ML20151H645

Potential for Water Hammer During Restart of RHR Pumps at BWR Nuclear Power Plants, AEOD Engineering Evaluation Rept

ML20151H645
Person / Time
Site:	Browns Ferry
Issue date:	04/30/1983
From:	Lanik G, Rubin S NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD)
To:
Shared Package
ML20151H635	List: ML20151H632 ML20151H637 ML20151H645
References
REF-GTECI-A-01, REF-GTECI-PI, TASK-A-01, TASK-A-1, TASK-AE, TASK-E309, TASK-OR AEOD-E309, NUDOCS 8305040745
Download: ML20151H645 (29)

Text

m . ~ e _

UNIT: Browns Ferry Unit 1 EE N0.: AE00/E309 DOCKET NO.: 50-259 DATE: April 21, 1983 LICENSEE: TVA EVALUATOR /CCilTACT: S. Rubin NSSS:AE: GE/TVA

SUBJECT:

THE POTENTIAL FOR WATER HAMMER DURING THE RESTART OF RHR PUMPS AT BWR NUCLEAR POWER PLANTS EVENT DATE: None b

4 O I

l t

1 t

i 8305040745 830422 DR ADOCK 0500025]

ENGINEERING ' EVALUATION THE POTENTIAL FOR WATER HAMMER DURING THE RESTART OF RHR PUMPS AT BWR NUCLEAR POWER PLANTS by the Office for Analysis and Evaluation of Operational Data April 1983 Prepared by: George Lanik Stuart Rubin NOTE: This report documents results of an evaluation completed by the Office for Analysis and Evaluation of Operational Data with regard to a potential operating condition. The findings and recommendations contained in this report are provided in support of other ongoing NRC activities concerning this subject. Since the evaluation is ongoing, the report is not necessarily final, and ,the findings and recommendations do not represent the position or requirements of the responsible program. office of the Nuclear Regulatory Commission.

m . --

.)

ABSTRACT This engineering evaluation addresses a concern related to postulated scenarios of water hammer in the residual heat removal (RHR) system of a boiling water reactor (BWR) caused by void formation due to interruption of RHR pump operation while in certain containment spray or cooling modes. Browns Ferry 1 is considered as a prototype for system arrangement, response and operati.onal analyses. The results of the evalua~ tion suggest that the possibility of void formation in the RHR system and subsequent water hammer should be considered by BWR plant operations personnel for all BWRs when developing RHR system operating procedures and operator training. No equipment modifications'are recommended.

G

~

- ii -

EXECUTIVE

SUMMARY

Water hammer is a-tenn commonly used in the engineering field to describe a -

large momentary imbalance of forces developed within a piping system. The force occurs as a result of abrupt changes of fluid states (e.g., pressure, temperature and flow rates). The magnitude of the water hammer force can vary over a wide range depending primarily on the rate of change of fluid states and the mass of fluid. The effects on the piping system can range from minor pipe movements and piping support failures to damage to the piping or attachments, and, in some severe cases, pipe rupture.

2 In the design and operation of commercial nuclear power plants, water hammer is viewed as an undesirable phenomena since it could potentially affect the intended safety function of a piping system during a plant accident and/or it might be the cause of a plant accident. Although extensive consideration has been given to prevent water hammers from occurring, many have occurred during the history of commercial operation.

A recent review of water hammer events perfonned by the Office for Analysis and Evaluation of Operational Data (AE0D) indicates that there are certain additional accident scenarios or operational sequences that could cause water hamers in selected boiling water reactor (BWR) piping systems beyond those already postulat-ed or observed. Because of a concern for potentially damaging water hammer, AEOD initiated a limited engineering evaluation of the situation and identified several scenarios involving the operation of the BWR residual heat removel (RHR) system during loss of coolant accident which might impact system functional and radio-logical containment capabilities. In general, Browns Ferry 1 was used as a prototype for system design, configuration and operation.-

At Browns Ferry the RHR heat exchanger and drywell spray piping are at a relatively high elevation compared with the test return line and torus

' spray discharges. Consequently, interruption of RHR pump operation when the RHR system is in the containment spray or suppression pool ' cooling l

mode with the absence of an immediate discharge valve closure would allow draindown from the high points in the RHR system into the suppression pool.

This would result in void formation at the high points of the RRR system piping. Subsequent restart of the RHR pumps, either manually or automatically in connection with an accident, can generate a separated, moving column of water.

Such conditions are conducive to water hammer generation. The consequences of these postulated scenarios were examined with respect to the effects on RHR system functional capability for containment heat removal and system integrity for radiological containment.

The recommendations of this evaluation focused on procedure development and operator awareness. The following recommendations would apply to those plants whose p,rocedures and operator training do not already adequately address the water hammer potential due to draindown of the RHR system:

(1) Procedure development and operator training should specifically address the potential for water hammer during operation of the RHR system in the containment spray and in the suppression pool cooling modes.

(2) Operating procedures and operator training should address methods to minimize the extent of water hammer damage, e.g., the desirability of

- - ~ -

iii -

operating only a single train of RHR in containment spray mode at any given time and the desirability of keeping any train cross-connect valves closed.

(3) Consideration should be given to the development of emergency startup procedures for restart of RHR system pumps following possible draindown conditions when-a need for rapid response or equipment accessibility considerations precludes use of the usual fill and vent operations.

AEOD does not believe that any immediate actions are requi/ed at this time beyond dissemination of the information and recommendations in this evaluation and a careful examiration of this concern by each licensee on a plant-specific basis. No hardware changes are recommended at this time pending the results of the licensee's examination.

I h 9 e

-~ - - -

1.0 INTRODUCTION

In recent years, licensees of operating reactors have reported a large r. umber of water hammer avents during commercial operation. Although most of these -

events resulted in little or no damage to the piping system pressure boundary integrity, there frequently has been damag'e to the pipe suppcrts and attach-ments. As a result, the NRC initiated a task activity to resolve the concerns associated with water hanmer. Objectives of the review were to identify the causes of water hammer events and, further, to recommend actions needed to reduce the likelihood of such events. The earliest results af this review were published in NUREG-0582 (Ref.1). More recent studies have been presented in NUREG/CR-2059 (Ref. 2) and NUREG/CR-2781 (Ref. 3). The latter report proposes recommendations which are intended to resolve the water hammer issue.

