ML17159A879

Application for Amend to Licenses DPR-32 & DPR-37,deleting Surveillance Requirements for Boron Injection Tank Level Instruments,Inadvertently Included in Amends 97 & 96.Fee Paid

ML17159A879
Person / Time
Site:	Surry
Issue date:	07/12/1985
From:	Stewart W VIRGINIA POWER (VIRGINIA ELECTRIC & POWER CO.)
To:	Harold Denton, Varga S Office of Nuclear Reactor Regulation
Shared Package
ML17159A880	List: ML17159A879 ML18142A565
References
82-342, NUDOCS 8507190306
Download: ML17159A879 (109)

Text

DAR TABLE OF CONTENTS VOLUME 1 NON-PROPRXETARY Chapter 1 GENERAL INFORMATION 1.1 Purpose and Organization of Report 1.2 History of Problem 1.3 SSES Containment Program 1.4 Plant Description Chapter 2

SUMMARY

2.1 Load Definition Summary 2.2 Design Assessment Summary Chapter 3 SRV DISCHARGE AND LOCA TRANSIENT DESCRIPTION 3.1 Description of Safety Relief Valve (SRV) Discharge 3.2 Description of Loss-of-Coolant Accident (LOCA)

Chapter 4 LOAD DEFINITION 4.1 Safety Relief Valve (SRV) Discharge Load Definition

4. 9 T,n~~-nf-C'nnl ant Accident {LOCA) Load Definition 4.3 Annulus Pressurization Chapter 5 LOAD COMBINATIONS FOR STRUCTURES, PXPING, AND EQUIPMENT 5.1 Concrete Containment and Reactor Building Load Combinations

'5.2 Structural Steel Load Combinations 5.3 Liner Plate Load Combinations 5.4 Downcomer Load Combinations 5.5 Piping, Quencher, and Quencher Support Load Combinations 5.6 NSSS Load Combinations 5.7 Balance of Plant (BOP) Equipment. Load Combinations 5.8 Electrical Raceway System Load Combinations 5.9 HVAC Duct System Load Combinations Chapter 6 DESIGN CAPABILXTY ASSESSMENT 6.1 Concrete Containment and Reactor Building Capability Assessment Criteria 6.2 Structural Steel Capability Assessment Criteria 6.3 Liner Plate Capability Assessment Criteria De.,"-~.ed> @5+

'E c~.';.,) 8 85jP+f goop Date t:,'ZGUIA OAY OCf(ETRLE Rev. 9, 07/85

TABLE OF CONTENTS (Continued)

6. 4 Downcomer Capability Assessment. Criteria 6.5 Piping, Quencher, and Quencher Support Capability Assessment Criteria 6.6 NSSS Capability Assessment Criteria 6.7 BOP Equipment Capability Assessment Criteria 6.8 Electrical Raceway System Capability Assessment Criteria 6.9 HVAC Duct System Capability Assessment Criteria Chapter 7 DESIGN ASSESSMENT 7.1 Assessment Methodology 7.2 Design Capability Margins Chapter 8 SSES QUENCHER VERIFICATION TEST (PROPRIETARY) 8.1 Introduction 8.2 Test Facility and Instrumentation 8.3 Test Parameters and Matrix 8.4 Test Results 8.5 Data Analysis and Verification of Load Specification Chapter 9 GKM IIM STEAM BLOWDOWN TESTS 9.1 Introduction 9.2 Test Facility and Instrumentation 9.3 Test Parameters and Matrix 9.4 Test Results 9.5 Data Analysis and Load Specification 9.6 Verification of the Design Specification Chapter 10 RESPONSES TO NRC QUESTIONS 10.1 NRC Questions 10.2 Responses Chapter 11 REFERENCES VOLUME 2 NON-PROPRIETARY Appendix A CONTAINMENT DESIGN ASSESSMENT A.l Containment Structural Design Assessment A.2 Submerged Structures Design Assessment Appendix B CONTAINMENT RESPONSE SPECTRA DUE TO SRV AND LOCA LOADS B.l Containment Mode Shapes B.2 Containment Response Spectra Rev. 9, 07/85

TABLE OF CONTENTS (Continued)

Appendix C REACTOR/CONTROL BUILDING RESPONSE SPECTRA DUE TO LOCA AND SRV LOADS Appendix D PROGRAM VERIFICATION D.1 Poolswell Model Verification D.2 Velpot Computer Code Description and Verification Appendix E REACTOR AND CONTROL BUILDING DESIGN ASSESSMENT Appendix F BOP AND NSSS PXPING DESIGN ASSESSMENT Appendix G NSSS DESIGN ASSESSMENT (DELETED)

Appendix H EQUIPMENT DESIGN ASSESSMENT (DELETED)

Appendix I SUPPRESSION POOL TEMPERATURE RESPONSE TO SRV DISCHARGE Appendix J VERIFICATION OF SRV SUBMERGED STRUCTURE DRAG LOAD PROPRIETARY)

Appendix K DRYWELL FLOOR VACCUM BREAKER (VB) CYCLING DURING CHUGGING Appendix L SUPPLEMENTAL DESIGN ASSESSMENT L.1 Assessment Methodology L.2 Assessment Results VOLUME 3 PROPRIETARY Chapter 4 PROPRXETARY LOAD DEFINITION Chapter 8 PROPRIETARY SSES QUENCHER VERIFICATION TEST Appendix D PROGRAM VERIFICATION APPENDIX J VERXFICATION OF SRV SUBMERGED STRUCTURE LOAD VOLUME 4 PROPRIETARY Chapter 9 SSES IOCA STEAM CONDENSATION VERIFICATION TEST Rev. 9, 07/85

PROPRIETARY 9-171 Comparison of the Amplitudes of Pressure Time Histories No. 3 and 5 with the Mean Values During the Full MSL Test, Nos 3 .. 10 9-172 Comparison of the Amplitudes of Pressure Time History No. with the Values During the 1/3 Test Nos.

ll and 6

12 Mean MSL 9-173 Comparison of the Amplitudes of Pressure Time History No. 9 with the Mean Values During the 1/6 MSL Test Nos.

13 .20 9-174 Oscillatory Frequencies of P ressure Time Histories No.

3 and 5 and Pressure Events During Full MSL Test Nos.

