of a valve to equalize with valve inlet pressure.

SAFFTY SIGNIFICANCE Flexible wedge and double. disc gate valves are used in a variety of applications in safety systems. Many of these valves are required to open to perform their safety functions during or immediately following postulated design basis conditions. Valve binding due to pressure locking or thermal binding is often synonymous with MOV failure to operate and inability of the associated safety systems to perform their safety functions. The safety implications of safety.related gate valve pressure locking are that it represents a potential common.cause failure that could render redundant trains of certain safety-related systems or multiple safety systems inoperable. These locking conditions generally develop from normal plant evolutions (Vogtle 1 and Grand Gulf events), leaking check valves (which is a well documented situation, FitzPatrick and Susquehanna 1 events), or actual line depressurization (San Onofre 1 and FitzPatrick containment spray events).

Thermal binding also constitutes a potential common-cause failure mechanism for safety-related valves. The valve disc binding generally is a result of normal plant evolution involving system temperature change (LaSalle 1, RHR inboard suction valve event).

Therefore, MOV failure to operate for any of these conditions represents a failure to comply with General Design Criteria 1,4,18, and 21 of Appendix A to 10 CFR Part 50-and Quality Assurance Criterion XI of Appendix B to 10 CFR Part 50.

Some safety systems are more vulnerable than others to potential unavailability depending on the associated valve susceptibility to the binding mechanisms. For the FitzPatrick event, the CCDP was generated by utilizing the ASP program event analysis.

The event was modeled as an unavailability of two of the four LPCI and LPCS injection valves for a 1 year time interval. Both LPCI valves were assumed to be unavailable.

Conditional failure probabilities of 03 and 0.5 were assigned to the two potentially operable LPCS injection trains. The impact of the valve failure on large-break LOCA sequences was included in the analysis. The CCDP for the event was estimated (Reference 6) to be 9.5 x 105. A sensitivity study was then performed assuming all four 4

ECCS valves were failed and the CCDP for this event was estimated to be 3.9 x 10 .

We would expect that valve binding in different safety systems would exhibit varying core damage probabilities.

14

4 FINDINGS

1. Comprehensive evaluation and extensive analysis, including consideration of plant system conditions and ambient conditions during all modes of plant operation, are needed in order to identify the valves susceptible to binding and determine the impact on safety 1.ystem function.

2. The binding problem is a result of inadequate design consideration and the most severe impact has not been accounted for in the accident analysis, or in typical probabilistic risk assessment.

3. Although the valve binding problem and corrective actions have been known in the industry for many years, the potential for this type of valve failure has not been fully evaluated or corrected.

4. The inadequacy in design or installation will not necessarily be found during subsequent plant startup testing or regular surveillance testing.

5. Two types of valve modification are used in the industry, depending on several considerations; maintenance, radiation, and leak rate test. These two modifications are: (a) drill a small weep hole in the upstream or downstream disc, and (b) install a vent line, with a block valve, from the bonnet to the main pipe.

6. Reliance on calculations of bonnet pressure relief in lieu of physical modification of the valve is not a reliable solution to the pressure locking problem. This is because several parameters used in the calculation are not constant and are subject to change. For instance,' leak rate can change following maintenance, repair or adjustment on the valve.

4 CONCLUSION As a result of this study, we conclude that previous NRC and industry efforts to-eliminate the gate valve pressure locking and thermal binding problems have not been ruccessful. Licensees continue to experience valve pressure locking or thermal binding events. The reviews conducted by licensees in response to the previous generic communications were not adequate. The scope and extent of licensee reviews varied widely. Some licensees completed their reviews based on engineering judgement without analysis; although others performed engineering analyses, the analyses were neither.

comprehensive nor conservative. Pressure locking and therrnal binding prevent valves from operating. The primary concern is that valve operators can not be sized to. account for the extra force needed to unseat the pressure locked valves. When a valve becomes locked in a closed position, actuation of the valve motor will result in locked rotor current which will rapidly heatup the motor internals. Within 10 to 15 seconds, it may either cause motor burnout or degraded motor torque capacity so that the valve may not be able to perform during design basis conditions.

15 4

Although TOL protection devices or cavity pressure decay due to leakage can prevent or limit damage to the MOV motor, neither of these is reliable. Most TOL devices are either permanently bypassed or significantly oversized. Cavity pressure decay is dependent of valve leakage rate which is indeterminate for any given condition. In any case, if leakage cannot reduce the cavity pressure within 10 seconds while the motor runs at locked-rotor current, then MOV safety function could not be assured.

Based on the history of industry problems with pressure locking and thermal binding, the following conclusions are provided:

1. Pressure locking or thermal binding can occur to gate valves used in safety-related systems of both PWR and BWR plants. These failure mechanisms have -

prevented safety-related systems from functioning when called on. Pressure locking is a potentially significant common mode valve failure mechanism with consequent ECCS system unavailability during accidents or severe depressurization transients.

2. Industry and NRC feedback on pressure locking has not been effective in precluding the potential problem.

3. Valve susceptibility to pressure locking may not be detected or identified during normal valve surveillance testing. A comprehensive evaluation to identify all gate '

valves that are susceptible to pressure locking or thermal binding is needed. ' The evaluation should include a detailed analysis to assure that all plant operating and accident modes are considered. Also, it appears that design modification is a simple, effective, and necessary step to assure proper operation of valves susceptible to pressure locking. 1 RECOMMENDATIONS The design characteristics of double-disc and flexible-wedge gate valves 'make them-susceptible to pressure locking and thermal binding under specific plant operating ._.

modes. This condition poses common-mode failure mechanisms that can render multiple -

ECCS systems inoperable. This situation has been known within the nuclear industry since 1977, but corrective action has been inadequate. Therefore, AEOD recommends '

that the Office of Nuclear Reactor Regulation address the following:

1. Licensees should evaluate all safety-related gate valves to determine potential susceptibility to pressure locking or thermal binding. The evaluation should- ,

employ indepth engineering analyses to cover all plant operating and accident modes.