Because of the frequency of water hammer events during plant operations and their potential impact-on the safety system operation, AE0D initiated its own limited review of the situation. The primary objective has been to <

assess certain accident scenarios that could result in conditions conducive to water hammer which, if allowed to occur, may impair the continued operation of the needed safety system. This effort supplements the aforementioned water hammer task activity which is primarily concerned with water hammer damage that has occurred and might continue to occur as a result of historical plant operational or transient sequences.

This report is limited to the results of an evaluation of possible scenarios that could cause water hammer in the BWR RHR system when-the system is aligned in the containment spray mode or the suppression pool cooling mode.

It should be noted that the particular scenarios dercribed in this report have not as yet been identified specifically in the reported water hammer experience of operating BWRs. However, other analogous water hammer events associated with the RHR drywell spray and torus spray lines and the test return lines have occurred (as reported in LERs) which form a basis for concern regarding the scenarios presented in this report. Other studies of water hammer events at BWRs (Refs.1, 2, 3 and 4) provide a broad review of scenarios that have resulte.d in reported water hammer events. We have made no effort to follow up on this work. This report is intended to simply supplement the collection of operationally based water hammer experience, already recognized I. and being addressed, with possible water hammer scenarios arising from emergency or accident conditions.

The Browns Ferry Nuclear Plant Unit No.1 was used as a prototype for this study. Although the Browns Ferry plant is used for purposes of analysis, we I believe that the concern raised may apply to most BWR facilities with multi-functional RHR systems. However, specific arrangements and values cited here with respect to pipe sizes and elevations, valve closure times, and diesel load sequencing times, etc., may be different for other plants.

i 2.0 RHR SYSTEM DESCRIPTION A simplified piping and instrumentation diagram of one train of the Browns Ferry Unit 1 RHR system is shown in Figure 1. The major equipment for the entire system consists of four heat exchangers and four RHR pumps, separated into two independent trains. The heat exchanger discharges may be cross-connected by opening a normally closed valve. Each train is capable of

~

performing all required RHR functions.

,-,,,----v -

-,,,m. --,-y-..,y--,--.,,-- - -

. m - "

The RHR functions'are:

(1) To restore and maintain the reactor vessel coolant inventory so that the core is adequately cooled to prevent excessive fuel clad temperatures after a loss of coolant accident (LOCA).

(2) To limit the suppression pool water temperature during HPCI or RCIC

. operation (hot standby condition) so that if a blowdown should occur, the temperature of the suppression pool water will not exceed that temperature necessary to maintain its primary role as the quenching medium of the pressure suppression containment system.

(3) To remove decay heat and' sensible heat from the reactor coolant system so that the reactor can be shut down for normal refueling and servicing operations, and to provide additional cooling capacity for the fuel pool cooling system when not required to function as an emergency core coolir.g system.

(4) To provide for primary containment spray cooling to limit containment pressure.

(5) To provide for full flow testing to demonstrate the availability of the system to fulfill the above ft;nctions.

(6) To provide an inexhaustible source of makeup for vessel and containment flooding after a postulated design basis LOCA. (This capability assures defueling capability after such an accident.)

To accomplish these functions, the RHR system has the following operating modes and features.

(1) Low Pressure Coolant Injection (LPCI)

(2) Containment Spray / Cooling (a) Drywell Spray (b) Torus Spray (c) Suppression Pool Cooling .

(3) Snutdown Cooling / Head Spray Other Features:

(4) Fuel Pool Cooling Augmentation

'(5) Standby Coolant Supply Cur concern in this study is focused on the LPCI mode and the containment spray cooling modes of operation. No further discussion of the other modes will be preser.ted.

2.1 Low Pressure Coolant Injection (LPCI) Subsystem The LPCI subsystem is an integral part of RHR and is the dominant mode and normal valve lineup configuration of RHR. LPCI operates to restore and, if necessary, maintain th,e coolant inventory in the reactor vessel after a LOCA

- ~

i so that the core is sufficiently cooled to preclude fuel clad temperatures in '

excess of 2200*F and subsequent energy ' release due to a metal-water reaction.

The LPCI subsystem operates in conjunction with the high pressure coolant

. injection (HPCI) system, the automatic depressurization system (ADS), and the core spray (CS) system to achieve this goal.

4 During LPCI operation, the RHR pumps take suction from the suppression pool and discharge into the core region of the reactor vessel via each of-the two recirculation loops. Spillage from any break in a reactor coolant system line within the primary containment drywell returns to the suppression pool through the pressure suppression vent lines. A minimum flow bypass to the suppression pool is provided so that the RHR pumps are not damaged if operated with the discharge valves closed.

Power for the RHR pumps and the RHR service water pumps comes from the 4 kV AC power shutdown boards. Power for these boards normally comes from the auxiliary power sources but if these sources are not available, power is provided by the standby (diesel) AC power source.

2.2 Containment Spray / Cooling Subsystem The containment spray / cooling subsystem is an integral part of RHR. The 1 cooling function limits the temperature of the water in the suppression pool so that even if the design basis LOCA has occurred, the pool temperature i

will not exceed 170*F. The selection of 170*F is based on tests which show that at this temperature complete condensation of blowdown steam from the design basis LOCA can be expected. .

.With the containment spray / cooling subsystem in the suppression pool cooling mode of operation, the RHR system pumps are aligned to pump water from the suppression pool up through the RHR heat exchangers where cooling takes place by transferring heat to the RHR service water. The flow returns

down to the suppression pool via the test return line. This alignment is 4 also used to cool the suppression pool during nomal operations when suppression pool temperature may rise due to RCIC, HPCI, safety / relief i valve (SRV) discharges, or high ambient temperatures.