3 ..10 9-175 Oscillatory Frequencies of Pressure Time History No. 6 and Pressure Fvents During 1/3 MSL Test No. 11 9-176 Oscillatory Frequencies of Pressure Time History No. 9 and Pressure Events During 1/6 MSL Test Nos. 13 and 19 9-1 77a Finally Selected CO Time Segment Pressure Time History No. 14 9-177b Finall y S el ected CO Time Segment Pressur e T i me History No. 14 177c PSD of the specified CO Time Segment 9-178 Oscillatory Frequencies of Pressure Time History No. 14 and Pressure Events During RCL Test Nos. 1 and 33 9-178a Comparison of the Envelope of the PSD 's of the Selected PTH Nos. 3, 5, 6 and 9 and the Envelope of the Test/Be pea t Me an PSD 's 9-178b Normalized PSD of P TH No. 3 9-178c Normalized PSD of PTH No. 5 9-178d Normalized PSD of PTH No. 6 9-1789 Normalized PSD of PTH No. 9 9-178f Fnvelope of Normalized PSD's of PTH Nos. 3, 5, 6 and 9 9-178@ Mean Values of Pressure Events and Time Spans from which PSD 's are produced Full MSL Break 9-178h Mean Values of Pressure Events and Time Spans From which PSD's are Produced 1/3 MSL Break Rev. 9, 07/85 9P-13

PROP R IETAR Y 9-178i Mean PSD's of Test/Repeat Test No. 3/4 for Corresponding Time Segments Mean PSD's of Test/Repeat Test No. 3/4 for 9-178'-1 Corresponding Time Segments 78k Mean PSD' of Test/Repeat Test No. 9/10 f or Corresponding Time Segemnts 9-178m Mean PSD 's of Test/Repeat Test No. 9/10 for Correspondinq Time Segments 9-178n Mean PSD's of Test/Repeat Test No. 11/12 for Corresponding Time Seqments 9" 1 78p Mean PSD's of Test/Repeat Test No. 11/12 f or Corresponding Time Segmen ts 9-178q 'Mean PSD 's of Test/Repeat Test No. 11/12 for Corresponding Time Segments 9-178r Envelope of the Mean PSD's of the Time Segments from Test Nos. 3/4, 9/10, ll/12, 13/14 and 19/20 9-179 Pressure Amplitudes Normalized to the Sliding Mean Values 13andpass 0.5-13 Hz - Test No. 5 9-180 Pressure Amplitudes Normalized to the Sliding Mean Values Band pass 0. 5-13 Hz Test No. 15 9-181 Probability Densities of the Normalized Pressure Amplitudes with Mean Values > 0 5 bar Full MSL Tests 9-1 82 Probability Densities of the Normalized Pressure Amplitudes with Mean Values > 0.5 Bar 1/3 MSL Tests 9-183 Probability Densities of the Normalized Pressure Amplitudes with Mean Values > 0.5 Bar 1/6 MSL Tests

/

9-184 Derivation of the Maximum Overturning Moment 9-185 Multi-Sinqle-Cell Model of the SSES Pool 9-186 Spatial Pressure Profile in the Circumferential Direction 9-187 Results of the Monte Carlo Calcu1ations 9-188 Load Exceedance Criteria for Bandpass 0.5 13 Hz and Zero Phase 9-189 Load Exceedance Criteria for Band pass10-100 Hz and Zero Phase Rev. 9, 07/85 9P-14

PROPRIETARY 9-190 Effect of Delynchronized Time Window on Bottom Pressure Amplitude Harmonic Time Histories with Identical Amplitudes in All Cells 9-191 Load Fxceedance Criteria for Bandpass 0. 5-13 Hz and Des yn chron ized T i me W indov 9-1 92 Load Exceedance Criteria for Bandpass 10 100 Hz and Desynchronized Time Window 9-193 KWU Source 303 Time History 9-19 4 KHU Source 305 Time History 9-195 KWU Source 306 Time History 9-196 KHU Source 309- Time History 9-197 KHU Source 314 Time History 9-198 Comparison of PTH No. 3 (P 6.8) and the Pressure Output From IWEGS 9-199 Comparison of the PSD of PTH No. 3 (P 6.8) and the PSD of the Pressure From ZWEGS 9-2 00 Comparison of PTH No. 5 (P 6.8) and the Pressure Output From IHEGS 9-20l Comparison of the PSD of. PTH No. 5 (P 6.8) and the PSD of the Pressure'utput from IWEGS 9-202 Comparison of PTH No. 6 (P 6.8) and the Pressure Output Prom IHEGS 9-2 03 Comparison of the PSD of PTH No. 6 (P 6.8) and the PSD of the Pressure Output From IWEGS 9-204 Comparison of PTH No. 9 (P 6.8) and the Pressure Output Prom IWEGS 9-205 Comparison of the PSD of PTH No. 9 (P 6.8) and the PS D of th e Pressure Output Prom IWEGS 9-2 06 Comparison of PTH No. 14 (P 6. 8) and the Pressure Output From IWEGS 9-207 Comparison of the PSD of PTH No..14 (P 6.8) and the PSD of the Pressure Output Prom IWEGS 9-208 IWEGS/MARS Coordinate System 9-2 09 Comparison of Horizontal Acceleration Response Spectra thLu Due to LOCA (DPPR) vs. LOCA (KWU) 9-218 Rev. 9, 07/85 9P-1 5

PROP RIETAR Y 9~2 19 Comparison of Vertical, Acceleration Response Spectra thru D ue to LOCA (DFFR) 's. LOCA (KHU) 9-227 9-228 Comparison of Horizontal Acceleration Response Spectra thru for the combination SSE + SRV + LOCA(DFFR) vs.

9- 237 SSE + SRV + LOCA {KMU) 9-2 38 Comparison of Vertical Acceleraticn Response Spectra thru for the Combination SSF. + SRV + LOCA(DFFR) vs.

9-246 SSE + SRV + LOCA (KRU)

~ 9-2 47 Schematic View o f P in ite E le me nt M odel 9-248 Schematic View of Structural Model 9-249 Schematic View of Pluid Elements 9" 250 Amplitudes of the Resultant Force at the Bracing Frequency Range: 0.5... 200Hz Test No. 3 6 4 9-2 51 Amplitudes of the Resultant Force at the Bracing Frequency Ranqe: 0. 5...200Hz Test, No. 5 6 6 9-252 Ampli tudes of the Resultant Force at the Bracing Frequency Range: 0.5..200Hz Test No. 7 6 8 9-253 Amplitudes of the Resultant Force at the Bracing Frequency Range: 0.5...200 HZ Test No. 9 6, 10 9- 254 Amplitudes of the Resultant Force at the Bracing Frequency Range: 0.5...200Hz Test No. 11 6 12 9-2 55 Amplitudes of the Resultant Force at the Bracing Frequency Range: 0. 5... 200Hz Test No. 13 6 14 9-256 Ampli tudes of the Resultant Force at the Bracing Frequency Range: 0.5...200Hz Test No. 15 6 16 9-257 Amplitudes of the Resultant Force at the Bra"ing Frequency Range: 0.5...200Hz Test No. 17 6 18 9-258 Amplitudes of the Resultant Force at the Bracing Frequency Range: 0.5... 200Hz Test No. 19 6 20 9-2 59 Frequency Distribution of the Resultant Bracing Forces Test No. 3..10 9-260 Prequency Distribution of the Resultant Bracin g Forces Test No. 11 6 12 9-261 Frequency Distribution of the Resultant Bracing Forces Test No. 13... 18 Rev. 9, 07/85 9P-16