2. For those valves identified as potentially susceptible to the binding mechanisms, licensees should implement effective salve modifications and appropriate procedures to prevent the binding from occurring.

16 l

s , 6 t

We suggest this action could be accomplished with either a supplement to Generic .

Letter (GL) 89-10 or a new NRC Bulletin. - Recommended action items a. and c. of GL 8910 should involve the types of analyses to satisfy recommendation (1) above to determine susceptibility to pressure locking or thermal binding. However, we recognize .

this option would represent Supplement 5 or 6 to that GL which may detract industry effort from specific concerns currently being addressed. Thus, a new bulletin may be a preferred option. Although selection of the generic communication method is optional, the operating experience strongly supports the need to assure corrective action is implemented.

Since most of pressure locking and thermal binding occurrences are a result of plant evolution, system transients, or unusual system alignment, we suggest that an NRC/ Industry workshop would be a constructive method to communicate the importance of system condition analyses in order to identify those vatves susceptible to the binding mechanisms.

4 0

0 1

17

1.

REFERENCES

1. . Generic communications on pressure locking and thermal binding issued by the NRC and industry.

a. Circa 1977 Westinghouse Issues Design Change to All-Westinghouse Supplied Gate Valves Requiring Vent Unes at Salem.

b. March 9,1977_

IE Circular 77 05," Fluid Entrapment in Valve Bonnets."

c. October 8,1981 IE Information Notice No. 81-31, " Failure of Safety Injection Valves to Operate Against Differential Pressure."

d. Circa 1981 Two Recirculation Discharge Isolation Valves Fail to-Open After a Scram at a BWR.

e. December 1,1981 General Electric Service Information Letter No. 368,

" Recirculation Discharge Isolation Valve Locking."

, f. Circa 1982 Foreign Utility Orders Valve Design Changes at All Power Reactors to Prevent Pressure Locking.

g. November 30,1983 Industry Communication on RHR Suction Valve Pressure Locking at a foreign reactor,

h. July 6,1984 AEOD Reissue of Study on Pressure Locking of Flexible Disc Wedge-Type Gate Valves.

i. December 14,1984 Industry _ Communication on Pressure Locking and -

Thermal Binding. ,

j. March 25,1988 _ Industry Communication on Pressure Locking of Both RHR Crossover Isolation Valves,

k. October 14,1988 General Electric Service Information Letter No. 368, Rev.1," Recirculation Discharge Isolation Valve Locking."

1. April 2,1992 NRC Information Notice 92-26," Pressure Locking of -

Motor-Operated Flexible Wedge Gate Valves."

2. C. Hsu, A. L Madison," Licensee Actions Taken to Address Pressure Lacking of:

Double Disk and Flexible Wedge Gate Valves," Office for Analysis and Evaluation of Operational Data, U. S. Nuclear Regulatory Commission, July 1992. -

18 e

3. U.S. Nuclear Regulatory Commission, Regulatory Guide 1.106, " Thermal Overload Protection for Electric Motors on Motor-Operated Valves."

4. U.S. Nuclear Regulatory Commission, ' Thermal Overload Protection for Electric Motors on Safety-Related Motor Operated Valves-Generic Issue II.E.6.1," U.S.

NRC Report NUREG 1296, June 1988.

5. E. Brown," Evaluation of Recent Valve Operator Motor Burnout Events,"

AEOD/S503, September 1985.

6. J. W. Minarick, J. W. Cletcher, D. A. Copinger, B. W. Dolan, Oak Ridge National Laboratory," Precursors to Potential Severe Core Damage Accident: 1991, A Status Report," U.S. NRC Report NUREG/CR-4674 (ORNL/NOAC-232), Vol.

15 and 16, September 1992.

7. L T. Guewa, Georgia Power Company," Plant Vogtle-Unit 1 NRC Docket 50424, Operating License NPF-68, Special Report on Pressure Locking of Motor-Operated Gates Valves," March 31,1988.

4 1

I 19 l

APPENDIX A PRESSURE LOCKING EVENTS The following pressure locking events occurred from 1969 to the present. Information relating to these events was obtained from LERs, licensee failure analysis reports, and international reports. LERs were identified by a search of the Sequence Coding and Search System (SCSS).

A. IAw-Pressure Coolant Injection and low-Pressure Core Spray System Injection Valves FitzPatrick- LER 333/91-014 Susquehanna Unit 1 - LER 387/91-013 LER 91-014 for FitzPatrick describes an event involving pressure locking of a flexible-we<1ge gate valve. The plant was shut down on May 7,1991, to repair valves in both LPCI lines. On July 17,1991, following corrective maintenance for valve and actuator problems with the outboard LPCI injection valve, a hydrostatic test of the piping was performed between the inboard (MOV-25B) and outboard (MOV-278) LPCI injection valves. The hydrostatic test pressure was around 2100 psig. Upon completion of the test, the piping between the valves was depressurized. A fill and vent of the system was initiated in preparation for returning the loop to service in the SDC mode.