The containment spray / cooling subsystem also provides drywell and torus spray for post-accident conditions. The RHR system water pumped up through the RHR heat exchanger may be diverted to spray headers in the drywell and/or the torus. The spray headers in the drywell could be used to condense steam in the drywell thereby lowering containment pressure. The spray and condensed steam would collect in the bottom of the drywell until the water level rises to the level of the pressure suppression vent lines where it then overflows to -

I the suppression pool . Water flowing there would then be pumped back through

. the heat exchangers. In addition, approximately 5% of RHR system flow may i be diverted directly to ',he torys spray ring header to cool any noncondensable i gases collected in the free volume above the suppression pool. These spray headers cannot be placed in operation unless the core cooling requirements of

the low pressure coolant injection (LPCI) subsystem have been satisfied.

These requirements may be manually bypassed under certain conditions.

l

The standby alignment of the two train (redundant) RHR system during normal power operation is the LPCI mode. In this mode the LPCI injection valves are closed and the RHR pumps are idle. Upon receipt of the LPCI initiation signal, the pumps' start and the injection valves are opened when reactor pressure has dropped to less than 450 psig. This allows LPCI water to flow directly into the recirculation loop discharge piping as soon as system pressure drops below the maximum pump head pressure.

7 At times during nonnal power operation, either or both trains of the RHR system may be aligned in the suppression pool cooling mode via the full flow test return lines. This could be done for purposes of running tests, for sampling, or to reduce high suppression pool temperature which might approach the technical specification limit of 95'F due to elevated ambient temperatures.

If an LPCI initiation signal is received while the RHR system is aligned in the suppression pool cooling mcde, the test return line valves automatically t

close and the system is realigned to the LPCI mode in preparation for LPCI injecticn. The stroke time of the test return line valves is 24 seconds which is sufficiently fast to provide the required realignment by the time LPCI flow is required for core flooding.

For the containment spray mode of the RHR system, the RHR pumps would be aligned to pump water from the suppression pool up through the RHR heat exchangers to the drywell free space via the drywell spray sparger and/or to the torus free space via the torus, spray-header.

Suppression pool cooling is eventually required following a LOCA to remove core decay heat deposited there. However, containment spray would only be required following a LOCA which leads to a delayed pressurization of the containment which approached the design limits. At Browns Ferry, containment i spray is required by procedure to be in operation when torus spray is in operation. Suppression pool cooling is required when steam discharges from RCIC, HPCI, SRVs, or a high ambient temperature raises the suppression pool temperature to near 95'F.

Although it is possible to align the RHR system with the test return line open to the torus while the containment spray is being used, ~ operators at Browns Ferry are trained not to use containment spray and suppression pool l cooling from the same RHR train. For purposes of analysis, this study addresses only those operating modes expected if the Browns Ferry procedures are used.

A " keep full" system is installed at the Browns Ferry plant to maintain the RHR piping filled with water from the pump discharge check valves to the

. outer (normally closed) valve in the low pressure cool ant injection, containment spray and vessel head spray flow paths. This is needed to make up for

' valve leakage or inadvertent draining and thereby prevent a possible water hammer which might occur on normal system initiation. The LPCI injection lines are normally pressurized to 48 psig by the keep full system. The pump discharge check valves prevent back flow through the pump suction lines to the suppression pool. The keep full system incorporates two pumps (capacity 40 gpa at 152 ft TDH/ pump) which takes a suction from the suppression pool through two normally open air-operated containment isolation valves.

The pumps discharge to a head tank which supplies the needed makeup water through connecting 2" piping to the reactor pressure vessel head spray and

_ _____ _ _ _ _ _ _ _ _ _ __-_l_____ _ _ _ _ _ ___ .

containment spray lines. The connections to the RHR pipe have double check valves to prevent reverse flow to the head tank during RHR system operation. -

This system maintains the RHR pipes full while in a nonnal standby condition.

However, the small capacity of the keep full system is not adequate to restore or maintain the RHR system completely filled if certain large valves associated with the drywell spray, torus spray or test return lines are open. The keep full system at Browns Ferry is not considered to be safety grade. If the keep full system is unable to maintain adequate pressure, a backup from the condensate transfer system can be manually valved into service.

3.0 POTENTIAL WATER HAMMER EVENT SCENARIOS A review of the Browns Ferry RHR system equipment arrangement reveals that if RHR pump operation is interrupted without discharge valve closure during some emergency containment cooling modes, draindown to the suppression pool will occur in a portion of the system piping. This potential for drainage can be seen from Figure 2, which is a schematic of the principal piping of one train of the RHR system with the elevations shown.

3.1 RHR Pump Trip with Containment Spray Valves Open If attendant to a pipe break within the drywell, the torus spray is operated concurrently with the drywell spray, the drywell spray header is open to the drywell at location A in Figure 3 and the torus spray is open to the torus at location B. With the RHR pumps running, there would be sufficient water pressure in the piping upstream of the spray valves to maintain flow into both the drywell and into the torus. However, if at some point following the accident, RHR pump operation was interrupted for any reason, there would no longer be sufficient water pressure in the piping to maintain these fl ows. In this situation, an open and vented flow path would develop from location A to location B for water in the containment spray piping to drain into the suppression pool and for drywell air to enter the piping. Air pockets would also form in other sections of the system piping which are located at elevations above the suppression pool due to percolation of the torus air through the piping. Alternatively, if percolation cannot occur, a water vapor pocket will form in the vacuum such that any water column is limited to about l 33 feet of vertical height at room temperature. If the pumps are subsequently restarted without first slowly refilling and venting the voided sections of -

piping, a water hammer might occur.

If the RHR pumps remain inoperable, a final steady state condition could develop with voids forming at the locations of the darkened piping shown in Figure 4.

Those piping sections which are open to drywell free space at the high end and open to torus free space at the low end should be expected to empty very l

rapidly ,and almost completely. Those piping sections which are not open at a high point but which are elevated more than 33 feet (the head of water at room temperature that can be held up by one atmosphere pressure) above the elevation of piping low point which can act as a trap against percolation i of air into the elevated section of piping will quickly form vacuun voids.

l Figures 5(a) and 5(b) portray the voiding locations in the RHR system. If air percolation or inleakage is possible, the entire pipe will eventually drain. To enter the drain down scenario, both pumps (or one, if only one is operating) within a given train of RHR would have to trip with the spray valves open. If one pump stays operating during two pump operation, sufficient pressure would be maintained to keep all of the piping filled.

