ML18026A410

Responds to NRC Request for Addl Info Re Requested Rev to Tech Specs for Facility Ses Related to Rwcu/Hpci/Rcic Temp Based Steam Leak Detection Isolation Setpoints

ML18026A410
Person / Time
Site:	Susquehanna
Issue date:	08/19/1991
From:	Keiser H PENNSYLVANIA POWER & LIGHT CO.
To:	Butler W Office of Nuclear Reactor Regulation
Shared Package
ML17157A805	List: ML17157A804 ML17157A806 ML17157A807 ML17157A808 ML17157A809 ML17157A810 ML18026A410
References
PLA-3630, NUDOCS 9108260071
Download: ML18026A410 (35)

Text

ACCELERATED DISTRIBUTION DEMONSTRATION SYSTEM REGULATORY INFORMATION DISTRIBUTION SYSTEM (RIDS)

ACCESSION NBR:9108260071 DOC.DATE: 91/08/19 NOTARIZED: NO DOCKET FACIL:50-387 Susquehanna Steam Electric Station, Unit 1, Pennsylva 05000387 50-388 Susquehanna Steam Electric Station, Unit 2, Pennsylva 05000388 AUTH. NAME AUTHOR AFFILIATION KEISER,H.W. Pennsylvania Power 6 Light Co.

RECIP.NAME RECIPIENT AFFILIATION BUTLER,W.R. Project Directorate I-2 R SUEGECT: Responds to NRC request for addi info re requested rev to I TITLE: OR tech specs for face.lity SES related to RWCU/HPCI/RCIC temp based steam leak detection isolation setpoints.

DISTRIBUTION CODE'OOID NOTES:LPDR 1 cy Transcripts.

COPIES RECEIVED'LTR Submittal: General Distribution 2 ENCL l SIZE:~ 3~~

05000387 D

/

LPDR 1 cy Transcripts. 05000388 A A4 D ECIPIENT COPIES RECIPIENT COPIES ID CODE/NAME LTTR ENCL ID CODE/NAME LTTR ENCL D

PD1-2 LA 1 1 PD1-2 PD 1 1 THADANI,M. 2 2 INTERNAL: ACRS 6 6 NRR/DET/ECMB 7D 1 1 NRR/DET/ESGB 1 1 NRR/DOEA/OTSB11 1 1 NRR/DST 8E2 1 1 NRR/DST/SELB 7E 1 1 NRR/DST/SICB8H7 1 1 NRR/DST/SRXB 8E 1 1 NUDOCS-ABSTRACT 1 1 /-LFM8 1 0 OGC/HDS2 1 0 REG F 01 1 1 RES/DSIR/EIB 1 1 EXTERNAL NRC PDR 1 1 NSIC 1 1 NOTES 2 2 t

D D

D NOTE TO ALL "RIDS" RECIPIENTS S

PLEASE HELP US TO REDUCE WASTE! CONTACT THE DOCUMENT CONTROL DESK, ROOM Pl-37 (EXT. 20079) TO ELIMINATEYOUR NAME FROM DISTRIBUTION LISTS FOR DOCUMENTS YOU DON'T NEED!

TOTAL NUMBER OF COPIES REQUIRED: LTTR 26 ENCL 24

0 .'. 0, Pennsylvania Power 8 Light Company Two North Ninth Street~Altentowri, PA 18101-1179~215/774-5151 AUG l9 $ 91 Harold W. Kelser Senior Vice President-Nuclear 215/774%194 Director of Nuclear Reactor Regulation Attention: Dr. W. R. Butler, Project Director Project Directorate I-2 Division of Reactor Projects U.S. Nuclear Regulatory Commission Washington, D.C. 20555

'ocket SUSQUEIIANNA STEAM ELECTRIC STATION RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION ON PROPOSED AMENDMENTS 138 TO LICENSE NO. NPF-14 AND 92 TO LICENSE NO. NPF-22: REVISIONS TO TEMPERATURE LEAK DETECTION RWCU/HPCI/

RCIC SETPOINTS Nos. 50-387 PLA-3 30 and 50-388

Dear Dr. Butler:

This letter is in response to the NRC Staff's request for additional information regarding our requested revisions to the Technical Specifications for Susquehanna SES related to RWCU/HPCI/RCIC temperature based steam leak detection isolation setpoints. The following material provides additional background explanation for our requests in PLA-3487 dated January 9, 1991. Later material provides specific responses to the Staff questions contained in an NRC letter dated June 13, 1991.

The actions requested by PP&L in its Proposed Amendments 138 (NPF-14) and 92 (NPF-22) resulted from efforts to reconstitute the design bases for the temperature based steam leak detection and isolation circuitry in rooms within secondary containment which interface with the reactor coolant system as a result of the response to NRC Violation 88-15-01. They were also based on operating experience considering the normal seasonal variations of reactor building temperatures and design bases temperatures listed in FSAR Chapter 3 for maximum room temperatures under postulated accident conditions.

Reconstitution of design bases started with modeling each room in which steam leak detection circuitry is installed, and calculating room temperature response to postulated leak rates. Initial analyses used leak rates of 5 gpm, and subsequently, 25 gpm. We used our Compartment

,9f0826007f 910819 PDR P

AOOCK 05000387 PDR

~

Transient Temperature Analysis Program (COTTAP) computer program, which is discussed

,0 x!

+ 4 T

V

FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler below in our response to Staff Request 2, to generate room thermal response curves under leak conditions. The response also provides summary results of the calculations. Our response (below) to Staff Request 1 provides a functional basis for considering design bases leakage rates of 25 gpm for automatic isolation, and our response to Staff Request 3 provides other precedents for selection of a 25 gpm leak rate as a design bases.

RE ET 01 It is not clear why a change to the existing temperature setpoints is necessary. The licensee stated that the existing setpoints will not allow for timely detection of a 5 gpm leak, but the licensee did not dePne what leak rates the existing setpoints will detect. The current capability must be dered and the acceptability of the existing condition must be addressed.

R~NE The majority of the Technical Specification temperature based isolation setpoints in the RWCU area, HPCI area, and RCIC area remain unchanged by our request. The changes that were requested include high ambient and high differential temperature setpoints in the RWCU penetration room, and high ambient setpoints in the HPCI/RCIC room cooler air inlets.

In the RWCU area, the proposed change would eliminate an excessively small margin between isolation setpoints in the RWCU penetration room and peak summer temperatures. (The setpoints for the RWCU Pump rooms and RWCU Heat Exchanger rooms will remain at their present values.) At Susquehanna SES, several inadvertent isolations of RWCU have occurred, with resultant impact on plant chemistry and pump seal performance. The most recent inadvertent isolation occurred in June 1991 on Unit 2 and was reported to the NRC. Since Susquehanna SES began operation, several isolations of RWCU have occurred because of inadequate margin between isolation setpoint and peak temperature in the RWCU penetration room. Even without occurrence of an actual isolation, summer temperatures in the RWCU penetration room without adequate margin invokes unnecessary challenges to plant operations personnel.

In the HPCI area the proposed change to the two high temperature isolation circuits would eliminate an inconsistency between those circuits with setpoints of 147'F (mounted on air inlets to the room coolers), and two other circuits with setpoints of 167'F (wall mounted). Room cooler outlet air does not impact directly on any of the temperature monitoring elements used for isolation actuation. Both the local and wall mounted temperature circuits would be subject to similar temperature environments in the event of a room steam leak and should have the same setpoint.

~a FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler The HPCI room temperature in the event of an accident is described in FSAR Table 3.11-6 as 130'F maximum. The smaller margin between setpoint and maximum room temperature under these conditions (when an inadvertent isolation would be unacceptable) argues that the two circuits in question should have their setpoints raised to match the two wall mounted temperature circuit setpoints. All of the remaining high ambient temperature and high differential temperature circuit isolation setpoints in the HPCI room and in the HPCI pipe routing area would remain at their present value.

The RCIC area steam leak detection circuity is functionally identical to the HPCI area circuity, and is covered by the discussion above.

The requested changes to selected Technical Specification setpoints have a different motivation from our requests to redefine the design bases for the bulk of the temperature based isolation setpoints as described in our FSAR. The requested setpoint changes are intended to eliminate anomalies and inconsistencies for a limited number of setpoints.