Approximately 9 to 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> after completion of the hydrostatic test, the loop had been filled to the inboard LPCI injection valve. The operator attempted to open the 24-inch flexible-wedge gate valve (MOV-25B) from the control room. The actuator remained energized for approximately 30 seconds after which the motor actuator circuit breaker tripped. The normal stroke time for this valve is 120 seconds. The licensee determined the root cause of the motor actuator failure to be fluid at pressure trapped between the?

discs of a flexible-wedge gate valve. This phenomenon is known as pressure locking.

This type of valve arrangement is common in BWRs for both LPCI and LPCS injection lines, rendering both systems susceptible to the failure.

The LPCI and LPCS systems have a testable check valve between the reactor and the inboard isolation valve (flexible-wedge gates). The check valves are only required to reduce reverse flow to less than 10 gallons per minute. Small amounts of leakage past the check valves will eventually bring reactor pressure to one side of the flexible-wedge disc. The wedge will then flex, allowing fluid at reactor pressure into the bonnet.

Pressure of the order of 1000 psig could be reached in the bonnets of all leur low pressure ECCS injection valves (LPCI and LPCS systems). The licensee's calculations showed that bonnet fwid pressure in the range of 600 to 700 psig could be sufficient to lock the affected valves shut, preventing low pressure ECCS injection; (i.e., t:ie valve operators would not provide sufficient thrust to overcome the double-disc drag forces caused by the bonnet pressure.)

1 20

k i

i A second hydrostatic test was performed on July 28,1991. Instmmentation installed

- during this test confirmed that pressure locking was taking place. During this test, as pressure was increased to 800 psig, the rate of pressurization dropped to zero for approximately 30 minutes, indicating compression of air in the valve bonnet. Target test pressure of 2100 psig was held for 10 minutes and released.' Thirty minutes after depressurizat'on, operators attempted to open the valve from the control room. The actuator motor line current went to locked rotor current and the circuit breaker was manually opened by an electrician monitoring line current. The bonnet was vented through the ste.n packing gland. Coincident with the bonnet depressurization, valve -

position indication in the control room changed from closed to intermediate. The valve then stroked normally from the control room.

All four LPCI and LPCS injection valves were modified, prior to plant start-up, to incorporate a bonnet vent to the high pressure side of the valve. .

The '.PCI and LPCS injection valves at the Susquehanna plants are also susceptible to ,

potential pressure locking during LOCA actuation. On October 18,1991, with both Units 1 and 2 in Mode 1 at full power, an extensive evaluation of all safety related air operated valves and MOVs was conducted to determine susceptibility to pressure locking and/or thermal binding. The licensee determined that a susceptibility to the pressure .

locking phenomenon exists for the RHR/LPCI and the LPCS system injection valves.

This potential has existed since plant construction.

Based on the licensee's evaluation, the pressure locking phenomenon could occur if inboard isolation check valves leaked, allowing the LPCI and core spray injection valve bonnets to potentially pressurize to recirculation loop pressure. Following a rapid depressurization during a DBA-LOCA, this trapped high pressure fluid'could lead to pressure locking of the valve, increasing the force required to open the valve. The:

evaluation indicated that opening times for LPCI and LPCS injection valves will delay 1 second and 4 seconds, respectively. Although these times remain within the required ECCS response times, the margin of safety is reduced. -

The licensee modified 'the Unit I valves' by drilling a small hole in the upstream discs and planned to make similar modification to the Unit 2 valves during the next refueling' outage.

B. Safety Injection System San Onofre Unit 1- LER 206/81-020 Both trains of the SI system were inoperable during an actual SI signal when the <

> injection valves could not be opened as required on September 3,1981. There were no adverse consequences in this event since there was no LOCA. The reactor pressure -

remained above the SI pump's shutoff head and therefore no actual injection _of. water would have occurred if the valves had opened. However, if reactor pressure had -

21

+

decreased and actual injection been required, injection flow would not have been automatically available as designed.

Subsequent evaluations concluded that the valves failure to open was due to the pressure locking phenomenon. The high pressure water trapped in the bonnet cavity caused an-excessive bonnet to body A p when the pressure in the injection line decreased. The high .

Ap increased the disc-to-seat contact forces and locked the valves closed. All of the injection system valves had been successfully tested periodically during cold shu.down.

However, these tests were not performed while the bonnet to-body Ap conditions existed.

The valves were 14-inch double-disc gate type manufactured by Anchor-Darling.

C. Containment Spray System FitzPatrick - LER 333/88-013 During a refueling outage at the FitzPatrick plant on November 17,1988, the operator motor of the A CS valve tripped on TOL when attempting to stroke the valve as part of a post-work test. Subsequent investigation revealed that the valve operator motor had failed on overload due to trapped pressurized fluid in the bonnet cavity. The trapped fluid resulted in op across the valve disc in excess of the design value for opening the valve. The investigation also found that the same . excessive Ap condition could exist for the corresponding valve in Loop B. There was a possibility that both loops of CS would have been inoperable when required. The valves involved were the double disc gate type-manufactured by Anchor-Darling.

Prior to the test which resulted in failure of the CS valve, the RHR/LPCI system had been operating and then shutdown with only the pump minimum flow valve open for cooling of the pump internals. When the RHR/LPCI pumps are shutdown, system pressure at the elevation of the CS isolation vnive decays from approximately 240 psig to '

the normal operating pressure of the " keep full" system, which is approximately 100 psig.

Due to the leak-tightness of the valve design, the fluid trapped between the valve discs stays at 240 psig. Pressure upstream of the valve was approximately.100 psig and pressure downstream of the valve was approximately zero psig. This resulted in a Ap across the upstream disc of approximately 140 psi and approximately 240 psi across the _

downstream disc, The valve was designed to open against a Ap of 325 psi across only the downstream valve disc. The force required to open the valve Ap of approximately 240.

psi across one disc in addition to approximately 140 psi across the other disc is more .

than a design Ap of 325 psi across one disc.