3.2 RHR Pump Tri'p with Test Return Line Discharge Valve Open Water hammer might also result due to a similar draindown scenario if RHR pump operation is interrupted when the RHR system is aligned for suppression pool cooling and the test return line discharge valve remains open. This mode is illustrated in Figure 6. The test return line is open but submerged at location C. This provides a flow path for fluid in the system to the

! draindown .into the suppression pool. Again, as long as either RHR pump i remains running, sufficient pressure is maintained in the piping upstream of the discharge valve to prevent voids from forming. However, .if RHR pump

operation is interrupted while the test return line discharge valve is open, the piping could begin to drain, foming a vacuta void at the highest j elevations. If the pumps remain off, the column of water in the RHR piping i will fall to the level that can be supported by the pressure difference between the pressure applied on the suppression pool surface and the saturation pressure applicable to the vertical column. Tne final state for a water column at 90'F and the containment pressurized to atmospheric pressure is shown in Figure 7. Here the column of water supported by atmospheric pressure is approximately 32.5 feet high. A higher torus pressure would lessen the amount of voiding. A different fluid temperature, such as at i the heat exchanger discharge, would also result in a different final steady i state elevation, as would any noncondensible gases present in the fluid i liberated when the RHR pumps trip.

t Suppression pool cooling may be required both during nomal operations and following an accident. If the pumps trip during normal operation while aligned in the suppression pool mode, sufficient time is available with easily accessibic system components to allow operator actions to fill and vent the system prior to restart of the RHR pumps. In fact, the keep full system should automatically act to refill the piping if the test return line valve is closed and sufficient time is available.

i Suppression pool cooling is required following an accident condition to remove decay heat. In this case, if a pump trip were to occur without discharge valve closure, it is likely that sufficient time would be avail-able for operator action to fill and vent the RHR system prior to RHR pump.

restart but radiological considerations might prevent access to areas inside i the reactor building where equipment to be manipulated is located. If operable, I the keep full system could accomplish the refill. On.the other hand, it is likely that only a single train of RHR would have been realigned in suppression pool cooling _ and the other train would be unaffected.

A scenario of concern with respect to the suppression pool cooling mode j is when one or both RHR trains are aligned in this mode due to high l

ambient temperature conditions at the time of accident initiation. If the t

accident results in RHR pump trip, piping draindown could occur. However, i as distinct from interruption of pump operation during non-accident conditions l where time and accessibility are favorable for corrective actions, if pump l operation were interrupted (such as caused by a loss of offsite power) during

[

accident conditions which generate a LPCI start signal, the pumps would automatically restart as soon as the diesel generators reached operating i vol tage. This would occur about ten seconds after the loss of offsite power.

BWRs with system and procedural differences may be subject to other scenarios.

For' example, operation of torus spray only would result in voiding due to percolation effects. ,0peration of drywell spray only could result in only i

L -

i a small drainage. A plant with a test return line discharge above the pool surface would be subject to percolation. effects. If a plant were operating ,

4 with the RHR trains cross-connected, draindown of both trains could result I due to the loss of pumps in only one train (if only one train's pumps had l been in operation).

, 4.0 ANALYSISOFWATERjiAMMERPOTENTIAL 4.1 Containment Spray' Mode l 4

RHR system operation in the containment spray mode at Browns Ferry Unit No.1 is unlikely except for conditions outside the design basis. The safety

analysis for Browns Ferry shows that containment spray is not needed following any postulated event. Expected operator actions in response to large

{ leaks in the reactor coolant system inside containment could include efforts to rapidly depressurize and shutdown the reactor before containment pressure  !

j or temperature reaches a level which might require containment spray.

3 This can be accomplished with HPCI and/or ADS. However, in a case where the reactor is not rapidly depressurized, a small break might continue to add energy .to the containment atmosphere and slowly pressurize containment. The Browns Ferry emergency procedures require the operator to employ the drywell j spray if the containment pressure reaches 35 psig or the drywell temperature is above 280*F 30 minutes after the incident began. It can be concluded that

there is a low probability that the drywell spray will be used, even in the

case of a small break LOCA. To date, at Browns Ferry, these systems have i only been used during pre-operational testing. The unexpected need for drywell or torus spray may not be the case for other BWRs.  ;

j It is also noted that during normal operations at Browns Ferry, suppression

-pool cooling is always accomplished using the RHR test return line. To

! date, only rarely has an SRY (or RCIC/HPCI) exhaust discharge resulted in the i use of the torus spray to lower torus pressure. Nomally, cooling of the torus

water when it has been needed has been accomplished by using one, two, three, or four RHR pumps discharging through one or both test return lines, depending on the heat load and on the service water temperature. The largest flows are required to maintain suppression pool temperature below 90*F when ambient '

temperatures are high and service water temperature is near 90*F. Although

! there is a cross connect between the two RHR trains, this is nomally closed and would have no effect on the pump trip scenario.

4.2 Suppression Pool Cooling Mode The suppression pool cooling mode of RHR is used for a variety of purposes such as: 1) cooling the suppression pool during high ambient temperature i conditions; 2) cooling the suppression pool following a SRV discharge or I

following HPCI or RCIC operation; 3) sampling of the suppression pool; 4) full flow testing ~ of the RHR ECCS function; and 5) cooling the suppression pool i following a LOCA. It is not unusual for the system to be aligned in suppression l pool cooling mode. It is reasonable to expect that at least one train of the l RHR system could be in suppression pool cooling operation (with the reactor I pressurized) on an average of several hours per month. If a loss-of-coolant accident with a loss of offsite power were to occur at such a time, partial voiding of the RHR system would be expected to occur. Subsequent restart of RHR pumps with a void in the piping might cause a water hammer. If the loss of ,

offsite power occurred sometime after the LOCA, when the system was aligned in

~ . , - . , - -. ~ - . ~ . .~.----__ -_..-m, . . . - , , - . ---.~.--__-.,e ........,#,-----,---.m__--,- .-,-,_-...---er--- wm---.-%*

<-- - ~ ~

.- ~ -

i i ~ either suppression pool cooling mode or contdnment spray mode (for heat removal following the accident), draindown may also occur. However, as discussed later in Section 4.3, if the diesels had been left in operation, RHR pump ,

operation would not be interrupted long enough to cause significant draindo m.