Redefinition of the design bases for all temperature based isolation setpoints represents the culmination of a two year effort to establish an analytically consistent and uniform design bases for the bulk of the existing temperature based isolation circuit setpoints without changing those setpoints.

Our analyses defined room temperature response for defined leak rates under both summer and winter conditions. The specific response in each room was different and is discussed for the HPCI, RCIC, and RWCU rooms below. For all three rooms, the temperature rise resulting from a postulated 5 gpm leak was distinct but not significantly different from temperature variations due to seasonal differences or to loss of HVAC functions. Actual room temperatures uhder non leak conditions vary on a seasonal basis by as much as 25 to 30 degrees. This variation is a substantial fraction of the temperature rise calculated for a 5 gpm leak. Thus, establishing high ambient isolation setpoints based on 5 gpm leaks under winter conditions would result.,in an unacceptably small margin between the isolation setpoint and the high room temperatures expected in summer. Conversely, establishing the setpoint based upon the initial room temperature under summer conditions would'produce an unacceptably long time to reach the trip point under a postulated 5 gpm leak in initial winter conditions. The opposite relationships exist for the high differential temperature circuits.

These conclusions, which raised question on the acceptability of a postulated 5 gpm leak rate as the design bases, were reported to the NRC as an emerging design issue under the provisions of 10CFR50.9 in PP&L letters PLA-3214 dated 7/24/89, PLA-3315 dated 1/16/90, and PLA-3443 dated 9/21/90. These letters also discussed concurrent analyses at higher projected leak rates which demonstrated the adequacy of the existing Technical Specification setpoints to isolate the leaking line within a reasonable timeframe, and to protect the plant and the public.

FILES R41-2, A 17-2 PLA-3630 Dr. W. R. Butler The process for establishing safe and effective setpoints for the leak detection circuitry followed the following strategy. We first established the status of using 25 gpm leak rates in the industry by contacting GE Company, and other licensees. (See response to Question ¹3.) We performed offsite dose calculations to show the acceptability of the leak size (see response to Question ¹5). We then calculated the room transient thermal response with 25 gpm leaks, and determined the temperature reached at the end of a 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> period. We considered that temperature as an Analytic'Limit and performed a setpoint calculation (see response to Question ¹4.) Where the resultant setpoint was greater than the present setpoint listed in the Technical Specifications, and where adequate margin to unnecessary isolations currently exists, we reselected the Analytic Limit at a time shorter than the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> point to retain the present setpoints. For most of the rooms, this time was four hours or less. For the RWCU penetration room, thermal response required setting of the Analytic Limit at the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> point.

Specific data for the thermal response capabilities of the three areas is discussed below:

HPCI Area:

The current technical specification high ambient trip points are 167'F. Calculations indicate that the temperature rise resulting from a 5 gpm steam leak at rated process conditions would take in excess of 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> to reach this setpoint. Calculations were reperformed for HPCI with a presumed leakage rate of 25 gpm at rated process conditions. Temperature response shows the existing setpoint (167'F) is reached at about 18 minutes and the analytic limit is reached in 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />. We estimate that a smaller leak rate of 12-15 gpm (rated temperature and pressure conditions) would cause the setpoint to be reached in 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />; The high differential temperature response is similar to the high ambient temperature response.

't t

It The HPCI room cooler inlet high ambient temperature setpoints are inconsistent with the HPCI wall mounted high ambient temperature setpoints as discussed above. Although sets of circuitry monitor the same environmental conditions, the air cooler inlet 'oth temperature circuit is set 20'F lower than the wall mounted circuit. The proposed change increases the margin above maximum design room temperature, and therefore reduces the possibility of an inadvertent isolation particularly during an accident when room temperatures are elevated.

k ~

t

'I

FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler RCIC Area:

The current technical specification high ambient trip points are 167'F. Calculations indicate that the temperature rise resulting from a 5 gpm steam leak at rated process conditions would require up to eight hours to reach this setpoint. Calculations were reperformed for RCIC with a presumed leakage rate of 25 gpm at rated process conditions. The temperature response shows the existing setpoint (167'F) is reached in about 10 minutes and the analytic limit is reached in under 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />. We estimate that a smaller leak rate of 10 gpm (rated temperature and pressure conditions) would cause the setpoint to be reached in.4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />. The high differential temperature response is similar to the high ambient temperature response.

The discussion for the RCIC high ambient temperature on the room cooler inlet is the same as that discussed for HPCI above.

RWCU Area:

The current technical specification high ambient trip setpoints are 147'F in the pump room circuits and heat exchanger room circuits. They are 118.3'F in the penetration room circuits from which most of the spurious isolations have occurred. Calculations indicate that the temperature rise resulting from a 5 gpm leak at process conditions, starting from initial winter room thermal conditions, would be insufficient to reach all existing setpoints within a 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> period. The analyses of a 25 gpm leak at rated pressure and temperature conditions, starting from a winter condition showed that the proposed setpoint of 131'F would be reached in less'than 1.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> and the analytic limit would be reached in 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. The time required to reach the isolation setpoint (147'F) for the RWCU pump room and the heat exchanger room would be 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> and 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> respectively. The analysis for the differential temperature setpoint is similar to the high ambient except that the longer time response is associated with initial summer conditions (low delta T).

RE ET 2 The licensee credits certain temperature calculations in its safety analysis, but the details ofthe temperature calculations were not provided for stag review. It is not clear to the staff what assumptions were made and whether those assumptions are acceptable or not. Also, specij7c details regarding the application of the computer model COTTAP were not provided for staff review.

1 1

l

~I t'

Y

FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler R

Attachment A contains a user's manual for the COTI'AP computer code and copy of a paper recently published in Nuclear Technology which describes the methodology used in the COTlAP program and presents some of the verification calculations which have been performed. The user's manual presents some of the calculations which were performed against problems that have exact analytical solutions. The referred paper presents the methodology along with calculations which have been benchmarked against calculations performed with the CONTEMPT computer program. In addition, the program and computation package have been independently reviewed by Gilbert Associates. PP&L also maintains a Quality Assurance file/package for the COTTAP computer code.

Attachment B contains a summary of the calculations which were performed for each room and upon which the revised temperature setpoints were based. Calculations were performed for each room under a variety of conditions (for example, summer and winter initial conditions and various break sizes) and have been independently reviewed. The attachment presents a summary which includes the methodology and assumptions for each calculation along with the representative results which were used to calculate the revised setpoints.

RE ET The licensee arbitrarily selected 25 gpm as the design basis leak rate for all areas, stating that the 25 gpm leak basis is consistent with GE design specifications, with the basis used at other BWRs and with the Technical Speci(Ication Improvement Program (TSIP). The TSIP cannot be credited since the program currently has not been approved; and details relative to the GE design spectftcations and other BWRs was not provided for stagconsideration. The effects of the design basis leak on equipment, emergency operating procedures and personnel were not discussed (assuming prolonged operation near the design basis'eak rate). Additionally, no discussion relative to ASME Code requirements was pr'ovided.

A of 25 gpm as the design bases leak rate was considered when calculations using i'election the FSAR values of 5 gpm leakage were unable to support a consistent methodology for setpoint selection in all rooms with steam leak detection functions. These issues were reported to the NRC (see response to Request I), and investigation into a 25 gpm leakage rate as design bases was started. Decision to select 25 gpm as the design bases leakage value for HPCI/RCIC/RWCU area rooms was based on the following:

4 V

I I

pt FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler A limiting criteria was that the resultant high ambient temperature setpoints for system isolation had to include significant margin above maximum room temperatures under all non leak conditions to preclude inadvertent isolations. The computer modelling capabilities demonstrated that the previously assumed leakage value of 5 gpm could not meet this criteria for most of the rooms examined. This conclusion applied equally to differential temperature setpoints. An assumed 25 gpm leakage rate allowed a uniform approach to setpoint calculation, retention of most of the existing setpoints which incorporate adequate margin, and allowed raising those few setpoints (as requested) where insufficient margin existed.