PROPRIETARY SECTIONS 9 4, 9 5 AND 9 6 G KM-II-M STE AM BLOHDO'8N TEST TABLE OF CONTENTS 9 4 TEST P.ESU LTS 9.4 1 Description of Break Transient 9 4 1.1 MSL Break Steam Mass Flow Transient 9.4 1. 1. 1 RCL Break Steam Mass Plow Transient 9 4.1.2 MSL Break - Air Content Transient

9. 4 1.2.1 RCL Break - Air Content Transient 9 4.1 3 MSL Break Drywell and Suppression Chamber Pressure Tra nsie nt 9.4 1.3.1 RCL Break Drywell and Suppression Chamber P re ssure Tr ansi en t 9 4.1 4 MSL Break Temperature Variations in the Suppression Pool 9-4.1.4.1 RCL Break - Temperature Variations in the Suppression Pool 9.4. 2 Dynamic Pressure Loads at the Pool Boundary 9 4.2. 1 General Observations 9.4.2 1.1 9.4 9 4 2

2 1.1.1

1. 1. 2 Osc illa ti f Frequency (o Oscillations) Analya s on Frequ en cies Event Fre que ncie s 9 4'.2 1. 2. ~

S tati sti cal Evaluat ion

9. 4. 2. 1 2.1 Amplitude Components of the Low-Frequency and High-Frequency Bands 9 4.2. 21 2 Frequency= (o f Occurence) Distributions 9 4-2. 1.2 3 S tatistical Characteristics

9. 4. 2. 1. 3 Pa ra me te r In f1 uen ces 9 4 2. 1.3. 1 In fluence of the Break Cross-Sectional Area 9 4 2 1 3.2 Influence of the Initial Pool Temperature 9.4. 2. 1 3. 3 Influence of the Initial Air Content in the Dr ywell 9;4.2.1 4 Correlation of Film and Pressure Recordings 9.4 2.1 5 Repeat Tests and Reproducibility of'he Results 943 Loads on the Vent-Pipe Bracing, Dummy Quencher and I-Beam 9 4 3 1 Statistical Evaluation 9 4 3 1 1 Frequency Distribution of the Resultant Vent-Pipe Bracing Forces 9.4 3.1-2 Distributions of Direction of the Resultant Vent-Pipe Bracing Forces 9 4.3 1 3 Frequency Distributions of the Resultant Bending Moments at the Dummy Quencher 9-4 3.1.4 Direction Distribution of the Resultant Bending Moments at the Dummy Quencher 9 4.3-1 5 Frequency Distributions of the Forces on the I-Beam 9 4.3 1 6 Statistical Characteristics

9. 4 3.2 Parameter Influences Rev. 9, 07/85 9P-1

PROPRIETARY

9. 4.3.2. 1 Influence of the Break Cross-Sectional Area 9.4.3.2.2 Influence of the Initial Pool Temperature 9.4 3-2.3 Influence of the Initial Air Content in the Drywell 9.4.3.3 Frequency (of Oscillation) Analyses 9.4 Summary 9 5 Data Analysis and Load Specification 9 5.1 Introduction 9.5.2 Data Analysis 9.5.2 1 Dynamic Pressures .in the Pool and Their Physical In terpreta t ion 9 5.2. 2 Evaluation of the Chugging Frequency at GKM II-M

9. 5. 2. 3 Evaluation of the Pressure Oscillation Frequencies in t he Pool 9.5.3 SSES LOCA Load Def inition 9.5.3.1 Selection of the GKM II-M Time Segments to Be Sourced

9. 5.3.1.1 Description of the Evaluation Procedure 9.5.3. 1. 2 Selection of the Chugging Pressure Time Histories 9.5-3. 1. 2. Selection of the CO Pressure Time History 9.5.3 1 3 Verification of the Selected Time Segments 9.5.3.1 3 1 Chugqing Pressure Time Histories 9.5 3. 1. 3. 2 CO Pressure Tine History 9.5.3 1 3 2 1 CO Evaluation 9 5.3 1.3 2 1 1 Comparison o.f the CO Amplitudes at GKM-M 9 5.3. 1 3. 2 1.2 PSD Comparison 9 5. 3.2 Determination of the Amplitude Factors 9 5.3 2.1- .Generation. of the GKM II-M probability Density Distributions 9 5'3 2 "2 -'Calculation of th*e 9 ymmetr ic= axd Asy mme tric

'A'mplitud~'actors 9.5.3. 3 Determination of the SSES-Unique Chugging/CO Sources 9.5 3.4 Application Procedure for Calculating the Susquehanna SES Boundary loads 9 5 3 4 1 Symmetric Load Case 9 5.3 4 2 Asymmetric Load Case 9.5.3.5 Verification o f the SSES LOCA Load Definition

9. 5 3.5.1 JAERI Comparison 9.5. 3.5. 1. 1 Introduction 9.5.3 5. 1. 2 Description of Comparison Method 9 5 3 5 1 2-1 The JAERI Test Data 9 53 5 1. 2.2 Ch ug So urces 9 5. 3. 5.1. 2 3 The JAERI Acoustic Model 9.5 3.5. 1 2. 4 Comparison Basis 9 5.3.5 1. 3 Results an d D iscuss ion

9. 5.3.5.2 Verification of the 50 msec Time Mindow 9 5.3 5.3 Response to the NRC Concerns Regarding the SSES Chugging Load Specification 9 5 3 5 3.1 Con tri bution o f the Asy mmet ric Ch ugging Load pecif ication to the Plant Structural Re sponse S

9 5.3 5 3.1.1 Comments on the Asymmetric Chugging Load Case 9 5 3 5 3 1 2 Generic and SSES Position HNP-2 Submittal 9 5 3-5.3. 2 Comparison'o f Minimum Variance Trials Using the SSES/GKM II-M Sources with the JAERI Data Rev. 9 07/85 9P-2

PROPRIET ARY 9.5. 3 5 .3 2.2 Comparison Procedure 9.5.3.5 .3. 2. 3 Results and Discussion 9.5.3.5 .3 2 3. PSD Comparison