Also, if the LPCI actuation signals were cleared, no water hammer would result because the RHR would not automatically restart. The operator could then take normal precautions to fill and vent the system prior to continued RHR operation but only if the reactor building remains accessible following the accident.

Browns Ferry procedures'do contain general precautions to assure that the RHR system remains filled and vented while in the standby ready condition. However, the current procedures do not specifically address the unusual circumstances

presented by the scenarios discussed in'this report.

4.3 Loss of RHR Pump Motor Power l An important potential cause of the postulated water hammer scenario is the

, loss of offsite power while the:RHR system is in the containment cooling i mode. To assess the severity of the event requires an examination of the sequence cf occurrences in the auxiliary power supply during a loss of offsite power. Again, for purposes of analysis, Browns Ferry auxiliary power system was used as a model. Two scenarios were considered: (1) loss of offsite power some time after a LOCA when the drywell and torus spray

, are being used and the rHR pumps are powered by the offsite power source, 1

and (2) loss of offsite power concurrent with a LOCA during a time when the RHR system is aligned in the suppression pool cooling mode.

The containment spray mode of RHR would' only be used in response to an accident which sufficiently pressurizes primary containment. As discussed 4 earlier in Section 4.1, RHR system operation in this mode would only be considered at least 30 minutes after the start of the accident. For purposes of analysis, a sequence of occurrences could be described by the following

- scenario. At the start of the accident (such as a small break LOCA which began to pressurize the containment),- the diesel generators (DGs) would receive a signal to start on high containment pressure. The operators would likely be most concerned with maintaining reactor water level, verifying that emergency

. equipment was functioning as required, and possibly attempting to locate and 1

isolate the break. Operators at Browns Ferry are instructed to leave emergency equipment running until plant conditions are stabilized. For the postulated.

scenario, it is likely that the diesels would be left running throughout i the mitigation phase of the event even if offsite power remained available.

The RHR pumps at Browns Ferry are supplied from 4160V shutdown boards.

Each of the four pumps is powered from a separate board. Normal power to the shutdown boards is from unit station service transformers. However, following a reactor trip and loss of offsite power, each shutdown board j would be ' powered by a separate diesel generator.

The sequence of occurrencs for a loss of normal power to the shutdown boards is discussed below. If the voltage to the shutdown boards drops 10%

i below normal, the diesel generator automatically starts. According to technical specifications, diesel generators are required to be ready to accept load within ten seconds or less after receiving an automatic start signal.

Industry average is about seven seconds. When the diesel generator reaches rated speed and voltage, and if voltage is sufficiently low or lost to the shutdown board, the diesel generator is automatically connected to the shutdown board. Thus, if the voltage on the shutdown board drops 30-40%, or if the l

1

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

voltage remains below 90% of the normal voltage for six seconds, the normal feeder breaker opens, all loads are stripped from the shutdown bus (es) and the '

feeder breaker from the DG closes. Once voltage is restored to the shutdown board by the diesel generator, the automatic load sequencer begins loading emergency equipment. The RHR pump is the first load connected by the automatic sequencer. It is loaded at time zero on the sequencer. From this sequence, it may be seen that the 'oss of offsite power would result in only a short interruption of power to an RHR pump.

With regard to the potential situation where the RHR system is aligned in containment spray mode with the RHR pumps running and powered by offsite power, it is likely that with the diesels left running, the interruption in power would be so brief that no voids would fonn in the RHR system piping.

If the diesels were returned to standby, the situation would be slightly different and some voiding would probably occur. An estimate of the RHR pump coast down time from Browns Ferry operations personnel is about three seconds.

As stated previously, the average expected start time for the diesels is approximately seven seconds. With the coast down flow characteristics of the RHR pumps and the timing of the power restoration described above, pump and flow inertia may not provide enough flow to prevent limited drainage of the RHR system piping, at least mcmentarily. Given the pump coast down time of three seconds and diesel generator loading time of approximately seven seconds, an estimate of the time during which voiding could occur is from four seconds to a maximum of seven seconds. In the absence of actual testing, the amount of voiding that would occur under such circumstances cannot be known with great certainty.

Another case of interest is a LOCA coincident with a loss of offsite power during nonnal operation at a time when the RHR system is aligned in the suppression pool cooling mode. This case is similar to the situation

. described above where power to the RHR pump is interrupted for seven seconds and, due to the effects of pump coastdown and loading times, RHR flow is interrupted for a maximum time of about seven seconds. Again, the amount of voiding is uncertain without testing.

4.4 Draindown Rate It is difficult to estinate the amount of voiding if RHR pump operation were to be briefly interrupted. The rate of void fonnation would depend upon such factors as fluid temperature, pipe flow characteristics, pump coastdown rate and torus pressure. It is expected that for cases of immediate pump restart, the extent of the voiding would not be as great as that previously described in Section 3.1 of this report. In that discussion air percolation and draindown continued to a final steady state condition and rapid RHR pump restart was not considered.

For the case of a rapid pump restart, the large uncertainty in the size of the void that would be formed is due to the inertial effects including flow coast-down and startup characteristic,s of the RHR pump. As described in Section 4.3, the maximum time during which insufficient RHR pump flow would be available to maintain the pipes full in a loss of offsite power is about seven seconds.

Also to be considered are the kHR heat exchangers at Browns Ferry which have a shell approximately five feet in diameter. The outlet nozzle is located on the side some distance below the top head. This situation provides a reservoir of

s

?

water at a high point in the system which will drain intb the piping to delay voiding in the piping. As a result high point voiding could be confined within j the top of the vertical RHR heat exchangers. .