Retention of the existing setpoints where analysis indicated some increase could be justified by a 25 gpm design bases leak also avoided possible conflicts with fire suppression initiation setpoints in those areas which included such systems.

2. Leak detection can be considered as a safety function with the purpose of minimizing or precluding the potential for a high energy line break (for which independent and diverse detection and isolation systems exist). FSAR Table 5.2-10 correlates leak rates to crack size up to cracks associated with unstable piping rupture for different pipe sizes and stresses. A leak rate of 25 gpm can be seen from that Figure to be less than those leak rates associated with the onset of unstable pipe rupture.

3. An assumed leakage of 25 gpm for calculating isolation setpoints was consistent with recommendations provided in GE document EDE-17-0689. GE advised that they were using 25 gpm as a design bases on all recent design activity.

Other existing BWR designs have been accepted using a leakage rate of 25 gpm as a design bases for leak detection. These include Perry, Grand Gulf, Clinton, and River Bend.

5. Off site dose calculations using leak rates of 25 gpm demonstrated acceptable safety consequences. See the response to Request 5.

After notifying NRC of our findings in 10CFR50.9 reports, room thermal analysis and setpoint calculations continued with the 25 gpm leak rate value. The effects of the new design bases leak rate on equipment, procedures, and personnel were assessed and found to be minimal.

The operating procedures require operator rounds into the HPCI and RCIC areas once per day. All areas with steam leak detection circuitry have their temperatures (and differential temperatures) available in the main control room for monitoring.

I 1,'i

FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler The alarm response procedures identify specific action required including observation, confirmation, isolation, and repair of leaks. Visual observation of a steam leak, or rising room temperatures, or the occurrence of a pre-isolation temperature alarm in the main control room would invoke operator action without attempting to quantify the leak rate, or waiting for the temperature to reach the isolation setpoint. Prolonged operation with any significant leak is not anticipated.

The emergency procedures are symptom based, and not event based. If a steam leak occurred, it would be isolated and repaired. Therefore, an increase in the analytic design bases leak rate would have no effect on emergency, alarm response, or operating procedures.

An increase in the defined design bases for leak detection should have no effect on personnel. Leaks will be detectable by the operational considerations mentioned above, and by diverse alarm systems such as area radiation monitors at levels far below design bases leakage rates. Operators would not enter areas with direct evidence of leakage, except if required as part of a planned evolution under controlled conditions with appropriate protective equipment. Therefore, the proposed changes should have no effect on station personnel.

Allequipment required to function within the environmental zone of the leak is included in our equipment qualification program. The equipment is qualified for the effects of a high energy line break - HELB.

If a leak were to occur, the system would be isolated and if appropriate an LCO (for HPCI and RCIC) would be entered. The faulty pipe will be repaired in accordance with the Susquehanna SES Welding and Non-Destruction Examination Manual. This manual is based on the requirements of ASME Section 3 (NC-2500, ND-2500) and Section 9.

The repair will be inspected in accordance with ASME Section ll. Prior to declaring the system operational, an evaluation of the leaks effect on other area equipment will be conducted and appropriate action taken.

E T 4 The methodology used in establishing the temperature setpoints was not described in detail, including consideration for instrument errors; the licensee did not describe to what extent industry standards were being used in establishing the temperature setpoints; and the logic used in selecting the system process conditions was not explained.

'C t F I'i lt f

1

FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler

~H R

The setpoints are calculated using the method in PP&L Design Guide, "Instrumentation and Control Setpoint Calculation Methodology". This safety related setpoint process utilizes the General Electric methodology and was used by General Electric in the calculation of the Susquehanna SES Technical Specification values.

Generic setpoint calculations based upon the PP&L Design Guide were prepared to establish the process for calculation of isolation setpoints for both high ambient temperatures and high differential temperatures. An analytical limit was defined based upon the thermal response curves calculated using the COTI'AP room models. The Allowable Value, Trip Setpoint (Tech Spec Setpoint), and Process Setpoint (as-installed setpoint) were defined starting from the Analytic Limit. The margin behveen the Analytic Limit and the Allowable Value accounts for instrument and calibration inaccuracy. The margin between Allowable Value and Trip Setpoint accounts for instrument drift. Drift values are derived from manufacturer's specified drift accuracy or from historical plant data if appropriate. The margin between Trip Setpoint and Process (as-installed) setpoint, usually based on the drift value, provides additional assurance that actual setpoints would not drift above Technical Specification Allowable Values.

Setpoint calculations were then completed for individual rooms, after selecting an Analytic Limit from the room thermal response curves calculated with the COTI'AP code. For the thermal calculations, leaking fluid was assumed to be at the process pressure and temperature conditions that would exist during normal power operation of the system that was presumed to be leaking. Because each individual room temperature response has a unique time dependent function, different times had to be selected for each room to determine the Analytic Limit.

Thermal response curves were calculated for both summer and winter conditions permitting the High Ambient Analytic Limit to be selected from the winter response curve, and the High Differential Temperature Analytic Limit to be selected from the summer response curve since these were the most conservative selections for the respective functions. The individual room setpoint calculations were then completed as described for the generic calculations above.

Iterations between setpoint calculations and room thermal response calculations were required when initially considering 5 gpm and subsequently 25 gpm as the design bases leakage. Where this process produced an Allowable Value and Trip Setpoint greater than prescribed by the Technical Specifications, and where no other anomalous factors existed (such as inadequate margin to worst case room temperatures), the existing setpoints were left unchanged. This essentially established an effective design bases leakage rate less than the 25 gpm used for the COTTAP room thermal response calculation, and would

FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler apply to'the HPCI and RCIC room setpoints. Our response to Request 1 provides estimates of the leakage rates which would cause isolation actuation after a four hour peflod.

RE T The licensee stated that the radiological consequences of a coolant leak outside primary containment was analyzed, but the details of the analysis was not provided for staff review.

R The radiological consequences of a coolant leak outside primary containment was analyzed in PP&L calculation SE-B-NA-078. Fifty gpm of reactor grade water was assumed to leak into secondary containment at a concentration of 4.0 uCi/gm Dose Equivalent Iodine-131. This'is the maximum allowable coolant concentration of iodine for Susquehanna SES operation. No credit for removal, holdup or decay was taken. The period of the leak was assumed to be 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> after which environmental leakage was terminated. The analysis concludes that the resultant offsite and control room doses fall far below 10 CFR 100 offsite dose limits and 10 CFR 50, Appendix A, GDC-19 control room dose limits.

An analysis of a reactor steam leak was also conducted in PP&L calculation FX-C-DAM-010. A 50 gpm water equivalent steam leak was assumed to occur for a 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> period.

No credit for removal, holdup or decay was taken. This analysis also concluded that the resultant offsite and control room doses fall far below 10CFR100 offsite dose limits and 10CFR50, Appendix A, GDC-19 control room dose limits.

Calculations SE-B-NA-078 and FX-X-DAM-010 which document the radiological analysis are included in Attachment C.

If you have any questions, please contact Mr. C.T. Coddington at (215) 774-7915.