9. 5. 3. 5 .3.2 3. Pre ssure Response Spectra (PRS) Com pari son 9.6 Ver ication of the Design Spec ifica tio n if 9 6.1 Evaluation of the DFFR CO and Chugging Load Specif ic ation 9.6 1. 1 Containment Acceleration Response Spectra Compazis on 9.6. 1. 1. 1 LOCA (DFF R) vs. LOCA (KHU) Acceleration Response Spectra Comparison 9 6 1.1 2 SSE + SRV{ADS) + LOCA (DFFR) vs. SSZ + SRV{ADS) +

LOCA {KRU) Acceleration R esponse Spectra comparison

9. 6. 1. 2 Evaluation of the Containment Response Specta Comp a rison 9 6.1. 2 1 Load Reduction Assessment 9.6. 1. 2 1 1 Reduction of Load Amplitude Factor 9-6. 1. 2.1. 2 Be-Selection of Pressure Time Histozies 9 6.1 2.1. 3 Adoption o f Mark II Owner' Group Load Metho dolo gy 9 6 1 2 Development of a New Chugging Load Methodology 9.6. 1 2 2 Plant Reassessment 9 6.1 3 Summary 9 6.2 Vezification o f the Mark II Single Vent Dynamic La terai Load Specification 9.6. 2. 1 Theoretical Determination of the Bracing Force at GKM II-M 9-6.2 1.1 Fini te El em en t Mo de 1 9 6. 2. 1 2 Model Assumptions 9.6. 2. 1.3 Fluid Representati.on 9.6 2 1 Structural Model 9.6.2. 1.5 Loading 9.6.2 1-6 Analysis Results 9.6. 2. 2 Bracinq Force Data at GKM II-M 9 6.2. 2 1 Measurement of the Bracing Forces 9 6.2. 2.2 Resultant Bzacinq Forces

9. 6. 2. 3 Comparison of the Theoretical and Measured Maximum Resu lta nt B racing For ce 9 6.3 Statistical Evaluation of the GKH II-M Resultant Bracing Force Data

6. 3.1 In trod uction 9

9 6. 3.2 Der ivat ion of a P rohabili t y th e measured Resultant B racing D en s ity Func ti f on rom For ces from the 1/6 MSL Tests 9.6. 3. 2. 1 General Consider ations

9. 6 .3. 2. 2 Application to the 1/6 MSL Tests 9.6.3.3 Determin ation o f t he Extra pola ted M azk II Impulse Rev. 9, 07/85 9 P-3

PROP RIET A RY SECTIONS 9.4, 9 5 AND 9.6 FIGURES Number Ti tie.

Steam Mass Flux vs. Time 1/6, 1/3 and Full MSL Breaks Test No. 5, 11, and 15 9-15a Comparison of the Measured Steam Mass Flux with the Theoretically Calculated Transient Pull HSL Break 9- 16 Steam Mass Flux vs. Time RCL Bzeak Test No. 1 and 2

9-1 6a Comparison of the Measured Steam Mass Flux with the Theoretically Calculated Transient - RCL Break 9-17 Steam Air Content vs. Time Fu11, 1/3 and 1/6 HSL Breaks Test No. 5, 11 and 15 9-1 8 Steam Ai r Content vs. Time 8 5% and 1 00% Ini tial Drywell Air Volume Test No. 5, 7 and 8 9-1 9 Steam Air Content vs. Time RCL Break Test No. 1 and 2 9-19 a Steam Air Content vs. Steam Mass Flux for Various Break Si zes 9-2 0 Dzywell and Wetwell Pressure Transien ts Full MSL Break Test No. 5 9-20a Drywell and Wetwell Pressure Transients 1/3 HSL Break Test No. 11 9-205 Drywell and Wetwell Pressure Transients 1/6 MSL Break Tes t No. 15 9-21 Drywe 1/6 ll Breaks twellTestPressure5, HSL and. We No Transients 7 and 8 Pull, 1/3 and 9-2 2 Dzywell and Wetwell Pressure Tra nsients RCL Break Test No. 1 9-22a Drywell and Wetwell P ressure Transients RCL Break Test No. 2 9-23 Pool Temperatur Variations vs. Time Fu11, 1/3 and 1/6 MSL Breaks Test No. 5, 11 and 15 9-2 4 Pool Temperature Variations Full HSL Bzeak - Test No.

3, 5 and 9 Rev. 9, 07/85 9P-4

PROPRIETARY 9-25 Pool Temperature Variations 1/6 MSL Break Test No.

14, 15 and 20 9-26, Pool Temperature Variations RCL Break Test Ho. 1 and 2 9-27 Pool Tempezature Variations RCL Break Test No. 1 and 33 9-27a Pool Tempera ture vs. Steam Mass Flux for Various Break Sizes Transducer T 6.3 9-28 Compressed Pressure Time Histories Full MSL Break - Test No- 3 9-2 9 Compressed Pressure Time Histories Full MSL Break Test Ho. 5 9-30 Compressed Pressure Time Histories Full MSL Break Test No. 9 9-31 Compressed Pressure Time Histories 1/3 MSL Break Test No. 11 9-32 Compressed Pressure Time Histories 1/3 MSL Break Test No 14 9-33 Compressed Pressure Time Histories 1/6 MSL Break Test No 15 N

9-3 4 Compr essed Pzessu ze Time Histories 1/6 MSL Break Test No 20 9-35 Compressed Pressure Time Histories RCL Break Test No. 1 9-3 6 Com pressed Pressure Time Histories RCL Break Test Ho. 33 9-36a Identification Sheet For Figures 9-37 thru 9-54 9-3 7 Visicorder Ho. I Traces Test Nc. 3 9-3 8 Visi"order No. I Traces Test No. 3 9-39 Visicorder No. I Traces Test No. 5 9-4 0 Visicozder No. I Traces Test Ho. 5 9-41 Visicorder No. I Traces Test No. 9 9-42 Visicozder No. I Traces Test No. 9 9-43 Visi"order No. I Traces Test No. 11 Re v 9, 07/85 9P-5

PROPRIETARY 9-4 4 Visicorder No. I Traces Test No. 11 9-45 Visicorder No. I Traces Test No. 14 9-46 Visi"order No. I Traces Test No. 14 9-47 Visicorder No. I Traces Test No. 15 9-4 8 Visicorder No. I Traces Test No. 15 9-49 Visicorder No. I Traces Test No. 20 9-50 Visicorder No. I Traces Test No. 20 9-51 Visi"order No. I Traces Test No. 1 9-52 Visicorder No. I Traces Test No. 1 9-5 3 Vi sicorder Ho. I Traces Test N o. 33 9-5 4 Visic order No. I Traces Test No. 33 Power Spectral Density (PSD) and Cross Power Spectral Densi ty tCPSD) Equations 9-56 Power Spectral Densities Sensor: P6.7 Test No. 5, 11 and 15 9-57 'Power Spectral. Densities Sensor: P6.7 Test No. 3,-

5 and 9 9-5 8 Power Spectral Densities Sensor: P6.7 Test No. 14, 15, and 20 9-59 Power Spectral Densities Sensor: P6.7 Test Ho.