For the case while in the containment spray mode, and spraying Lbth the l drywell and torus, the draindown could be more pronounced, s The voids would start to develop _ at the high point " vent" location 1'.c., at the containment spray nozzles. For this case, the pipe voiding would first occur in the highest elevations of the drywell spray line. The piping at -i the highest elevation is 12" in diameter. Again it is dffficult to determine l with certainty what piping would be vo~ided because thereswould also be a j draindown of fluid from the head spray piping, etc. ,N / j; The effects of elevated containment pressure would have little effect^on l the draindown characteristics for the loss of pump flow while in the n 'I containment spray mode. Howev.er, the affects of torus pressure on the. case of the suppression pool cooling mode where the piping discharges to the-pool [ ^ ,l under water would be important. For example, an elevated torus pressure could -

be sufficient to completely counteract the driving head due to water elevation at the RHR heat exchangers. Torus pressurization will tend to retard drain ~down for all cases where the RHR system is in the suppression pool cooling mode. -

It might be possible to show that for the complete LOCA break spectrum, " ' ,

no break results in a combination of containment pressure and reactor de; ~

pressurization that would cause significant RHR pipe" voiding at the time of ,

RHR pump restart. This mitigation effect would not apply if the RHR water discharges above the pool surface whil'e in suppression pool cooling mode.

u 4.5 Water Hammer Magnitude and Location Two aspects of water hammer which have bearing on the safety significance of ..-

the postulated scenarios are the magnitude of the water hammer and the location d of the water hammer. '!ith respect to the magnitude of the water hammer generated, some general comments can be made without detiiled a' n alysis of the mechanics of water hammer. First, the draindown of the,RHR system sile in the containment spray mode can result in a larger voldme (longer run of yoided pipe) than would be expected for the system in sunpression' pool coolinbode.

Thus, many more valves, tees, elbows and containment penetrat' ions would be potentially susceptible to the effects of a column of water' moving into the .

voided piping. Consequently, water hammer damage might be.more likely if the RHR system was operating in the containment spray mo~de than~if the RHR system was in suppression pool cooling mode. Also, when the RHR system is operating - a in suppression pool cooling mode, the high points of the system piping involve mostly horizontal pipe runs. Piping in such horizontal pipe runs would likely  ;

cause a less severe water hammer than voiding located ,in vertical piping runs.

With respect to location, the concern relates to whait parts 6f the system would most likely be damaged. Again, for the case when the RHR system is in -

containment spray mode, the potential for more voids results insthe potential -

for more valves and instruments to be affected. A locaticn where a large water hammer load might occur (if the postulated event occurred while the, RHR system is in containment spray mode) and be of significant concern is s' - .

near the point where the containment spray line penetratdsithe drywell. At -

this point, at Browns Ferry, the containment spray line completes a vertical rise, bends around with some elbows, passes through two containment isolation -

valves; and tees off to the drywell sprays. These tr'ansitions occur in a pipe '

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

c -

u -

~n-

y. - -

.- -=-

f- _ -

11 .

n ~

~ ,

r Y  %- g A ' run of 4 proximately 25 feet. A failure that might be.of concern at this location is the failure of_ the primary containment penetration due to movement

~

I -

of the drywell spray Une;because of water hammer. -

'e-If it is true that damage to a vent, drain or instrum'ent line or interference G, '

- with valve operation 'is more likely-than a large pipe"ri.pture. then the m iocation of the water hammer effects aFe especially important . If it is

'~

c assumed that the water hammer effects are most intense 'near tne location of the 7 _ original voided regi%more general cosme'ntslcan be made. As the pumps restart and repressGr15e the system, the vcid size is reduced as it is carried

^

j L' through the piping. ..Most of the valves controlling operation of the RHR system are located at the enas of the piping runs in the torus or near the containa. ant spray injection poihts. Thus, it.is likely that these locations would not be affected as severely as for the case where the RHR systein1s in suppression 7

pool coofing mode because the original voiding is near the RHR heat exchangers.

It. is also noted that, for Browns Ferry, for the case where the RHR system is in:the suppression pool cooling mode, the original location would not appear to be in close proximity to vent, drain or instrucent lines. Thus, for the postulized water hammer to damage-such components, any water hammer induced

, motion = or pressure forces would have to be transmitted along the piping or generated at a _ location remote from the original void location. Although the i piping ~ system restraints and snubbers may not be designed to withstand such

water hammer forces without damage, it is likely that over some distance, the

~

However, it is

> effects recognized-tnat the location of water hammer effects and the ir propagation is of the postulated water hammer would be attenuated; difficult to predict. ,

4.6 Safety Significance  ;

To evaluate'the safety significance of the postulated wnter hammer events in l . the RHR system, it is necessary to consider two separate aspects: (1) system l

functional capability with respect to core and containment cooling, and (2) system integrity with respect to radiological releases." "

/

The RHR system is d'esigned with redandancy in that its safety functions are

, provided by two' separate trains of piping, humps, heat exchangers, valves, etc.

! Thus, the loss of one' train would not result in the loss of any safety function.

Forcexample in the postulated scenario of'draindown of the piping while the RHR system is in the containment spray mode, it is likely that only one train of.RHR would be operating in the containment' spray mode at any given time. The

, other train would likely be operating in either the_ suppression pool cooling N mode or in standby. Also, the reason for using~ containment spray is because the containment is pressurized. Thus, in this situation, tripping of all RHR pumps would result in'draindown of the RHR train which is in the containment spray mode but the containment pressure, wou1A likely prevent voiding in the RHR piping which is aligned in' the suppressionfpool cooling.

There are circumstances where the postulated draindown scenario with resultant E. water hammer could effect both trains simultaneously: Consider the case where the plant operators find it necessary during normal power generation conditions to operate both trains of RHR in the suppression pool cooling due to high ambient temperatures. If during this period the plant experienced a small -

break LOCA with coincident loss of offsite power, and if containment was not rapidly pressurized, both RHR trains would experience draindown. Diesel

_ generator start and ' automatic loading of the RHR pumps would compress and move -

4

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

the voids through'the piping system. Thus, in this example, a water hammer might occur in both RHR trains.

Operating experience would indicate that a water hammer which occurs in a low pressure piping system will likely not cause the rupture of a large diameter pipe. It would appear more likely that this type of water hammer event could generate breaks in attached drain, vent or instrument 1.ine piping. Loss of certain instrumentation could result in control problems during post accident operation of the RHR system. The loss of RHR water through even small breaks could cause operational problems due to the resulting adverse environment.