Very truly yours,

. W. Keiser Attachments

FILES R41-2, A17-2 PLA-3630 Dr. W. R. Butler cc: NRC Document Control Desk (original)

NRC Region I Mr. G. S. Barber, NRC Sr. Resident Inspector Mr. J. J. Raleigh, NRC Project Manager

0 ATTACHMENT A

,.9108260071

C I

f A l'

COTTAP: A COMPUTER CODE ~ ~ ~ ~

FOR SIMULATION OF THERMAL TRANSIENTS IN SECONDARY CONTAINMENTS OF BOILING WATER REACTORS 'ARK A. CHAIKO and MICHAELJ. MURPHY Pennsylvania Power dc Light Company, Allentown, Pennsylvania 18101 Received December 1, 1989 Accepted for Publication September 12, 1990 The Compartment Transient Temperature Analysis out the calculation on an IBM 3090 computer. The Program (COTTA P) was developed by the Pennsylva- COTTAP code considers natural convection and radi-nia Power & Light Company for postaccident boiling ation heat transfer between compartment air and walls water reactor (B WR) secondary containment thermal through a detailed finite difference solution of the slab analysis. The code makes use ofpreviously developed conduction equations. Heat addition from hot piping implicit temporal integration methods and sparse ma- and operating equipment, and cooling effects associated trix inversion techniques to allow modeling of an en- with ventilation flows and compartment heat removal tire B IVR secondary containment. Investigations were units are also included. Additional capabilities of made with a model consisting of 121 compartments COTTAPinclude modeling of compartment heatup re-and 767 heat-conducting slabs. The simulation pre- sulting from steamline breaks and simulation of nat-sented involves the numerical integration of 20 101 or- ural circulation cooling in compartments with flow dinary differential equations over a 30-h simulation paths at differing elevations.

period. Two hours of CPU time were required to carry

~M@<-'e~~%@+.%kN " ~:e M48&KL4CF5FZ~~'"-""">V~>ZMSÃ I ~ INTRODUCTION ventilation system operates in a recirculation mode to promote air mixing between compartments and Under postaccident conditions, boiling water reac- to dilute locally concentrated radioactive isotopes.

tor (BWR) secondary containment ventilation systems Original design calculations for Pennsylvania Power typically isolate to prevent fission product release to & Light Company's (PP&L) Susquehanna Steam the environment. Since cooled air is no longer circu- Electric Station (SSES) assumed that air recircula-lated through the secondary containment, increased tion provided enough mixing to produce a fairly compartment temperatures result. Predictions of post- uniform temperature distribution throughout all sec-accident compartment temperatures are necessary to ondary containment compartments. For this reason, determine whether safety-related equipment is sub- a single-compartment transient model was used in the jected to temperatures that'exceed its maximum design simulation of postaccident conditions. Recent investi-values. Safety-related equipment must be operable un- gations based on steady-state calculations have shown, der postaccident conditions in order to effect the safe however, that significant temperature variations can shutdown of the reactor. exist between compartments. These temperature After an accident, the secondary containment variations were large enough to prompt a detailed NUCLEAR TECHNOLOGY VOL. 94 APR. 1991

Chalko and Murphy POSTACCIDENT BWR SECONDARY CONTAINMENTTHERMAL ANAL/$1$

multicompartment transient analysis of the secondary and containment.

To reanalyze the postaccident transient behavior of the SSES secondary containment, PP&L developed the V dpw di

=g Nu WuJ(1 YuJ) +

NI g Wij(I YIJ) j~i J=l Compartment Transient Temperature Analysis Pro- JJc gram (COTTAP). Development of this program began after an evaluation of available codes revealed that

+ Z Wcj(Y j~l Ycj) + Wbs Wcond Wro none were capable of performing a sufficiently detailed (2) simulation owing to the large number of heat-conduct-where ing structures found in the SSES secondary contain-ment. For example, the CONTEMPT code,'hiCh is V = compartment volume (m3) probably the most widely used containment analysis program, can model as many as 999 compartments but l = time (s) is limited to 99 heat-conducting slabs. In contrast, pp= compartment air and water vapor COTTAP can model up to 1200 heat-conducting slabs densities, respectively (kg/m3) and 300 compartments. It also contains models that describe heat dissipation from operating electrical WJ, Wlj, H~ = mass flow rates associated with equipment and process piping. A COTTAP model of j'th ventilation, leakage, and cir-the SSES-1 and -2 secondary containment structures culation paths, respectively (kg/s) consists of -120 compartments and 800 heat-conduct- Y = mass fraction of air within com-ing slabs. partment The CONTAIN codex's a more recently developed Y j, YIJ air mass fractions in donor com-containment simulation program with complex mod-eling capabilities. It is, however, designed specifically partments for ventilation path and leakage path j, respectively j

for primary containment simulation and is not well suited for secondary containment modeling because it Yj mass fraction of air in adjoining has no provisions for energy input to compartments compartment associated with cir-from heat loads such as electrical panels, lighting, mo-tors, and hot piping.

culation path j Wb, = rate of steam addition due to pipe A description of the COTTAP code, including as-breaks (kg/s) sumptions, governing equations, numerical solution methods, and code limitations is given in Sec. II. Rep- Wd = steam condensation rate (kg/s) resentative results of the SSES-1 and -2 secondary con-tainment analysis are presented in Sec. III, and code W, = rain-out rate (kg/s).

verification is discussed in Sec. IV. j The values W and W~ are positive for flow into the compartmerit and negative for flow out of the com-II. OESCRIPTION OF THE COTTAP CODE partment, whereas the circulation rate Wj is always a positive quantity. Ventilation paths are described by II.A. Compartment Mass and Energy Balances their associated mass flow rates and identification The COTTAP code allows for air and water vapor numbers of source and receiving compartments. Ven-mass transfer between compartments by means of tilation flows can be tripped off or on at any time dur-forced ventilation, leakage, and natural circulation ing a transient by supplying appropriate trip-logic data.

flows. A forced ventilation flow model describes heat- Leakage, circulation, and pipe break models are dis-ing/ventilating/air conditioning systems, and a leakage cussed in Sec. II.C along with other special purpose model simulates intercompartment flows that are gen- models.

erated by pressure differentials. In addition, a natural In formulating the compartment energy balance, it circulation model simulates gravity4riven flows between is assumed that air behaves as an ideal gas. Moreover, compartments connected by flow paths at differing for the transients of interest, partial pressures of wa-elevations. Steam can also be added to a compart- ter vapor are typically (I atm. Therefore, it is assumed ment as a result of pipe breaks or removed through that the steam speciflic enthalpy depends only on tem-condensation and rain-out. Air and water vapor mass perature, i.e., the vapor enthalpy is equal to the en-conservation equations for a compartment with N thalpy of saturated steam at the temperature of the gas ventilation paths, NI leakage paths, and N, natural cir- mixture. The partial pressure of water vapor within a culation paths are given by compartment is computed from the ideal gas equation of state, and the total compartment pressure is calcu-V dp dt

=Z Ivu WuJYuj+

W Q WgYIJ+ Z Wcj(YcJ JSR 1 Y) lated as the sum of the air and water vapor partial pressures. With these assumptions, the compartment energy balance becomes NUCLEAR TECHNOLOGY VOL. 94 APR. 1991 45

Chaiko and Murphy POSTACCIDENT BKVR SECONDARY CONTAINMENTTHERMAL ANALYSIS dCpa(T) Pb,k total compartment pressure if pipe paT paCp,(T) contains saturated liquid (Pa) pipe fluid pressure if pipe contains

't dT Pb,z

+p dhg(T) pR saturated steam (Pa) p lu a Iv a a dT hg(Pb,h) = specific enthalpy of saturated water

= VTCp,(T) dt

' Vlt (T) " vaPor at Pressure Pbreah (J/kg)

= specific enthalpy of saturated liquid ltf (T) water at temperature T (J/kg)

+ VT R dt

'"+R dt T J, Tij donor compartment temperatures for

+ Qh'ghs + Qpanei + Qmoror + Qcooier + Qpiping respectively (K) j ventilation path and leakage path j,

+ Qmisc + Qslab + Qbreal' IVbsltg(Pbreak)

IVroJtf (T) IVcond J>f (T)

Tj = temperature in adjoining compart-ment associated with circulation path

+

>u Z Vvj[YvjTvJCpa(Tvj) + ( vj)llg(Tvf)]

j (K)

J=l Compartment heat loads from lighting, electrical pan-NI els, motors, and miscellaneous equipment are main-

+ g ~ij[YiiTtjCpa(Tij) + ( I Jmi i)]

Ytj)ltg(Tj tained constant unless they are tripped on, off, or exponentially decayed during the transient. Hot piping and room cooler loads vary with compartment temper-

+ Z IVcj[ YjTcj Cp j~l (Tcj ) YTCpa (T) ature and can also be tripped on or off. In addition, hot piping heat loads can be exponentially decayed

+ (I Ycj) ling(Tcj) using the heat load decay model discussed in Sec.