9-6 0 Power Spectral Densities Sensor P6.7 Test No. 33 9-6 1 Compressed Time Histories - T5. 6, LP5.1....LP5.5, P5.4 and P6. 4 Test No. 5 9-61a Identification Sheet For Figures 9-62 Thru 9-79 9-6 2 V isicorder N o. III Traces Test No. 3 9-63 Visicorder No. III Traces Test No. 3 9-64 Visi"order Ho. III Traces Test No. 5 9-65 Visicorder Ho. III Traces Test No. 5 9-66 Visicorder No. III Traces Test No. 9 Rev. 9, 07/85 9P-6

PROPRIETARY 9-67 ~

Visic ord e r No. III Tr aces Test No 9 9-68 Visicorder Ho. XII Traces Test No. 11 9-69 Visicorder No. IXI Traces Test No. 11 9-70 Visicorder No. IXI Traces - Test No. 14 9-71 Visicorder No. III Traces Test No. 14 9-72 Visi" orde r No. III Traces Test Ho. 15 9-73 Visicorder No. III Traces - Test No. 15 9-74 Visicorder No. III Traces Test No. 20 9-75 Visic order Ho. IIX Traces Test No 20 9-76 Visicorder No. XIX Traces; Test No. 1 9-77 Vi sic order No. III Traces Test No. 1 9-78 Visicorder No. III Traces Test No. 33 9-79 Visicorder No. IXI Traces - Test 'No. 33 9- 80 Time between Chug Events Mean Values. vs. Break Size

. 4 9-81 - Amplitudes and Mean Values of-Pressure Events Test ~ - =--

Ho. 5 and 6 9" 82 Amplitudes and Mean Values of Pressure Events Test No. 11 and 12 9-83 Ampli tudes and Mean Values of Pressure Events Test Ho. 15 and 16 9-84 Amplitudes and Mean Values of Pressure Events Test Ho. 1 and 2 9-85 Mean Values of Pressure Events Test No. 3...10 9-86 Mean Values of 'Pressure Events Test No. 11...20 9-8 7 Mean Values of Pressure Events Test No. 1, 2, 33 and 34 9-88 Amplitudes and Mean Values of Pressure Events Frequency Ranqe: 10...100, 12...100 and 15...100 Hz-Test Ho. 13 9-89 This fiqure has been intentionally left blank Rev. 9, 07/85 9 P-7

PROP RIETARY 9-9 0 This figure has been intentiohally left blank 9-9 1 Positive Pressure Amplitude Histoqrams Sensor: P6.4 Frequency Ranqe: 0 5...13 Hz Pull, 1/3, and 1/6 ISL Breaks 9-92 Ideative Pressure Amplitude Histograms - Sensor: P6.4 Frequency Range: 0. 5... 13 Hz Full, 1/3, and 1/6 MSL.

B'r eaks 9-93 positi ve pressure Amplitude Histoqrams Sensor: P6. 7 Frequency Ranqe: 0.5...13 Hz - Full, 1/3 and 1/6 MSL Breaks 9-94 N~eative- Pressure Amplitude Histcgrams Sensor: P6.7 Frequency Range: 0.5...13 Hz Full, 1/3 and 1/6 MSL Breaks 9-95 Positive Pre ssure Amplitude Histogra ms Sensor: P6.8 Frequency Range: 0. 5...13 Hz Full, 1/3, and 1/6 HSL Breaks 9-9& Negative Pressure Amplitude Histograms Sensor: P6.8 Frequency Range: 0.5...13 Hz Full, 1/3 and 1/6 MSL Breaks 9-97 Positive and. Neaative Pressure .Amplitude. Histoqrams-Sensor: P6 4 Frequency Range: 0 1...13 Hz RCL Break 9-98 Positive and Negat iye- Pressu re Am pli t ude 8 is to gra ms-Senso r: PG. 7 Frequency Ranqe: 0.1...13 Hz RCL Break 9-99 Positive and Negative Pressure Amplitude Histograms-Sensor: P6.8 Frequency Range: 0.1...13 Hz RCL Break 9- '] 00 Positive Pressure Amplitudes vs. Break Size Sensor:

P6.7 Frequency Range: 0.5...13 Hz Full, 1/3 and 1/6 MSL Breaks 9-101 N~eative Pressure Amplitudes vs. Break Size - Sensor:

P6.7 Frequency Range: 0. 5...13 Hz Full, 1/3 and 1/6 MSL Breaks 9-102 positive Pressure Amplitudes vs. Initial Pool Temperature Sensor: P6.7 Frequency Range: 0.5.. 13 Hz Full MSL Break 9-1 03 negative Pressure Amp1itudes vs Initial Pool Temperature Sensor: P6.7 Frequency Range 0.5...13 Hz Full HSL Break Re v. 9, 07/85 9P-8

PROPRIETARY 9-104 Positive- Pressure Amplitudes vs. Initial Pool Temperature - Sensor: P6 7 Frequency Range: 0.5...13 Hz 1/6 MSL Break 9-105 Negative Pressure Kmplitu6es vs. Initial Pool Temperature Sensor: P6. 7 Frequency Range: 0.5... 13 Hz 1/6 MSL Break 9-1 06 Positive Pressure Amplitudes vs. Initial Pool Temperature Sensor: P6.7 Frequency Range: 0.1... 13 Hz RCL Break 9-107 Negative- Pressure Amplitudes vs. Initial Pool.