There are circumstances which could reduce or eliminate the safety concern associated with this common mode mechanism. First of all, the size of the LOCA must be considered. If the LOCA is large enough to quickly pressurize the containment atmosphere while in the suppression cooling mode, little voiding would occur before tne RHR pumps restart. In any case, containment pressurization of a lessor magnitude would tend to reduce the size of the void and the risk of damaging water hammer.

Also, with respect to break size, LOCA analyses show that the LPCI mode is not required for a small break LOCA below a certain size. For these situations, HPCI alor.e or HPCI with ADS and core spray would be capable of maintaining peak clad temperatures below the regulatory limits. However, these accidents would require that the RHR system be used some time after the accident for suppression pool cooling and decay heat removal. . At this point, RHR system operability to perfonn that function would depend on the character of any water hammer damage and whether or not the reactor building was accessible and the damage was repairable.

With respect to system pressure boundary integrity and radiological release, j two separate concerns would be of interest: (1) the releases due to failures in RHR system piping and components, and (2) the releases due to possible induced failure of containment integrity. A failure in the RHR system piping instrument lines outside primary containment could be isolated as soon as detected such that transport of radioactive water would be limited to that which was in the RHR piping. Depending on the severity of core damage, this could be negligible or fairly significant. It is likely that this limited leakage would not generate plant releases in excess of 10 CFR 100 limitations l if detected and corrected in a timely manner. However, if piping interaction with primary containment resulted in loss of containment integrity, serious high level releases could be generated.

4 .'7 Event Sequence Considerations The risk -from the postulated water hammer scenarios depends, in part, on the likelihood of the individual events which make up the scenario. Without assigning probabilities to the terms of the analysis, it is useful to describe the sequence of occurrences leading to the consequences identified. For purposes '

i of illustration, two specific event sequences are listed: (1) a LOCA which leads to the need for containment spray which results in a loss of containment integrity, and (2) a LOCA which leads to RHR system piping or instrument line failures resulting in the loss of decay heat removal capability.

Sequence for Scenario No. 1 (1). LOCA ,

._ . _ _ _ _ _ _ , _ a

(2) Core damage with large releases to containment and delayed pressurization of containment due to failure to remain within . <

response envelope predicted for design basis events leading to a need for containment spray. Alternatively, containment spray is initiated by the operator to reduce the containment leakage rate for a LOCA which pressurizes containment below the required spray initiation pressure.

(3) Tripping of RHR pumps in the RHR train which is aligned in

. containment spray mode.

(4) Restart of RHR pumps without being able to take actions to assure void free piping.

(5) Generation of a high force water hammer.

(6) Failure of containment spray penetration due to water hammer forces.

(7) Release of significant amounts of radioactivity to the reactor building environs.

Sequence for Scenario No. 2 (1) High ambient temperature conditions during normal power operation which require operation of both trains of RHR for suppression pool cooling.

(2) Small break LOCA with coincident loss of offsite power results in partial voiding of RHR system with subsequent automatic restart of RHR pumps.

(3) Generation of a severe water hammer in both trains of RHR.

(4) Failure of instrument lines in both trains of RHR which disables instrumentation needed for control.

(5) Failure to take manual corrective actions resulting in inability to remove decay heat.

These are only two of many possible sequences. To determine the total risk due to the general concern of water hammer in the RHR system caused by draindown following a pump trip, a complete analysis of the probabilities and consequences of each of the of possible event sequences would have to be performed.

5.0 OPE' RATING EXPERIENCE A review of information on water hanmer events as reported in LERs indicates that water hammer has occurred during operation of the RHR system in the containment spray and suppression pool cooling modes. It should be noted that the damage to containment spray piping occurred during pre-operational testing. Below are some of the events reported by the licensees.

v ,,,

~

U V Events specifica'lly identified as water hammer in containment spray mode of RHR:

FitzPatrick 1 - 3/21/75; 5/24/75; 6/17/75 Dresden 3 - 10/5/79 Events specifically identified as water hammer in the suppression pool cooling mode of RHR:

Brunswick - 4/13/77 There were many LERs reporting " water hammer" in the RHR system which do not specify the operating mode. It is likely that at least some of these events occurred while the RHR system was in either the suppression pool cooling or containment spray modes. It is not clear that any of these water hammers involved partially drained piping.

6.0 CONCLUSION

S AND RECOMMENDATIONS Because of the high elevation of RHR heat exchangers and containment spray piping and the low elevation of the test return line and torus spray discharges, interruption of RHR pump operation when the RHR system is aligned in the containment spray or suppression pool cooling modes allows draindown from the high points into the suppression pool causing high point voiding of the piping.

The subsequent restart of the RHR pumps, either manually or automatically, can generate a voided moving water column which is conducive to water hammer generation.

The consequences of several postulated scenarios were examined with respect to the effects on system functional capability for containment heat removal and system integrity for radiological containment. There appears to be a basis for concern that should first be addressed by operator awareness and procedure. The following recommendations are made to apply to those plants whose procedure development and operator training do not already adequately address the water hammer potential due to draindown of the RHR system.

(1) Procedures and operator training should specifically address the potential for water hammer during operation of the RHR system in the containment spray and in the suppression pool cooling modes.

(2) Procedures and operator training should address the desir-ability of operating only one train of RHR at a given time in the containment spray mode and the desirability of keeping any train cross-connect valves closed.

(3) Consideration should be given to developing emergency proce-dures for restarting the RHR system pumps following a drain-down when the need for rapid response or equipment acessibility considerations precludes use of nomal fill and vent operations.

AE0D does not believe that any immediate actions are required at this time beyond dissemination of the infomation and recommendations in this study l

and a very careful examination of this concern by each licensee on a plant-l

15 -

specific basis. No hardware changes are recommended at this time pending the results of licensee examination. ,

It should be noted again that the analysis above uses the Browns Ferry plant as a prototype. Other plants may have specific RHR piping configurations which either increases or decrease the potential for or magnitude of water hammer ,

probl ems. Also, different diesel loading characteristics and timing could '

affect the s:enarios. Operating proceduret, at einer plants could lead to different scenarios than those discussed in this report. Such items would have to be reviewed on a plant-specific basis. ,

h l

l I

0 9

- 16 -

7.0 REFERENCES

1. " Water Hammer in Nuclear Power Plants," USNRC NUREG-0582, dated July 1979.