( I Y) ltg (T)], (3) II.C.7.

where II.B. Slab Model T = compartment gas temperature (K)

In the secondary containment of a BWR, compart-Cp,(T) = specific heat of air at temperature T ment walls, ceilings, and floors are generally concrete (J/kg K) slabs that range in thickness from -0.3 to -2 m. To

=

ltg(T) specific enthalpy of saturated water determine the heat transfer rate between a compart-vapor at temperature T (J/kg) ment atmosphere and the bounding concrete slabs, the R, = ideal gas constant for air (288.7 J/

R= ideal kg K)

J/kg K) gas constant for water (461.4 IJT, at 'x'4) one-dimensional heat conduction equation cl2'T, is solved for each slab. Here, T, (K) is the slab temper-Qlighs, Qpanelt Qmosor~ Qcoolers Qpiplngs Qmisc ature, and x (m) is the spatial coordinate. Since the

= compartment heat loads due to light- thermal diffusivity e, (m /s) is supplied as input for each slab, materials other than concrete can be mod-ing, electrical panels, motors, air coolers, hot piping, and miscellane- eled provided that slabs are of uniform material com-ous equipment (J/s) position. This one-dimensional description assumes that slab edge effects do not significantly affect the Q,i,b rate of heat transfer to compartment overall rate of heat transfer.

air/water vapor mixture from sur- Boundary conditions on slab temperature are given rounding slabs (J/s) by Qb,h = heat transfer rate to air/water vapor cJT,

= hl mixture from liquid exiting break as it cools to compartment temperature Bxo k, [T, (t) Ts(0, t)]

(J/s) and 8'b, = mass flow rate of steam exiting break IJT,

= hg

[Ts(Ls t) T2(t)], (6)

(kg/s) BxL, k, 46 NUCLEAR TECHNOLOGY VOL. 94 APR. I99t

Chaiko and Murphy POSTACCIDENT BIVR SECONDARY CONTAINi4IENTTHERMAL ANALYSIS where free convection from a vertical plate. For horizontal Ti (t), Tz(t) = temperatures of compartments ad- slabs, free-convection coefficients depend on whether jacent to the slab the surface is being heated or cooled by the surround-ing gas mixture. As recommended by Holman," the k, = slab conductivity (J/m s K) correlation of Fujii and Imura is used with the mod-L, = slab thickness (m) ified characteristic length proposed by Goldstein et al 6 to compute the coefficient for an arbitrarily shaped hi llz heat transfer coefficients (J/ slab with heated surface facing upward or cooled sur-m- s K). face facing downward. In cases where the upper sur-The solution of Eq. (4) subject to Eqs. (5) and (6) gives face is cooled or the lower surface is heated, the the rates of energy transfer from the slab surfaces to correlations of Lloyd and (vloran are used.

the adjacent gas mixtures. Diatomic gases such as nitrogen and oxygen are es-The coefficients hi and h. account for natural sentially transparent to thermal radiation; however, the convection, radiation, and condensation heat transfer. emissivity of water vapor with respect to therinal radi-In the absence of condensation, the coefficient hi can ation is significant.a In COTTAP, radiant energy ex-be expressed as change between a slab surface and water vapor contained within the surrounding gas mixture is modeled through hi = hi+ hi,, (7) the use of an effective radiation heat transfer coeffi-where hiand hi, arc the natural convection and ra- cient [see Eq. (7)). For the applications of interest, tem-diation components, respectively. perature differences between a slab surface and the Natural convection coefficients are expressed in surrounding gas mixture are relatively small (typically terms of the Nusselt number, which in turn is a func- <5 K). Therefore, the following approximate relation tion of the Rayleigh and Prandtl numbers. For the co- proposed by Hottel and Sarofirn for small tempera-efficient hi, the appropriate relation is ture differences is used to compute thc radiation coef-ficient:

where Nu = " = f(Ra,Pr),

hiCt.

(8) hi, = '4+

(<s+ I) 2 a+ b c)e,<<,aT, (10)

Ct = slab characteristic length k = gas thermal conductivity where a = Stefan-Boltzmann constant (5.669 x mz s K4) 10 '/

and the Rayleigh and Prandtl numbers for the gas mix-ture are, respectively, defined by e, = slab emissivity Pr , =p k

T,= average T~= f(T" + T,((,g)/2) ',

temperature, which is defined by (I I)

(9) where where T = gas temperature (K) g = acceleration due to gravity (9.8 m/sz) Ts,/ slab surface temperature (K)

P = coefficient of thermal expansion (K ') c,= emissivity of water vapor evaluated at T,.

v = kinematic viscosity (mz/s)

The Cess-Lian'quations, which give an analytical a = thermal diffusivity (mz/s) approximation to the emissivity charts of Hottel and Egbert," are used to compute the water vapor emis-t( = dynamic viscosity (kg/m s) sivity. In Eq. (10), c has the value 0.45, and a and b are Cp = specific heat of the air/water vapor mixture obtained through differentiation of the Cess-Lian emis-(J/kg K). sivity equations Gas mixture properties used in the calculation of free convection coefficients arc evaluated at the thermal a8 ln (e,( T, P, P., P,L,)) (12) boundary layer temperature, which is taken as the av-8 ln(PL,)

erage of the slab surface temperature and the bulk gas and temperature.

For vertical slabs, coefficients are calculated from Bin[a,(TP~ P P L( ))

(13) the correlation proposed by Churchill and Chu'or 8 In(T)

NUCLEAR TECHNOLOGY VOL. 94 APR. I99i 47

Chaiko and Murphy POSTACCIDENT BIVR SECONDARY CONTAINMENTTHERMAL ANALYSIS where isolation of a pipe break (due to valve closure, for in-stance) a compartment begins to cool and condensa-P, = air partial pressure (Pa)'=

tion continues to occur on surrounding walls. For a water vapor partial pressure (Pa) sufficiently fast cooldown rate, condensation alone does not prevent compartment air from becoming sat-L = average mean beam length (m).

urated, and thus moisture droplets (rain-out) form Condensation on a slab surface occurs when the within the gas mixture. To maintain compartment rel-surface temperature drops below the dew point (the ative humidity less than or equal to unity, the rainout saturation temperature of water evaluated at the par- rate W(kg/s) is calculated from the following empir-tial pressure of water vapor in the compartment) of the ical model:

air/water vapor mixture. Heat transfer coefficients for condensation conditions are calculated using the exper- Wp = 200 (RH 0.99)max( WC,~ )

imentally determined Uchida" correlation, which in- if RH) 0.99 (15) cludes the diffusional resistance effect of noncondensible gases on steam condensation rates and In COTTAP, initial compartment temperatures, W, = 0,0 if RH ( 0.99, (16) pressures, and relative humidities are specified as in-put data. An initial slab temperature profile is deter- where mined by computing the steady solution to Eqs. (4), RH = relative humidity (5), and (6) corresponding to the initial compartment conditions. This implies that compartments have been W, = total steam flow rate into the compartment maintained at their initial conditions long enough for (kg/s) slabs to attain steady-state temperature profiles. C,) = constant that is supplied as part of the input data (kg/s).

II.C. Special Purpose Models The COTTAP code includes specialized models to II.C.2. Hot Piping Model simulate the effects of pipe breaks, hot piping, and In many secondary containment compartments, compartment air coolers. Leakage and natural circu- the major heat source consists of piping that contains lation models are also included to describe intercom- reactor steam or coolant. The heat addition rate to a partment mass transfer. In addition, the code includes compartment air/water vapor mixture from a hot pipe a simplified slab model, a heat load decay model, and is calculated from a compartment model in which temperature, pressure, and relative humidity are specified as a function of Q~l g= t t, t[ f T(t)] (17) time. where II.C.I. Pipe Break Model U~ = overall heat transfer coefficient (J/m s K)

Within the scope of the present model, pipes may L~ = pipe length (m) contain steam or saturated liquid water. Input data de- D~ = outside diameter of the pipe (or insulation if fine the total mass flow through the break Wt(kg/s) the pipe is insulated) (m) along with the time at which the break develops and the length of time over which fluid loss occurs. For Tf = pipe fluid temperature (K) pipes containing saturated liquid, the steam flow rate T = compartment temperature.