Temperature Sensor. P6.7 Frequency Range: 0.1...13 Hz RCL Break 9- 108 Correlation of Pictures an d Pressure Traces P6.4 and P6.7, 269 sec After Test Start Test No. 15 9-1 09 Correlation of Pictures and Pressure Traces P6.4 and P6.7, 269 sec After Test Start - Test No. 15 9-110 Correlation of Pictures and Pressure Traces P6.4 and P6. 7, 269 sec A ft er Test Start Test No. 15 9-1 11 Correlation of Pictures and Pressure Traces P6.4 and P6.7, 269 sec After Test Start Test No. 15 9-1 12 Correlation of Pictures and -Pressure- Traces P6.4 -and,-

P6.7, 269 sec Af ter Test Start Test No. 15 9-1 13 Pictures of Vent Clearinq Process Test No. 3 and 18 9-114 Amplitudes of the Resultant Forces at the Bracing Frequency Range: 0 5...200 Hz - Test No. 5 and 6 9-1 Amplitudes of the Resultant Forces at the Bracing 15 Frequency Range: 0. 5... 200 Hz -'est No. 11 and 12 9-1 16 Amplitudes qf the Resultant Forces at the Bracing Frequency Range: 0.5...200 Hz Test No. 15 and 16 9-117 Amplitudes of the Resultant Forces at the Bracing Frequency Range: 0.5...200 Hz - Test No. 1 and 2 9-1 18 Histogram of the Resultant Forces at the Bracing-Frequency Ranqe: 0 5...200 Hz Test No. 5 9-1 19 Histogram of the Resultant Forces at the Bracing Frequency Ranqe: 0 5...200 Hz Test No. 1'1 9-120 Histoqram of the Resultant Forces at the Bracing Frequency Range: 0.5...200 Hz Test No. 15 Rev. 9, 07/85 9 P-9

PROP RZETAR Y 9-1 21 Histo'gram of the Resultant Forces at the Bracing Frequency Range: 0. 5... 200 Hz Test No. 1 9-122 Direction Distribution of the Resultant Forces at the Bracing Frequency Range: 0.5 ..200 Hz Test No. 5 9- 123 Direction Distribution of the Resultant Forces at the Bracinq Frequency Range: 0. 5... 200 Hz Test No. 11 9-1 20 Resul tan t Forces at the Bracing

- Test F re qe uncy Range:

0.5...200 Hz No. 15 9-125 Resultant Forces at the Bracing Frequency Range:

0.5...200 Hz - Test No. 1 9-126 Direction Distribution of the Resulting Force on the Vent-Pipe Bracinq Frequency Range: 05... 200 Hz Test No. 3 20 9-127 Amplitudes of the Resultant Bendinq Moments at the Dummy Quencher Frequency Ranqe: 0 5 ..200,Hz - Test No. 5 and 6 9" 128 Amplitudes of th Resultant Bending Moments at the Dummy Quencher Frequency Range: 0. 5 .. 200 H z Test No. 11 and 12 9-129 Amplitudes of the Resultant Bending Moments at the Zummy Quencher Frequency Range: 0.5-...200 Hz Test-No. 15 an d 16 9-130 Amplitudes of the Resultant Bending Moments at the Dummy Quencher Frequency Range: 0.5...200 Hz Test No. 1 and 2 9-131 Histogram of the Resultant Bending Moments at the Dummy Quencher Freguency Range: 0. 5... 200 Hz Test No. 5 9-1 32 Histogram of the Resultant Bending Moments at the Dummy Quencher Frequency Range: 0.5 ..200 Hz Test No. 11 9-133 Histogram of the .Resultant Bending Moments at the Dummy Quencher Frequency Range: 0.5...200 Hz Test No. 15 9-134 Histogram of the Resultant Bending Moments at the Dummy Quencher Frequency Range: 0.5...200 Hz Test No. 1 9-135 Direction Distribution of the Resultant Bendinq Moments at the Dummy Quencher Frequency Range: 0.5...200 Hz-Test No. 5 Rev. 9, 07/85 9P-10

PROPRIETAR Y 9-136 Direction Distribution of the Resultant Bending Moments at the Dummy Quencher Frequency Range: 0.5...200 Hz Test No. 11 9-137 Direction Distribution of the Resultant Bending Moments at the Dummy Quencher Frequency Range: 0.5...200 Hz-Test No. 15 9-1 38 Direction Distribution of the Resultant Bending Moments at the Dummy Quencher - Freguency Range: 0.5...200 Hz-Test No. 1 9-139 Amplitudes of the Forces on the I-Beam Frequency Range: 0.5...200 Hz - Test No. 5 and 6 9-140 Amplitudes of the Forces on the I-Beam Frequency Fanqe: 0.5...200 Hz Test No. 11 and 12 9-141 Amplitudes of the Forces on the I-Beam Frequency R anqe: 0. 5...2 00 Hz Test No. 15 an d 16 9-142 Amplitudes of the Forces on the I-Beam Freguency Ranqe: 0.5...200 Hz Test No. 1 9-143 Histoqrams of the Positive and Negative Force at the I-Beam Frequency Range: 0.5...200 Hz Test No. 5 9-144 Histograms of the Positive and Negati'ye Force at the I-Beam Frequency Range: 0. 5...200 Hz Test No. 11 I-p 9-145 Histoqrams of the Positive and Negative Force at the Beam Freguency Range: 0.5...200 Hz Test No. 15 9-146 Histoqrams of the positive and Negative Force at the.I-Beam Frequency Range: 0.5...200 Hz Test No. 1 9-147 Resultant Forces Measured at the Bracing vs. Break Size and Initial Pocl Temperature Frequency Range:

0. 5... 200 Hz MS L Br eak 9-148 Resultant Bending Moments at the Dummy Quencher vs.

Break Size and Initial Pool Temperature Frequency Range: 0. 5... 200 Hz MSL Break 9-14 9 Absolute Forces Measured at the I-Beam vs. Break Size and Initial Pool Temperature Frequency Range:

0.5...200 Hz MSL Break 9-150 Power and Cross Power Spectral Densities of the Bracing Loads Test No. 13 9-1 51 Power and Cross Power Spectral Densities of the Dummy Quencher Loads Test No. 1 Rev. 9, 07/85 9P-11

PROPRIETARY 9- 152 Pressure Time Histories at the Pool Wall and Vent Test Xl 9-1 53 Pressure Time Histories at the Pool Mall and Vent Test Xl 9-154 Pressure Time Histories at the Pool Wall and Vent Test Xl 9-155 Observed Oscillation Frequencies in the Pool vs. Time Test Xl 9-1 56 Velocity of Sound vs. Pool Air Content 9-157 Air Content in the Mater Derived From the Pool Frequencies Test Xl 9-158 Pressure Time Histories at Pool Wall and in Vent Test No. 33 fancies 9-1 59 Chugging Frequencies Test Nos. 4 and 10 9-160 Chugging Frequ Test No. 11 9-161 Chugqing Frequencies Test Nos. 13 and 20 9-162 Oscillatory Frequen'cies of Pressure Events Test Nos.

1 and 33 9-163 Oscillatory Frequencies of Pressure Events Test Nos.

3. .10 9-1 64 Oscillatory Frequencies of Pressure Events Test No.