The valves were modified by drilling a hole approximately 1/8 inch in diameter through the upstream disc to relieve the pressure in the bonnet and also preclude Ap across the disc with the hole so that the valve motor would only need to overcome A p across a single disc when opening the valve.

22

4 D. Residual Heat Removal Shutdown Cooling Suction Isolation Valve Foreign Reactor (August 7,1982)

These 900-MWe PWRs have two separate suction paths for RHR. With the primary system at 175 *C (347 'F) and 2.5 MPA (362 psig), two valves in series in one of the RHR suction paths failed to open. No problems resulted from these failures because the two valves in the second RHR suction path opened as designed. The two malfunctioning valves could not be opened until the insulation was removed from the valve bodies and the temperature was decreased to 44 C (111 *F).

The RHR valves were flexible-wedge gate valves. The utility determined that a major contributor to the failure was leakage past the upstream disc that allowed the bonnet cavity to reach primary pressure. When primary pressure was reduced, the high pressure water in the bonnet cavity was trapped, causing the disc to press more tightly against the valve seats. The utility also believes thermal binding between the valve disc and seat contributed to the failure. The high friction created between the disc and seat prevented the valve operators from opening the valves.

Turkey Point 4- LER 251/89-004 On May 23,1989, with the unit in Mode 4, the RHR normal suction isolation valve MOV-4-751 failed to open with its control switch. The valve was a Copes Vulcan double-disc gate valve. The reactor was being cooled down from Mode 2 in order to test .

the ability of the plant to shutdown the reactor using the alternate shutdown panel. This .

valve was to be opened from the alternate shutdown panel to place RHR in service for its normal heat removal function. Following the valve failure to open, the licensee used the steam generators to continue cooling the reactor. Two subsequent attempts to open the valve with the control switch were unsuccessful. Both attempts resulted in breaker tripping on TOL The valve was then manually opened.

The licensee's evaluation concluded that the valve failure was caused by pressure locking.

Leakage past the upstream disc allowed the bonnet cavity to reach primary pressure.

When primary pressure was reduced, the upstream disc seated tighter than expected due to the cavity pressure. This trapped the high pressure water in the bonnet cavity and between the discs causing the discs to press more firmly against the valve seats. The motor operator was not able to overcome the friction force which was higher than expected. Valve MOV-4-750 downstream of MOV-4-751, was also determined to be susceptible to pressure locking.

The licensee's corrective action included installation of a pressure equalizing line on each valve. The same problem was also identified to occur in the corresponding valves in Unit 3. These valves were also modified by installing an equalizing line.

-23

l E. Residual Heat Removal llot Leg Cross-Over Isolation Valve Vogtle Unit 1 (Reference 7)

On January 28,1988, at Vogtle Unit 1, both RHR hot leg crossover isolation MOVs failed to open during surveillance testing. The unit was in Mode 4 (approximately 320 F and 350 psi). Prior to entering Mode 3, a routine quarterly RHR pump and valve test was planned. This test required opening the valves that had been closed during the outage when system temperature was approximately 100 *F to ensure isolation between the two RHR trains. During performance of the test, both valves were sequentially given a signal to open from the control room. Within about 30 seconds, an area smoke alarm was received, and the circuit breakers of the two motor operators tripped on overcurrent.

Both motors had overheated and failed, and the valves remained in the fully closed-position.

The licensee's subsequent evaluation concluded that pressure locking was the most likely cause of both motor operator failures. Prior to January 28, both valves had been closed under ambient temperature conditions. The temperature at the valves then increased to approximately 300 'F due to heat conduction over a time period. The temperature increase caused expansion of the fluid in the valve bonnets, and the increased bonnet

- pressure forced the discs against the seats with such force that the motors stalled and failed before unseating the discs. The valves are Westinghouse 8-inch double-disc gate valves with Limitorque SMB-00 actuators.

To prevent recurrence of pressure locking, the valves were modified by drilling a 3/16 inch diameter hole through the upstream disc. _ This hole provides pressure equalization between the bonnet area and upstream piping. In this case the drilled upstream disc for.

the valves are on the opposite sides of the header such that leak tight isolation of a passive failure of the header can always be obtained.

The licensee's review of the system operating history revealed that a motor on one of the actuators failed in October 1987. The failure was attributed to a broken torque switch, but the failure conditions were similar to those for the January 1988 event. It is now -

believed that the motor failure in the 1987 event was also the result of pressure locking.

The pressure locking phenomenon ir. the stated operating condition was not identified in the review that the licensee conducted in response to an industry generic communication issued in 1984 (Reference li). The licensee did not assess all of the potential operating conditions.

F. Residual Heat Removal Containment Sump Suction Isolation Valve Ginna On December 16,1969, the Ginna plant experienced a pressure locking of thelRHR containment sump suction isolation valve. The valve could only be opened one-third of 24

. . l l

l its total travel during regular operability testing. The valve was a 10-inch double-disc Anchor Darling gate valve. Upon disassembly both discs were found to be bowed outward, preventing full opening. The direction of the bowing indicated that there was a buildup of pressure between the discs. It was determined that the pressure increase was caused by heatup of the water entrapped in the bonnet and between the discs during plant operation. Vent lines were subsequently installed on both the sump isolation  !

valves to preclude pressure locking. This modification was recommended by Westinghouse and Anchor-Darling.

G. Residual Heat Removal Suppression Pool Suction Valve Grand Gulf- LER 416/92-001 The Division 1 RHR system pump suction valve from the suppression pool failed to open when operators were realigning the system from the SDC mode to the standby LPCI mode as reactor heatup commenced following a plant outage on January 8,1992. .