Wtexiting the pipe (kg/s) is calculated from the en-ergy balance The overall heat transfer coefficient is calculated by the code based on initial compartment conditions; the co-W>IAf(Pp) Wbsirg(P) + ( Wpl W5$ )irf(P) s (I4) efficient is then maintained constant throughout the which describes the isenthalpic expansion of fiuid from transient.

pipe pressure P~ to compartment pressure P. The liq-uid fraction, which does not flash as it leaves the pipe, II.C.3. Air Cooler Model is assumed to cool to compartment temperature, and Cooling units are used in a number of secondary the dissipated sensible heat is transferred directly to the containment compartments to remove heat generated compartment air/water vapor mixture. For the case by equipment such as emergency core cooling systems where a pipe contains steam, all of the mass and energy (ECCS) injection pumps and high-voltage buses and exiting the break is deposited directly into the compart- transformers. Heat removal rates of cooling units are ment gas mixture. calculated from Rain-out phenomena can be important in compart-ments containing pipe breaks. For example, following Q,,(t) = C, [T(t) T,oot(t)j (Ig) 48 NUCLEAR TECHNOLOGY VOL. 94 APR. I99I

C"a+o an" Murphy POSTACCIDENT BWR SECONDARY CONTAINMENTTHERMAL ANALYSIS where This model also describes intercompartment, gravity-driven circulation flows that can develop at open door-T<<t(t) = average of the inlet and outlet cooling ways (see the analysis of Brown and Solvason'.

water temperatures C,<<1 = constant that is computed from spec- II.C.6. Thin Slab Model ified initial values of the cooling load The detailed slab model discussed in Sec. II.B is Q p/ the inlet cooling water tempera- not required to describe heat transfer through thin ture, the cooling water flow rate, and thc compartment temperature T. slabs that have little thermal capacitance. Slabs of this type, e.g., refueling floor walls, have nearly linear tem-An energy balance on the cooling water yields the out- perature profiles, and thus the heat flow through a thin let cooling water temperature. slab can be calculated by the use of an overall heat transfer coefficient U>>. The rate of heat transfer II.C.4. Leakage Models through a thin slab is obtained from The COTTAP leakage model simulates pressure- q>>(t) = U>>A>>[T1 (t) T,(t)]

induced intercompartmental mass transfer through openings such as doorways and ventilation ducts. In- where tercompartment leakage is calculated by balancing the A>> = thin slab heat transfer area (m")

pressure differential between the compartments with an T1 Tp = temperatures of the compartments sepa-irreversible pressure loss. Thus, the leakage rate sat-rated by the slab (K).

isfies Values of U>> (J/m s K) are supplied as part of the Ktk IVg. ( t ) IVrk( t )

J code input data (one value for each vertical slab and two values for each horizontal slab). For horizontal where slabs, two values of U>> are required because free-convection film coefficients depend on the direction, P1, Pz pressures of the compartments associated upward or downward, of heat flow through the slab.

with the leakage path (Pa)

~ik leakage rate (kg/s) ll.C.7. Heat-Load Decay Model Ktk irreversible pressure loss coefficient Cooling of a component such as a pipe filled with Alk leakage area (m2) hot stagnant fluid or a pump that has ceased operat-ing is simulated through the use of a lumped-param-Plk gas density within the compartment sup- eter heat transfer model. Most compartments in the plying the. leakage flow (kg/m3). secondary containment have a large thermal capacity because of the bounding concrete slabs. It is therefore It is assumed that inertial effects do not significantly assumed that the component temperature changes on affect leakage rates. a faster time scale than the compartment air temper-ature; i.e., the air temperature is assumed to remain II.C.5. Natural Circulation Model fairly constant during the cooldown of the component.

With this assumption, the component heat dissipation A natural circulation model simulates gravity- rate Q,(t) is governed by driven mixing in compartments connected by flow paths at differing elevations. The circulation rate W, dQc(t)

Q ( ) (22)

(kg/s) is obtained from dt 2g(pz(t) Pt(t)] (~ <) where

t. Kt/IAtpz(t)] + E<</(Ap,(t)] j Qc(tO) = Qco (23) where and 7, (s '), the thermal time constant of the compo-

= densities of the air/water vapor mixtures nent, is given by pt, p2 within the two adjacent compartments Mc Cm (kg/m ) (here it is assumed that p2 is the Yc (24)

UcAc gas density for the cooler compartment) where E, Ft = elevations of the upper and lower flow paths (m) M, = mass of the component (kg)

A, At upper and lower flow path areas (mz). C~ = specific heat of the component (J/kg K)

NUCLEAR TECHNOLOGY VOL. 94 APR. 1991 49

Chaiko and Murphy POSTACCIDENT BIVR SECONDARY CONTAINMENTTHERMAL ANALYSIS U, = overall heat transfer coefficient (J/m2 s K) tions, fourth-order central difference formulas are used to compute T,; at interior grid points:

A, = component heat transfer area (m~).

In Eq. (23), to (s) is the time at which the cooldown I process begins, and Q, which is supplied as input Tsxxi = ( Tsi-2 + 16Tsi-i 30Tsi+ 16Tsi+i 128, 12 data, is the heat dissipation rate prior to cooldown. So-lution of Eqs. (22) and (23) gives the exponential-decay Tst+z) + O(~") (26) approximation used in COTTAP to model heat dissi- where pation of cooling components. The component time constant 7, is specified as input data except in the case i = 3,4,...,N 2 of hot piping, where it is calculated by the code from 6 = spacing between grid points.

the piping description data.

II.C.8. Tt'tne-Dependent Cotnparttnent Model A six-point sloping difference formula is used to ap-proximate T,; at i = 2 and i = N I:

With the time-dependent compartment (TDC) model, environmental conditions within a compart- I inent are specified as a function of time; i.e., temper- Tsxx2 = 2 (10Tsi I 5TsZ 4Ts3 + 14Ts4 ature, pressure, and relative humidity versus time are supplied as tabular input data. This model is particu- 6Tss+ Ts6) + O(h ) (27) larly useful for representing outside air conditions, in- and cluding solar and thermal radiation effects. The influence of solar and long-wave atmospheric radiation on exterior buildup surfaces can be described by spec-ifying the effective Sol-Air temperature'4 in the TDC TsxxN- i = 1 12622

( I OTsiv 15 Tsiv-i 4Tstv-3 instead of the actual outside air temperature. In sec- + 14Tsiv-3 6Tstv-4 + Tsiv-s) ondary containment analysis, the TDC model is also +O(~4) . (28) useful for describing transient conditions within the primary reactor containment, which are generally For the end points, where the normal derivatives known from the results of detailed licensing basis cal- are specified through convective boundary conditions, culations. the following finite difference approximations, recom-mended by Pirkle and Schiesser,'s are used to com-II.D. Numerical Solution Methods pute T~i.

An energy balance and two mass balances are solved for each compartment to determine gas temperature, air mass, and water vapor mass. In addition, the one-Tsxxi = I 12' 415 Ts i + 96T2 s 36T3+ s 32 3

T4s dimensional heat conduction equation is solved for 3 each slab. Before computing the numerical solution of Tss 50/3.'Tsxi) + O(h 4 ) (29) the governing equations, partial differential equations describing heat flow through slabs are approximated and by sets of ordinary differential equations (ODEs). This is accomplished through application of the method of lines (MOL). In the MOL, a finite difference approx-Ts iv =

1 1282 415 Tiv + 96Ttv 6

s i 36Tiv s z imation is applied only to the spatial derivative in Eq. (4), giving +

32 Tsh! 3 3

TsÃ-4 + SOATsxW 2

d Tsi

= oisTsxxi i (2S) +O(~4) . (30) where In Eqs. (29) and (30), thc normal derivatives Tsxi and T~> are evaluated in accordance with Eqs. (5) and (6),

i = 1,2,3,...,N, the number of equally spaced the convective boundary conditions; i.e.,

grid points Ti = slab temperature at grid point i Ts i = hi (Ti Tsi) ks T i = finite difference approximation to the second-order spatial derivative at grid pointi. and Following the approach used by Pirkle and Schiesser'n the MOL solution of parabolic equa- Tsx3 = h2 (Tsiv Tg) (31) ks 50 NUCLEAR TECHNOLOGY VOL. 94 APR. I99i

Chaiko and Murphy POSTACCIDENT BWR SECONDARY CONTAINMENTTHERMAL ANALYSIS All governing equations are now expressed in terms of 2. Cooler modeling does not describe moisture re-ODEs of the form moval under conditions where the cooling coil temper-ature is below the dew point of the inlet gas mixture.