11 9-165 Oscillatory Frequencies of Pressure Events - Test Nos.

13 and 19 9-166 Evaluation ideally of the Traces 9-167 Finally Selected Chugging Time Segment Pressure Time History No. 3 9-168 Fin ally Selected Chugging Time Segmen t Pressure Time History No 5 9-169 F Selected Chugging Time Segment P ress ure Time H istory No 6 9-170 Fin all y S el ected Chu ggi ng Time Seg men t Pressur e Time History No. 9 9-170 a PSD's of the Specified Chug Time Segments Rev. 9, 07/85 9P-1 2

PROPRIETAR Y 9-262aGh CO Pressure Traces From Various Test Break Sizes 9-263 Dependence of Largest Mean CO Amplitude on Steam Mass Flux 9-264 Compa.rison of the Envelope of the PSD's of PTH No. 14 with the Mean of the PSD Envelope of Test No. 1 8 2 9" 265 PSD~s of PTH No. 14 with the time expansion and contraction factors 9-2 66 Comparison of Chug Strengths Observed in the JAERI 9- 267 JAERI Geometry- Actual and the One Used in the Acoustic Model 9-268 Comparison of the SSES Chugging Load Definition with the JAFRI Test 0002 Data at 1800 mm Elevation 9-269 Comparison of the SSES Chugging Load Definition with the JAERI Test 0002 Data at 3600 mm Flevation 9-270 C om pa ri son of the Theoretical Frequency D istributi on of the Resultant Bracinq Forces with the Test Data Test Nos. 13-18 9-'271 Schematic of the First Asymmetric Pool Node

'-272 Schematic -of ANSYS Model 9-273 Comparison of KWU Symmetric and Asymme tric ARS Curves Direction z - Node 841 9-274 Comparison of KWU Symmetric and Asymmetric ARS Curves Direction z Node 411 9-275 Comparison of KWU Symmetric and Asymmetric ARS Curves Direction z Node 771 9-276 Comparison of KWU Symmetric and Asymmetric ARC Curves Direction z Node 235 9-277 Comparison of KWU Symmetric and Asymmetric ARS Curves Direction x Node 841 9-2 78 Comparison of KWU Symmetric and Asymmetric ARS Curves - Direction x Node 411 9-279 Comparison of KWU Symmetric and Asymmetric ARS Curves Direction y Node 135 9-280 Comparison of KWU Symmetric and Asymmetric ARX Curves Direction x Node 131 Rev. 9, 07/85 9P-17

PROP R IETAR Y 9-2 81 Comparison of Flexible Hall Sonic Speeds in GKN II-N and JAERI Test Facilities 9-282 Comparison of the Minimum PSD Envelope of 20 Nin. Var. STS Trials with Max.

PSD Envelope of JAERI Data at 1 8 M Elevation . 1.3 Amplitude Factor 9-2 83 Comparison of the Minimum PSD Envelope o f 20 Min. Var. STS Trials with Nax.

PSD Envelope o f J AERI Da ta at 1.8 N Elevation 1.3 Amplitude Factor 9-28 4 Comparison of the Minimum PSD Fnvelope of 20 Min. Var. STS Trials with Max.

PSD Envelope of JAERI Data at 3.6 M Elevation 1.3 Amplitude Factor 9-285 Comparison of the Minimum PSD Envelope of 20 Hin. Var. STS Trials with Nax.

PSD Envelope of JAERZ Data at

3. 6 N Elevation 1.3 Amplitude Factor 9-2 86 Comparison of the Minimum PSD Envelope of 20 Min. Var. STS Trials with Max. PSD Envelope of JAERI Data at 1.8 H Elevation 1.95 Amplitude Factor 9-2 87 'Comparison of the Minimum- PSD Envelope of 20 Min. Var. STS Trials with Max PSD Fnvelope of JAERI Data at 3.6 H Elevation 1.95 Amplitude Factor 9-2 88 Comparison of the Minimum PRS Envelope f oz the

20. Nin. Var. STS, Trials with the Max. PRS Fnvelope of the JAERI Data 1.3 Amplitude Factoz with 4ii Damping 9-289 Comparison of the Minimum PRS Envelope for the 20 Nin. Var. STS Trials with the Nax. PRS Envelope of the JAFRI Data 1.95 Ampli tude Factor wit h '%

Damping 9-290 Comparison of the Minimum PBS Envelope for the 20 Nin. Vaz. STS Trials with the Max. PRS Envelope of the JAERI Data 1.3 Amplitude Factor with 7% Damping 9-291 Comparison of the Hin. PRS Envelope for the 20 Hin. Var. STS Trials. with the Nax. PRS Envelope of the JAERI Data 1.95 Amplitude Factor with 7T Damping Re v. 9, 07/85 9P-18

PROP RIETAR Y SECTIONS 9 Qt 9 5 AND 9.6 TABLES Number Title 9-1 See Non-Proprietary Section 9.0 to 9-5 9-6 Evaluation Seg me nts 9-7 Dominant Frequencies Measured at Pressure Transducer P6.7 Mean and Maximum Values of the Positive Pressure 9-8'-9 Amplitudes - Frequency Range: 0.1...13 Hz Mean and Maximum Values of the Negative Pressure Amplitudes Frequency Range: 0.1...13 Hz 9-10 Mean and Maximum Values o the f P ositiye Dynami c Pressures Frequency Range: 0. 5...13 Hz Test Xl 9-11 Loads On Submerged Structures Frequency Range:

0;5...13 Hz 9-12 Effect of Steam Mass Flux on the --Mean CO Pressure--

Amplitudes According to Break, Size 9-13 CO Time Seqments Chosen for PSD Comparison 9-1 0 Nodal Points Enveloped at Each Elevation Re v. 9, 0 7/85 9P-19

PROPRIETARY 9-. 4 TE ST I? ESULTS This section provides a compilation of the test results for the LOCA steam condensation tests conducted in the GKM II-M test facility in Mann'heim, Hest Germany by Kraftwerk Union. Twenty-two test runs were completed during the period from October 1979 to Mare h 1-980..

Subsection 9.4.1 presents the time histories of important test, parameters (i.e., steam mass flow, steam air content,Recirculation etc.)

measured durinq selected Main Steam Iine {NSL) and Line (RCL) break test runs. Subsection 9. 4. 2 contains a description of the test results for the dynamic pressure loads measured at the pool boundary. Subsection 9.4.3 documents the loads on t he vent pipe bracing a nd sub me rged st'r uct ures in the water pool. Finally, Subsection 9.4.4 summarizes the results of the tests. This information is provided in the form of tables, fiqures and actual Visicordez Traces.