The motor operator TOL circuit tripped on the first attempt to open the valve from the control room. The overload device was reset and immediately tripped again. When the' trips occurred, reactor coolant temperature was approximately 175 'F and suppression pool temperature was 74 *F. The valve is a 24-inch Powell flexible-wedge gate valve.

Operators reestablished Division 1 RHR SDC operation and reduced reactor coolant temperature to approximately 135 *F. The TOL circuit tripped again during another -

attempt to open the valve. Operators then attempted unsuccessfully to open the valve manually. The temperature difference across the valve was 43 'F.

The valve was successfully opened via the motor operator when reactor coolant #

temperature was further reduced to a value corresponding to a 20 *F difference across the valve. The valve failure to open was determined to be caused by pressure locking. ,

As a result of heating the trapped water volume in the valve bonnet during plant startup, the trapped water pressure increased, causing additional forces on the valve wedge. The licensee's subsequent evaluation concluded that both Division 1 and 2 RHR suppression pool suction valves have the potential to become inoperable due to pressure locking when subjected to a temperature differential of.approximately 15 *F or greater across the-valve. Such conditions may exist with the respective RHR train in the SDC mode configuration. An RHR train in SDC mode operation with this limitation is inoperable for the LPCI safety function.

In a previous evaluation, the licensee, based on engineering judgement,~ had considered these two suction valves not to be susceptible to thermal binding or pressure locking because of their distance from potential heat sources.

Both RHR suppression pool suction valves were modified with vent lines to prevent pressurization of the water trapped in the bonnet. The vent lines were connected from the valve bottom to the upstream pipe.

25

'H. Residual Heat Removal Heat Exchanger Outlet Valve LaSalle Unit 1- LER 373/83-117 At LaSalle Unit 1, on September 20,1983, and again on November 12,1983, with the unit in cold shutdown, the "B" RHR heat exchanger outlet valve failed to open by either "

the motor operator or manually.- This valve was a flexible-wedge gate valve manufactured by Anchor Darling. The valve failure made the B SDC loop and B suppression pool cooling loop inoperable. It is believed that the valve became -

inoperable in the closed position due to high pressure water trapped in the bonnet cavity.

The bonnet cavity does not have a mechanism to vent the trapped high pressure water and, therefore, the valve discs became pressure locked. At the recommendation of the valve manufacturer, the valve limit switches were temporarily adjusted such that the disc travel was stopped by position and not by torque. This change kept the valve from going completely closed and, thereby, prevented high pressure water from Zag trapped in the bonnet cavity. This mode of operation was allowed on these particular valves because they are not required to provide zero leakage isobtion.

The permanent modification was to drill a weep hole in one of the discs to provide a -

vent path from the bonnet cavity.

I. High-Pressure Coolant Injection Steam Admission Valve Brunswick Unit 1 - LER 325/88-012 On May 28,1988, while Brunswick Unit I was at full power, the HPCI steam admission valve failed to open when the switch was energized during a periodic test. The failure caused a total loss of system function. The cause was determined to be pressure locking.

The valve was previously closed with fluid trapped in either the bonnet cavity or between-the discs. When the valve was subsequently heated, the trapped fluid expanded or flashed to steam and caused pressure to increase. The pressure increas_e inhibited opening of the valve by causing the wedges to press tightly against the valve seats, resulting in binding of the valve, To prevent recurrence the licensee drilled a small hole

in the upstream disc to provide a pressure relief path. The valve involved was an Anchor-Darling flexible wedge gate valve.

J. Emergency Feedwater Isolation Valve Foreign Etwaur (March 22 and August 16, 1990) his event occurred at a foreign reactor. The utility's inservice test procedure for the EFS required that the isolation gate valve should be tested prior to running the pump.

On March 22, and August 16,1990, during two inservice tests, the utility revised the procedure, running the pump before exercisir.; the valve. In both instances, the isolation gate valve failed to open. The valve was a double-disc gate type. A subsequent

.26

. . . - _ - . . -.. - - . - ~ . . . - - . . - - ., - - . . . - _ .-

evaluation indicated that the valve failure to open was due to pressure locking. During the pump test, water was pumped against the closed valve's disc on the pump side, forcing this disc away from the valve seat, which permitted high pressure water to enter the valve body. The water at pump discharge pressure remained inside the valve following the pump shutoff, and pressed the double discs strongly against the valve seats.

The actuator motor torque was insufficient to unseat the valve. A calculation, without considering the safety factor, showed that the motor torque needed to open the locked l-valves would be about 355 Nm (262 lbs ft) which is higher than the torque limit switch'

.' setpoint of 265 Nm (195 lbs ft). The reported inservice test procedure will not detect the

pressure locking condition, because the valve is exercised before the pump test is run, -

and under such conditions, no pressurized water is entrapped inside the valve.

4

+

27

APPENDIX Il TilERMAL IIINDING EVENTS The thermal binding events listed below were obtained from a search of the NPRDS and SCSS database files. The search covered the period from 1983 to the present. Some of these events found through the NPRDS search were not reported via the LER process.

A. Reactor Depressurization System Isolation Valve Big Rock Point- LER 155/84-001 At Big Rock Point, three of four reactor depressurization system isolation valves failed to open during surveillance testing on February 22,1984. The plant was in Mode 4 with reactor pressure at 50 psig. The valves involved were air-operated flexible-wedge gate valves manufactured by Anchor-Darling. The failure was caused by the disc wedging tightly into its seat so that the air operator spring force was insufficient to open the valves. The valves had been closed while hot. When the valves cooled during plant shutdown, thermal binding occurred.