2. " Compilation of Data Concerning Known and Suspected Water Hammer Events in Nuclear Power Plants," USNRC NUREG/CR-2059, dated May 1982. 9

3. " Evaluation of Water Hammer Events in Light Water Reactor Plants,"

USNRC NUREG/CR-2781, dated July 1982.

4. " Review and Evaluation of Potential Water Hammer Events Nuclear Plants" by R. L. Chapman, 0. M. Hammer, Jr., and M. E. Wells, EG8G Idaho, CAAP Report dated September 1979 (Rev. 1).

t .

f e

t it ti ictn ca.c: :p:t es.c s Da ,- @CG i.l r

n is

. iss.

z rx1 :

!!*I' :

e -

-O"

D : @Q'ilil

i

s.  := ,

?\

i. .g -

• b.

I i

E PlL,P

-i g

e .. , .i a i ,,r.g -- gi d.

.:  :' . m j

d 1 e e k '

5 c .s l.K' , .

Q.

C E 0 _.: j b

I __ r-_ m l .-5 m.

A E

.- g , - -

E o ,,

l hY jf E

• fi l-* m Tf]; '. -

b  !

2-x ww E - .  := r i c +T-E i 1 c

e

-(,.

{ (' e i.

r' (%J l

..g I ,,,

.h. N EH;

,- i' id il I e H

EK g- / - <

g4 l!* g- ,

,_ .! @, M? f EN ErX $

4" *

$h ,8 e

1 i

r s'

e Mif  !!i e

56:i!,

s y

cb i 9 !ss ke rT :-9 4

A

)

,k

+ 3 4

?!i e-tis; er .

~lt

p Ei!

. m -

e

J k Head Spray Z

EL - 629' EL - 589' -

Containment Spray ~

RHRS EL - 580' g Service Water

~

LPCI : 'M LPCI Line EL - 555' l $ Y Full Flow Test Return L,ine EL - 549' g Torus Spray 0

, ETI557 -

Suppressic,n'

_ Pool _ _ C RHR Torus Pumps l

Figure 2 Relative Verticle Elevations of the RHR System Cornponents

)k Head Spray

, r L,  :

i EL - 629 i

EL - 589' Containment Spray RHRS EL - 580' Service Water A

, / g EL - 577' LPCI Line i

Full Flow Return Line EL - 549' C Tcros Spray k

B EL - 537' _

Suppression Pool D O

l .

RHR

, Torus Pumps i ,

Figure 3 Containment Spray / Torus Spray Mode Flow Path

. m -

I s

/k Head .

Spray

. Z-EL - 629' .

EL - 589' -

EL - 587.5'  ;

Containment

" RHRS Spray -

EL - 580' Service Water X -

LPCI : 'M b LPCI Line N

EL - 555' u Y Y -'

c l

Full Flow 4 Test Return EL - 549' l Line C Torus Spray O

EL - 537' Suppress. ion Pool p RHR Torus Pumps Figure 4 Worst Case Pipe Voiding following RHR Pump Trip.~w~ hile in ~~

Containment Spray / Torus Spray Mode

~

O

r

- W m*" 8 _

o m

. g2Nw %E o -

g* v' ._

(

\ ,

'0 2

o,

/

i s D lde p n nnn i iai a y L Rm le r Hu tairC RP i h

. "4 o pP w

, 2 TSL

, p s i)

E 2o5 %j =

[ rc 1

Ti r pt e NW *

• Ps m m u o 0 I 2 R(

He R do gM n

i y

wa lor p loS f

s gu io nr idT o/y B Var p ep Rm pS Hu i Pt R. P e

se n Cn am 0

t st ia A r

2 ide s o n o nnai n C WC

%'2 0 ia p r y L )

_ tairC Rm Hu a

4

. l opp S RP

- TSL e Wf

~

4 r

% 2 k u ig F

bbc Mo '

v 0

2

@ V J

2Nw %$ <

[ -

0 2

Rm A

p Hu RP

1 y

Cont 'i"""'"', T Spray I \

Head Spray

(,fx s!'

' 12" , _<

. J C

q...wa '

'~

p,.

18'. 'N ! - ' i

% (s/ ullTest flow g F Return i c i

6"

/

'N

. .i I

i Torus From RHR Spray I ^#"'~~ t Pumps - A e C '

t s

.f/ [i1 LPCI Lines i e 11 18.. //

• 24"

'/;

\,\.

\\ (fg.

s\

f e

s i

12"  %  ;

'\ '\

s

\i x' .

24.. 'y e 24" \ \ .! C0ntainment ,

9 p.

t Spray I iro Y'S '\ S

\\\\ g-

\

Gg \\,7f >

k . .

1g . k .

g  :

r~" I

( ,

(t -

-B g.,'

,; gg*

' ~

24" -

j b.-,...,,,.......,,...,..Sg*

. . .,. .... L%w

, Torus Spray H 'cult Flo w '

~ Test Return H 24" From RHR mps B e O .

Figure 5(b) Worst Case Pipo Voiding following RHR Pump Trip while in ,

Contninment Spray / Torus Spray Mode {lsometric)

Jk Head Spray Z

EL - 629' EL - 589' Containment

-~

Spra9 RHRS EL - 580' Service Water

-X

, /% EL - 577' R RS LPCI Line  ;

1 4

4 3

EL - 555' W W Full Flow _

Test Return EL - 5t9' ,

Line -

l C Torus Spray l

o i EL - 537' .

C Suppression Pool P s .

RHR Torus

• Pumps Figure 6 Suppression Pool Cooling Mode Flow Path

-. o 24 -  !

Jk Head Spray Z i EL - 629' EL - 589' _

Containment Spray RHRS -

EL - 580' Service Water I

LPCI : M l

LPCI Line EL - 569.5' n

in EL - 555' W V W "

Full Flow Test Return EL - 549' Line g Torus Spray 0

EL - 537' y .

Suppression Pool C RHR Torus Pumps Figure 7 Worst Case Pipe Voiding following RHR Pump Trip while in Suppression Pool Cooling Mode i