= F(y, t) with y(0) = yo (32) 3. Pipe break modeling is valid only for lines con-taining steam or saturated liquid; breaks involving the release of subcooled liquid cannot be described.

Solutions of Eq. (32) exhibit rapid initial adjust- /

ments in compartment air temperature caused by the 4. Compartment flooding events cannot be simu-relatively small thermal capacitance of the air contained lated because all liquid is assumed to exit through com-within the compartment. Moreover, slab temperatures partment floor drains.

undergo rapid initial changes in narrow regions near the boundaries, resulting in the formation of spatial thermal boundary layers. In the numerical integration III. RESULTS OF SSES SECONDARY CONTAINMENT of Eq. (32), small time steps are required to simulate ANAIYSIS FOR POSTACCIDENT CONDITIONS these initial transients. As the initial transient response decays, however, it is desirable to increase step sizes in This section gives representative results for a COT-order to reduce the computation time required to fol- TAP simulation of the combined SSES-I and -2 sec-low the slowly varying part of the solution. Equations, ondary containments under postaccident conditions.

such as Eq. (32), which exhibit initial temporal bound- The thermal responses of the Units I and 2 secondary ary layer structures are termed stiff differential systems containments are coupled by heat transfer through (see the discussion in Ref. 16), and because of stabil- common walls that separate the two structures. The ity limitations, they cannot be solved efficiently with SSES model consists of 105 compartments, 16 time-explicit integration schemes. For this reason,.an im- dependent compartments, 767 slabs, 38 thin slabs, and plicit scheme was selected for COTTAP. 505 heat loads. The simulation was carried out for 30 h Numerical integration of the governing Eq. (32) is and required 124 min of CPU time on an IBM 3090 carried out with the LSODES code,'hich uses the computer. Note that most of the CPU time is required implicit backward differentiation methods proposed by to simulate the rapidly varying part of the transient Gear for the solution of stiff systems. The LSODES that occurs within the first few hours of the event.

code also employs sparse matrix inversion techniques Thus, substantially longer simulation times do not sig-in solving the implicit finite difference equations. With nificantly increase CPU time requirements.

these numerical integration features, it is feasible to For this analysis, it is assumed that a loss-of-carry out the integration of the large differential sys- coolant accident (LOCA) occurs in SSES-I and a false tems that arise in the simulation of secondary contain- LOCA signal (a spurious signal that indicates loss of ment transients. As an illustration of the problem reactor coolant and leads to ventilation system'sola-dimension, simulation of the SSES-I and -2 secondary tion and operation of ECCS injection pumps) is gen-

'ontainments under postaccident conditions required erated on SSES-2. Under postaccident conditions, the solution of 20101 coupled ODEs. ECCS injection pumps comprise the key equipment For these large-scale problems, reevaluation of within the secondary containment structure. The ECCS code-calculated slab heat transfer coefficients at every consists of the residual heat removal (RHR), core time step leads to unacceptably long computation spray, and high-pressure coolant injection (HPCI) sys-times. To alleviate this difficulty, the frequency of re- tems. These systems receive electrical power from high-evaluation (number of steps between reevaluation of voltage buses contained within emergency switch gear coefficients) is a parameter supplied as input to the and load center rooms. Figure I shows the calculated code. Sensitivity calculations on small-scale problems temperature response within a SSES-I RHR pump representative of postaccident secondary containment room (each unit contains two RHR pump rooms and transients indicate that coefficients can be reevaluated two core spray pump rooms). Initially, the air temper-as infrequently as once per ten steps without introducing ature increases rapidly because of the small thermal ca-significant errors in the results. The CPU time require- pacitance of the air within the'compartment. As air ments were reduced by a factor of 4 when coefficients temperature increases, a balance between compartment were reevaluated at every tenth time step. heat sources and losses to compartment air coolers and slabs begins to develop. At this time, air temperature II.E. Code Limitations in Modeling Accident Scenarios starts to increase on the slow time scale governed by the slab thermal capacity and transport properties. An The following modeling limitations have been iden- initial rapid temperature rise followed by a much tified in the current version of the COTTAP code: slower temperature increase is characteristic of all com-partment heatup transients. After I h of operation, this

1. Fission product transport among compartments particular RHR pump switches from the injection is not modeled. mode of operation to the suppression pool cooling NUCLEAR TECHNOLOGY VOL. 94 APR. I99t 51

Chaiko and Murphy POSTACCI DENT BiVR SECONDARY CONTAINMENTTHERMAL ANALYSIS 322 . 317 316 320

~~ 315.

~ 31S.

I- I- 314 E

E o 316. o 313.

312 cL 314 .

31 1 z 310 .

312 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Time (h) Time (h)

Fig. 1. Simulation of postaccident temperature response Fig. 3. Simulation of postaccident temperature response within SSES-I RHR pump room for LOCA on within SSES-I HPCI pump room for LOCA in SSES-I and false LOCA on SSES-2. SSES-I and false LOCA in SSES-2.

mode. As a result of increased compartment heat loads HPCI system, which also begins operation at the start associated with the change in operating mode, the tem- of the accident. In this case, however, compartment perature again increases rapidly until a ncw balance temperature continues to increase when the system between the heat-generation and heat-loss rates is at- ceases operation at I h into the transient. This occurs tained. because piping heat loads within this compartment are The temperature response within a SSES-I core substantial. When HPCI pump operation stops, an as-spray pump room is shown in Fig. 2. Core spray op- sociated room cooling unit also ceases operation. Upon eration begins at the start of the event and ceases I h shutdown of the cooling unit, slowly decaying piping later. Temperature decreases rapidly at this point be- heat loads rapidly increase compartment temperature cause, once pump operation is terminated, no signif- until a balance between heat generation and heat losses icant heat loads remain in the compartment. Figure 3 to compartment slabs is approached. Figure 4 gives the illustrates the temperature response of the SSES-I temperature within a SSES-I load center room that

~ 317 309 hC L

c 316 E

I- P 308 E

315 E o

o

~ 307.

~ 314-O 0

313 306 O 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Time (h) Time (h)

Fig. 2. Simulation of postaccident temperature response Fig. 4. Simulation of postaccident temperature response within SSES-I core spray pump room for LOCA in within SSES-I load center room for LOCA in SSES-I SSES-I and false LOCA in SSES-2. and false LOCA in SSES-2.

52 NUCLEAR TECHNOLOGY VOL. 94 APR. 1991

Chaiko and Murphy POSTACCIDENT BIVR SECONDARY CONTAINMENTTHERMAL ANALYSIS supplies electrical power to emergency equipment. In 315 this compartment, heat loads remain essentially con-stant throughout the transient.

From the results of this analysis, it is determined that under postaccident conditions, some of the equip-ment within the secondary containment would be ex- ~ 310 posed to temperatures that exceed their qualification L values. Consequently, components were reassessed for operation at higher temperatures, and in some in-stances equipment was relocated to compartments with less severe environmental conditions. Furthermore, a ~~ 305 procedure was developed to instruct plant operators to COTTAP shed nonessential electrical loads within 24 h after an CONTAIN accident in order to moderate the temperature re-sponses within secondary containment compartments.

300 0 2 4 6 8 10 IV. EVALUATION OF CODE ACCURACY Time (h)

Fig. 5. Comparison of COTTAP and CONTAIN compart-As part of the verification process for the COT- ment temperature simulations for test problem I, TAP code, calculational results were compared with those obtained with the CONTAIN (Ref. 2) program, which has been verified through comparison with ex-perimental data.' Although the CONTAIN code 0.20 does not accommodate a direct heat input (such as from operating mechanical or electrical equipment) to a compartment, useful problems can nevertheless be 0.18 formulated in order to investigate the modeling and computational accuracy of COTTAP. Two such prob-lems were formulated for code verification. The first 0.16 problem tests the COTTAP compartment mass and en-ergy balance calculations and the slab heat transfer simulation. This problem consists of a single compart- I 0.14 ment that has a 1000-m3 volume and contains air at COTTAP 300 K and 101325-Pa initial temperature and pressure.