B. Residual Heat Removal Inboard Suction Isolation Valve LaSalle Unit 1- LER 373/83-142 On November 4,1983, at LaSalle Unit 1, while the unit was in hot shutdown for planned maintenance, the inboard suction isolation valves could not be opened by either the motor operator or manually. The valve was a flexible-wedge gate valve manufactured by Anchor-Darling.

The valve was manually seated to stop leakage during the previous operating period.

Subsequent'y, with the plant in hot shutdown and temperatures significantly less than when the valve was manually seated, the valve failed to open due to thermal binding.

The valve was opened by externally heating the valve body.

C. High-Pressure Coolant Injection Steam Admission Valve Brunswick (Jnit 1- LER 325/88-017 On July 1,1988, at Brunswick Unit I with the unit at 68 percent power, while performing the operability test of the unit HPCI system, the HPCI turbine steam supply isolation valve would not open. The unit was then shutdown. Subsequent evaluations found that the valve failure to open was due to thermal binding and that efforts to open the valve had damaged the motor windings. The valve was a flexible-wedge gate valve manufactured by Anchor-Darling.

28

D. Power Operated Relief Valve Block Valve At a domestic plant on June 16,'1990, while performing the PORV cycle test, the PORV isolation valve RC-11 failed to open from the closed position on demand from the control room. When this occurred the plant was in preparation for start up following a refueling outage and had to cooldown for a valve replacement. Following the plant cooldown, the valve stem was found separated from the discs when the operators managed to open the valve manually. Subsequent inspection showed a visible crack on the face of the valve seat. The cause of the valve failure was determined to be due to thermal binding. The valve involved was a Vulcan gate valve.

E. Reactor Coolant System Letdown Cooler Isolation Valve Davis Besse- LER 346/90-002 On January 26,1990, the letdown cooler isolation valve MU-2B at Davis-Besse was closed to isolate reactor coolant letdown following a reactor trip. On January 29, MU-2B was opened but no flow was observed. The operators then closed the valve and observed the following: (1) erratic position indication, (2) motor operator went to stall current, and (3) stem to disc separation. The valve failure was determined to be caused by thermal binding. When the valve was closed during the ttip, flow through the valve stopped, the valve cooled relative to its temperature when there is flow through the valve. The rigid disc wedged into the seat as the valve cooled. When operators attempted to open the valve, the stem separated from the disc. The valve was a Velan solid wedge gate valve.

F. Residual Heat Removal Suppression Pool Suction Valve ,

Six thermal hinding events occurred to four of the RHR suppression pool suction valves at both units at a domestic site. Two events occurred at Unit 1_on Jammry 28 and March 7,1991. The four other events occurred at Unit 2 on June 26,1989 and -August :

17,1990. At Unit 1, the RiiR suppression pool suction valves 1-E11-F004B and 1 E11 F004D would not open and the operator motor tripped on TOLs when an open signal was given on two separate occasions. The unit was at zero power when the event .

occurred. At Unit 2 with the unit at zero power, on June 26,1989, during periodic testing, the RHR suppression pool suction valves 2-E11-F004D and 2 E11 F004D would' not open when the signal to open was given. The valves were then found to histuck in

~

the seats The failures were due to thermal binding. The valves were disassembled,~-

cleaned, and satisfactorily tested following reassembly. _On August 17,1990, these two suction valves again would not open from the reactor turbine gauge room due to valve operator motors tripping on the TOL The root cause of the failures was thermal .

binding. The system operating procedures were revised to allow the ' valve to be manually cracked off the seat prior to motor operation to alleviate thermal binding. The valves involved were Anchor-Darling flexible-wedge gate valves.

29

q G. Containment Isolation Valves At a domestic plant on December 18,1988, with the unit in cold shutdown and preparing to heatup, the chemistry technician attempted to sample the pressurizer steam space and found that the containment isolation valve for this sample line.would not open. This is a solenoid operated valve and no electrical problems could be found, The licensee -

determined the failure was due to thermal binding. The valve was completely rebuilt and returned to normal service. The valve was a Rockwell gate valve.

At another domestic plant site, on October 23,1988, with the plant in cold shutdown, it was discovered during surveillance testing that the letdown heat exchanger inlet header containment isolation valve would not stroke to the closed position. The cause of the failure was determined to be thermal binding. An investigation revealed that the valve.

discs expanded greater than the body seats when heated. A design change was implemented to replace the existing gate valve with a globe valve. The valve involved was a WKM valve.

H.- Condensate Discharge Valves Thermal binding occurred on the two condensate discharge valves (MO 299813 and MO-2998A) at a plant on two separate occasions (April 2 and April 26,1991).. With the plant in the cold shutdown condition for refueling, personnel performing preventative maintenance activities for the valve were unable to get the valves to stroke open. Since the system was not in service at these times, the failures had no effect on plant operation. The cause of the failures was seat binding as a result of thermal cycling. The valves were closed when the valves were still hot. This was contrary to_ the plant -

procedure. As they cooled, the valve bodies contracted, causing the disc to bind. - A-contracting firm was brought in to heatup the valve to allow it to be opened. They unstuck when the temperature reached approximately 225 *F.

30

e t

h APPENDIX C SUMMAltY OF AEOD PIANT SURVEY Despite actions taken by licensees in response to previous generic communications issued by the NRC and industry, the pressure locking problem still occurs to gate valves i installed in safety-related systems. In order to understand the extent and adequacy of the corrective actions taken by licensees, AEOD conducted a survey of six selected plants (four DWRs and two PWRs). The site visit at one plant (Nine Mlle Point) was canceled i from the survey process because of an ongoing outage and other NRC inspections.

Ilowever, tl.c evaluation documents from the plant were reviewed. Survey results are documented in AEOD report " Survey of Licensee Actions Taken to Address Pressure -

Locking of Double Disk and Flexible Wedge Gate Valves" (Reference 2). This Appendix provides a summary of the survey results. .

i James A. FitzPatrick: -

1. Ilad not considered problem credible because:

a. Imi plant specific pressure locking events. ,

b. Lack system specific failure experience,

c. 'lhe specified phenomena in industry communications are too general, not specific to LPCI system.- l

d. Engineering could not provide a comprehensive analysis and calculation to convince management to initiate corrective action.

c. Not able to identify valves with the potential for pressure locking or thermal binding conditions,

f. Engineers did recommend modification to the LPCI injection valves in 1985. Management declined to take action due to lack of actual operating event.

2. Following the 1991 event (LER 91014)

A consulting firm (Failure Prevention Inc.) was hired to evaluate impact on other motor-operated, ficxible-wedge gate valves. Eleven valves were sek,.ted to be examined for the following three scenarios:

^

a. 'Ile water trapped in the valve bonnet expands as a result of heating during a normal plant startup.

31

, _- __ . - _ _ _ ._- __ _ _ ~ _ . . . . - - - . _ _ _ -

4

b. The water trapped in the valve bonnet expands as a result of heating during a llELilA.

c. One side of the valve is inidally pressurized by check valve leakage and then suddenly depret.surized as a result of a LOCA.

'Ihe analysis concluded that none of the 11 valves is subject to motor stall current under the first two conditions. This conclusion is based on the expectation that leakage rates, as derived from the current local leak rate test (LLRT) leak rate rnensurements, are greater than the bonnet water thermal expansion rates for all eleven valves. Among the  !

11 valves, four valves are affected by the third condition.

LPCI: 10 MOV-25A 12.3 see LPCS: 14 MOV 12A 30.0 sec ,

10 MOV 251155.4 min 14 MOV.1213 2.33 min All four valves will not function until the pressure of the fluid trapped in the valve bonnet decays to a level less than the maximum allowable bonnet pressure the valve actuator can overcome to open the valve. 'Ihe depressurization times for these valves are estimated to be from 12 seconds to 55 minutes. These time periods are larger than the required valve response times during a LOCA (two valves: 10 MOV 2513 AND 14 ,

MOV-1211).

Although the analysis provided analytical justification for delaying installation of a.

bonnet vent for the remaining seven valves, the contractor recommended modifications during the next outage to eliminate the possibility that maintenance or repair on the subject valves could result in a reduction of leak rates and an increase in leak off times.

The licensee is still evaluating reactor core isolation cooling,11PCI, containment cooling, and sump isolation valves for modification.1-lowever, they have indicated that modification for these valves may not be necesn'y due to opening the valves during rapid depressurization being unlikely and the o p ...ross the IIPCI valve being low during '

a small break LOCA. (This may not be true for the conditions during a large break -

LOCA).

3. Training Program The licensee originally did not provide training on this failure mechanism. Following generic communications, the issue was placed in required reading and maintenance lesson plans. The item was not incorporate.d in training courses for operations personnel until 1988. In 1989, a training plan was provided for maintenance personnel. _ The item ,

was then added to the continuing training program for engineering support groups following the 1991 event.

32  ;

4. Surveillance Test

%e LPCI valve IST surveillance was performed quarterly. Operability testing was performed monthly.11efore cycling the valve, nitrogen ghs was injected into the upstream side of the valve to reduce the o p across the valve disc if downstream pressure was too high. The nitrogen pressure could go up to 400 psig. The licensee believed the downstream high pressure would decrease or diminish during a LOCA; therefore, a large A p condition was not possible.

Ginna:

1. An RilR containment sump suction isolation valve experienced a pressure lockup during a monthly test in 1969. This was a pressure locking scenario due to expansion of entrapped water in the valve bonnet cavity as a result of heating.

The valve was a 10 inch double-disc gate type manufactured by Anchor Darling.

Vent lines were subsequently installed on both sump isolation valves. His modification was recommended by Westinghouse and Anchor Darling. Following the event, the licensee initiated an inspection of all other valves of the same type provided by Anchor Darling and Alloyco. Thirty valves were motor-operated and fourteen were manual. Modifications were recommended for several normally closed MOVs which are provided with heat tracing. These valves are: (a) refueling water storage tank supply to Si pumps, (b) boric acid supply to Si pumps, and (c) emergency boration line. The results of this inspection were used in their response to IG Circular 77 05 in 1977.

2. Industry Communication:

The licensee considered that valves manufactured by a company other than Anchor Darling were not susceptible to pressure locking. The evaluation done for IE Circular 77 05 was determined to be adequate for their response to the -

industry communication.

3. The licensee was not aware of the pressure locking potential for the low head injection valve until the AEOD visit. They had never considered the possibility of check valve 'eakage. The recent LLRT for both the low head injection valve and downstream check valve showed that the valves were leak tight with near zero leak rate. The low head injection valves are 6 inch Velan gate valves.

The licensee planned to initiate a review of all other gate valves for potential pressure locking. Subsequent to our visit, the licensee hired an independent consultant to assess the problem. ne assessment concluded the low head pressure injection valves are susceptible to pressure locking.

33

4. Training Program The failure mechart .oncern was provided in the training instructions for operations and malmenance personnel. However, no such training was provided for the engineering staff.