Concrete slabs, which range in thickness from 0.1 to 0.12 CONTAIN I m, form the walls of the compartment. All slabs have a uniform, initial temperature of 300 K. To add heat 0.10 to the compartment, the air in contact with the outer 0 2 4 6 8 10 surface of one slab (the slab that is 0.1 m thick) is sud- Time (h) denly increased to 400 K at t = 0. In addition, at 50 s into the transient, air with a temperature of 500 K is in- Fig. 6. Comparison of COTTAP and CONTAIN.compart-jected into the compartment at a 0.26 kg/s flow rate. ment pressure simulations for test problem I.

Outer surface temperature rise and air injection con-ditions were selected to effect significant, but not ex-cessive, temperature and pressure response.

Figures 5 and 6 present a comparison of the COT- ification involves modeling of compartment tempera-TAP and CONTAIN calculation results for the first ture and pressure behavior under conditions where test problem. The temperature and pressure simula- high-energy steam is injected into the compartment. In tions both show excellent agreement; note that the this problem, condensation effects strongly influence pressure response curves given in Fig. 6 completely the rate of temperature and pressure increase. Com-overlap. In Fig. 5, the initial temperature increase, partment physical description data are the same as that which is due to injection of hot air into the compart- for test problem l. In this case, however, the only heat ment, begins to level off at -0.5 h. Heat addition by source is the steam entering the compartment at a means of conduction through the externally heated slab 0.20 kg/s flow rate and a 2.7756 x 10'/kg enthalpy.

then begins to occur, causing a further but less rapid This flow rate and enthalpy are characteristic of a increase in temperature. small steam leak within a secondary containment com-The second test problem considered for code ver- partment. Figures 7 and 8 show a comparison of the NUCLEAR TECHNOLOGY VOL. 94 APR. t99t 53

Chaiko and Murphy POSTACCIDENT BiVR SECONDARY CONTAINMENTTHERMAL ANALYSIS 450 ACKNOWLEDGMENTS Thc authors thank Jack G. Refling, James E. Agnew, Mark R. Mjaatvedt, and Leonard J. West for their many helpful suggestions during the course of this work. We also

" 4oo thank Lisa Walsh for typing the manuscript.

n, REFERENCES E

350 1. C. C. LIN, C. ECONOMOS, J. R. LEHNER, G.

MAISE, and K. K. NG, "CONTEMPT4/MOD4: A Multi-COTTAP compartment Containment System Analysis Program,"

CONTAIN BNL-NUREG-51754, Brookhaven National Laboratory (1984).

300 0 5 10 15 20 2. K. K. MURATAet al., "User's lvfanual for CONTAIN Time Ih) I.l: A Computer Code for Severe Nuclear Reactor Accident Containment Analysis," NUREG/CR-5026, Sandia Na-Fig. 7. Comparison of COTTAP and CONTAIN compart- tional Laboratories (1989).

ment temperature simulations for test problem 2.

3. S. W. CHURCHILL and H. H. S. CHU, "Correlating Equations for Laminar and Turbulent Free Convection from a Vertical Plate," Int. J. Heat Mass Transfer, 18, 1323 (1975).

0.6

4. J. P. HOLMAN, Heal Transfer, 4th ed., p. 250, McGraw-Hill Book Company, New York (1976).

0.5

5. T. FUJII and H. IMURA, "Natural Convection Heat

~ =

g 0.4 Transfer from a Plate with Arbitrary Inclination," Inr. J.

Heat Mass Transfer, 15, 755 (1972).

6. R. J. GOLDSTEIN, E. M. SPARROW, and D. C.

JONES, "Natural Convection Mass Transfer Adjacent to N

C7 03- Horizontal Plates," Ini. J. Hear Mass Transfer, 16, 1025 o- (1973).

0.2 r CONTAIN COTTAP

7. J. R. LLOYD and W. R. MORAN, "Natural Convec-tion Adjacent to Horizontal Surface of Various Planforms,"

ASME 74-WA/HT-66, .American Society of Mechanical 0.1 Engineers (1974).

0 5 10 15 20

8. D. Q. KERN, Process Hear Transfer, p. 690, McGraw-Time (h) Hill Book Company, New York (1950).

Fig. 8. Comparison of COTTAP and CONTAIN compart-ment pressure simulations for test problem 2. 9. H. C. HOTTEL and A. F. SAROFIM, Radiative Transfer, McGraw-Hill Book Company, New York (1967).

10. R. D. CESS and M. S. LIAN, "A Simple Parameteriza-tion for the Water Vapor Emissivity," Inr. J. Hear Transfer, 98, 676 (1976).

COTTAP and CONTAIN simulation results. The re-sults show good agreement even though the codes em- 11. H. C. HOTTEL and R. B. EGBERT, "Radiant Heat ploy considerably different approaches in the Transmission from Water Vapor," Am. Insl. Chem. Eng.,

calculation of condensation rates on slab surfaces. The 38, 531 (1942).

COTTAP code uses the experimentally determined 12. H. UCHIDA, A. OYAMA, and Y. TOGO, "Evalua-Uchida'ondensation coefficient, while CONTAIN tion of Post-Incident Cooling Systems of Light-Water carries out a detailed computation of the thermal re- Power Reactors," Proc. 3rd Inr. Conf. Peaceful Uses of sistances associated with the gas boundary layer and Atomic Energy, Geneva, Switzerland, 1964, Vol. 13, p. 93, the condensate film. United Nations (1965).

54 NUCLEAR TECHNOLOGY VOL. 94 APR. 1991

~ ~ ~ /

Chaiko and Murphy POSTACCIDENT BWR SECONDARY CONTAINMENTTHERMAL ANALYSIS

13. W. G. BROWN and K. R. SOLVASON, "Natural Con- 17. A. C. HINDMARSH, "ODEPACK, A Systematized vection Through Rectangular Openings in Partitions-I Ver- Collection of ODE Solvers," Scientific Computing, Vol. I, tical Partitions," Int. J. Heat Mass Transfer, 5, 859 (1962). p. 55, R. S. STEPLEMAN et al., Eds., IMACS Transac-tions on Scientific Computation, North-Holland Publishing

14. ASHRAE Handbook 1985 Fundamentals, American Company, Amsterdam (1983).

Society of Heating, Refrigerating and Air-Conditioning En-gineers, Atlanta, Georgia. 18. K. K. MURATA and K. D. BERGERON, "Experimen-tal Validation of the CONTAIN Code," Proc. Iltlt LIVR

15. J. C. PIRKLE, Jr. and W. E. SCHIESSER, "DSS/2: A Safety Information Mtg., Gaithersburg, Maryland, October Transportable FORTRAN 77 Code for Systems of Ordinary 24-28, 1983, SAND-83-1911C, Sandia National Laborato-and One, Two and Three-Dimensional Partial Differential ries (1983).

Equations," presented at 1987 Summer Computer Simula-tion Conference, Montreal, Canada, 1987. 19. K. K. MURATA et al., "CONTAIN: Recent Highlights in Code Testing and Validation," Proc. Int. Mtg. Light IVater

16. C. W. GEAR, Munerical Initial Value Problems in Or- Reactor Severe Accident Evaluation, Cambridge, Massa-dinary Differential Equations, Chap. 11, Prentice-Hall, En- chusetts, August 28-September I, 1983, American Nuclear glewood Cliffs, New Jersey (1971). Society (1983).

Mark A. Chaiko [BS, 1980, and MS, 1983, chemical engineering, Penn-sylvania State University (PSU); PhD, applied mathematics, Lehigh Univer-sity, 1989] is a project engineer-nuclear systems at the Pennsylvania Power &

Light Company. His current technical interests include boiling water reactor stability analysis and thermal-hydraulic modeling of reactor systems.

Michael J. Murphy (BS, mechanical engineering, 1982, and MS, nuclear engineering, 1986, PSU) is a project engineer-nuclear systems with the Penn-sylvania Power & Light Company. He is currently involved in simulation of anticipated transient without scram and severe accident analysis.

NUCLEAR TECHNOLOGY VOL. 94 APR. 1991 55