5. Surveillance Test Surveillance testing of the low head injection valves was performed by cycling the valves within 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> after plant cold shutdown, f

Nine Mile Point:

l. The licensee pro"Ided documents that revealed it had reviewed all double disc and flexible wedge gate valves manufactured by Anchor Darling, Rockwell, Crane,

• and Powell in response to industry communications on pressure locking issues.

Additionally, the engineering organization supporting Nine Mile Point had recommended modifications to prevent valve pressure loeking. However, plant managers did not approve the implementation of the recommended modifications because valve pressure locking was not perceived as credible since they lacked site specific experience with the phenomena. Therefore, no modifications were performed at Nine Mile Point.

2. The licensee had provided training to operations and maintenance personnel but did not appear to have provided training for engineering personnel.

Salem: ,

1. All of the double disc, or flexible wedge, gate valves installed in other than cold '

water systems at Salem have been previously surveyed for susceptibility to thermal binding and pressure locking by the licensee and Westinghouse during the plant-design and construction phase (commercial operation: Unit 1 in June 1977, Unit 2 in October 1981). A 3/4 inch external pressure vent line, with a block valve, was installed between the valve bonnet cavity and the upstream pipe in susceptible valves. Examples, are the RHR and Si systems.

2. In their evaluation in 1986, the licensee concluded that the previous modifications

~

in the plant design and construction stage were adequate to prevent pressure locking or thermal binding from occurring; There is a very low probability for the problem to occur to these gate valves without modification. This is based on: (1) documented history of no valve failure due to pressure locking or thermal binding, and (2) all gate valves in safety systems were procured to a design specification that cites an allowable seat leakage of 2 cc/hr/in of nominal valve size for its hydrostatic seat leak test. (This is the maximum leakage rate at hydrostatic test pressure. The actual out leakage rate would be less. A conservative assessment

_34 __ __ _

O I

l should assume an out leakage rate based on a Eate valve with new, rebuilt, or reworked valve discs and seat under an actual test with higher A p across the valve). The licensee indicated that generally, there is a low probability that liquid entrapment or pressure locking of gate valves will occur since the trapped liquid will leak past the seat or possibly through the packing gland area to prevent i pressure buildup. '

3. Training Program The item has been included in the training program for maintenance and operations personnel since the time they performed their evaluation. In process of developing training instructions for engineering staff subsequent to our survey,

4. Surveillance Test Surveillance test of low head injection valve was conducted during plant cold shutdown.

Hope Creek:

1. Based on the engineering evaluation conducted in response to an industry communication, the licensee did not consider the gate valves at the plant to be i

susceptible to pressure locking or thermal binding. Therefore, the licensee chose not to modify any of the gate valves.

The licensee evaluation indicated that: I

a. There was no history of gate valve failure due to this mechanism.

b. Generally, the gate valves function as equipment isolation maintenance valves and thus, they are locked open, or normcIly in the open position during plant operation. Gate valves operated in the open position preclude liquid entrapment, pressure locking or thermal contraction.

c. There is a low risk of liquid entrapment or bonnet pressurization because the fluid will leak past the seat, or possibly through the stem packing to w relieve bonnet pressure.

- d. Although General Electric had recommended modifications to several gate -

valves in the recirculatirm and RHR system (CS, RHR injection), the licensee chose to use maintenance and operating procedures to prevent the problem from occurring, such as removal of insulation to allow valve -

cooling, loosen the valve packing to relieve bonnet pressure, or alternately open and close the affected valves during cooldown or heating up to prevent thermal binding. (The measures chosen by the licensee may not_

35

.,,-.-4 -,,w-w.+ ----w -, .. ~ ,m -c.m , - . . . . , , - . . y%. -r . . ,, vn..-. ,, e. . - - . . e

9 .

be effective in preventing occurrence of pressute locking or thermal binding).

2. Training programs on this item had been provided for maintenance and operations personnel. The licensee plans to provide the same training to engineering personnel.

3. The licensee admitted that depending on engineering judgement to determine depressurization rate for fluid entrapped in the bonnet is not adequate. They said they plan to initiate a complete reevaluation of the plant gate valves.

Susquehanna:

1. An initial review in 1985 identified that only lulR SDC valves HV F008 and HV-F009 were susceptible to thermal binding and/or pressure locking. As to Si valves, the licensee considered that the associated check valves would release the pressure in these valves, therefore pressure locking would not occur to them.

Although the review recommended modification to valves HV F008 and IIV F009, management chose to modify some procedures to avoid occurrence of the problem due to lack of operating experience.

2. Following the 1991 FitzPatrick valve pressure locking event, the licensee re-evaluated all motor-operated flexible wedge and double-disc gate valves with technical consulting from an engineering firm. The evaluation determined susceptibility to the pressure locking phenomenon existed for the LPCI and LPCS injection valves (F015 A & B, F005 A & B) on both Units 1 and 2. The licensee modified the Unit 1 valves by drilling a small hole in the upstream discs. The Unit 2 valves will be modified during the next refueling outage.

3. After the discussion of the assumptions ased in the licensee's re evaluation during our visit, the licensee considered that some assumptions may not be conservative and planned to review those assumptions.

4. The nonconservative assumptions given in the licensee's evaluation of potential pressure locking due to system depressurization were: (1) pressure increase as a result of bonnet volume heatup was not considered in the calculations for the maximum pressure in the valve bonnets; (2) pressure decay rate of entrapped fluid in the bonnet due to valve leakage was based on as-found LLRT data, which is higher than that of actual conditions; (3) the available actuator thrust was based on vendor supplied data rather than actual test data.

5. Training programs for maintenance and operations personnel were provided.

However, no such training was provided for the engineering staff.

I 36

_ -