9.4 1 Description 5 of Break Transients Once the test facility is prepared per Subsection 9.2.1.1 for the MSL break or Subsection 9.2.1.2 for the RCL break, the rupture disks are broken and the steam flows via the discharge line into the drywell of the test tank S3. The steam mixes with the aiz initially in the dzywell, pressurizes swell the drywell aiz and initiates the vent clearing and pool phases of the tra nsie nt.

Subsequently, the condensation phase occurs. This phase can be subdivided into a seqment of air-poor condensation (air content of steam sufficiently reduced such that further reduction in air content does not produce higher pressure loads) and later into a seqment of gir-free condensation.

The processes involved in the condensation of the air poor and air-free team in the water pool and the dynamic pressure loads that occur during these processes were the main points of investiqaton in the GKM II-M Condensation Tests 2.4,1 1 NS$ Break Steam Nass Flow Transient Fiqure 9-15 shows representative plots of the blowdown histories for a full NSL break {Test Ho. 5), a l/3 NSL break (Test No. 11) and a 1/6 MSL break (Test Do. 15) . Each plot represents the steam flow transient into the drywel'1 tank S3. Since quasi-steady-state conditions prevail a f ter the initial drywell pressurization, this can be interpreted as the mean mass flow through the vent pipe.

Upon comparing the "large>> and "small" MSL transients, evident that the small transient is covered by the ead it is phase (t 55 sec) of the large transient. Of course, a time interval of Rev. 9, 07/85 9P-20

PRGPRIETARY approximately 400 sec is then compressed to approximately 20 sec.

In addition, the "medium" MSJ. transient is covered by the end phase of the large transient in which a time interval of approximately 200 sec is compressed into approximately 40 sec.

Figure 9-15a compazes the envelopee of the full MSL transients run in the test stand (Test Nos. 3-10) to the theoretically calcula ted f ull MSL transient for SSES (see Subsection 9.3) .

This figure indicates that the test transients match the theoretical curve very closely. In. the 10 to 30 sec time frame, the full MST. transients exhibit some values below the theoretical curve, reachinq at the most about 20%. However, since the theozetical curve represents a conservative upper bound, this sliqht deviation is not considered significant.

9 4,1,], 1. RCL Break Steam'ass -Plow Transient

~

Figure 9-16 illustrates the steam mass flux transient for the RCL break Tests No. 1 and 2..

Both transients display an initial monotonic decrease that changes into a quasi-steady-state plateau at about 8 sec after test st'art and then at about 18 sec after test start resumes the monotonic drop until the termination of the blowdcwn. The tzansients in Tests No. 33 and 34, which were run with a higher initial pool temperature of 55oC, are identical to the blowdown history in Test No. 2.

'-Fiqure 9-16a shows a comparison of the envelopee -of--the RCL---

transients with the theoretically calculated transient for SSES (see Subsection 9. 3) . The test curves conservatively bound the theoretical tra'nsient in both duration and mass flow density.

The time elonqation in the monotonically decreasing steam flow prior to the plateau in the test curves results in the mean temperature of the pool after 10 sec being a few degrees higher than for the desiqn transient and the drywell air -being flushed over into the suppression chamber earlier than the design transient. The steeper monotonically decreasing flank after the plateau in the test curves only reduces the duration of this phase from about 23 to 18 sec. Thus, the test runs conservatively represent the RCL break transient.

J 9,4,1,2. MSI.. /reap Air Content Transients Another important test parameter is the air content in the steam.

It depends on the air content of the'rywell and on the steam supply, and has an influence on the chugging loads in the pool.

Foz that reason,- emphasis was placed on adequately simulating the plant parameters when planning the tests.

Fiqure 9-17 shows the time histories of the air content of the steam flowing throuqh the vent pipe. This figure indicates the Rev. 9, 07/85 . 9P-21

PROP RlET ARY beqinninq of air-free condensation for each break size as follow s:

Larqe transient beginning at about 15 sec

,tedium transient beginning at about 50 sec Small Transient beginning at about 100 sec From these times on, the composition of the steam remains essentially consta nt.

As previously described in Subsection 9.3 tests were also performed with a drywell air content reduced by about 15% (see Table 9-6):

Test No. 7 and 8, large transien t Test No. 17 and 18, small transient As examples, Figure 9-18 shows the aiz content transients foz the full NSL blowdown with 100% (Test No. 5) and approximately 855 air in the drywell at test start (Test No. 7 and 8) . A comparison of the duration of air flow reveals no significant difference in the beginninq of air-free condensation. The same is also true for the small transients.

9 4.1. 2.1- RCL Break Air Content Transients Figure 9-19 shows the time variation of the air content in the vent pipe for RCL break Tests No. 1 and 2.

The entire quantity of air has been flushed over after approximately 12 sec, so that only in the initial phase of constant mass flow rate (see Subsection 9. 4.1.1.1) is there still a very low percentage of aiz. Therefore, the largest portion of the steam transient plateau and subsequent decreasing steam mass flow occur in the phase of aiz-free condensation.

Figure 9-19a correlates the air content in the steam to the steam mass flow densities. This figure clearly shows the differences between the RCL transients and the other transients.

9.4.1.3 HSL Break Drywell and Suppression Chamber Pressure Transients The pressure time histories in the drywell and suppression chamber for the full HSL break, 1/3 NSL break and 1/6 NSL break are compared in Figures 9-20, 9-20a and 9-20b (Test No. 5, Test No. 11, and Test No. 15, respectively) .

The transient increase of the pressures is terminated when the air from the dzywell has been completely purged into the Rev. 9, 07/85 9P-22

PROPRIETARY suppression chamber (see Subsection 9.4. 1. 2) . The static pressures in the two chambers then remain approximately constant.

In addition, a closer examination of the data shows distinct pressure fluctuations (at about 0 5 to 1 Hz) in both pressure time histories throughout most of the transient. They reach a maximum of +0.15 bar in the drywell and a maximum of +0.05 bar in the suppression chamber and are out of phase with each other.

Two examples of the tests with reduced initial air content in the drywell are shown in Fiqure 9-21. Here, the pressure-time histories for Tests No. 7 and 8 and for Test No. 5 are compared with each other. There are practically no differences. The same holds true for the corresponding tests with the small transient.

The static pressures in the two chambers at test start and test end are compiled in Table 9. 5. The final pressures in the air reqion of the suppression chamber are 2.7 to 2.9 bar for a 100%

air content in the drywell and'.4 to 2.5 bar for an air content of approximately 85%.

9.4.1.3.1 RCL Break t)rywell and Suppression Chamber Pressure

.-Transients The pressure-time histories in the drywell and suppression chamber for Tests No. 1 and 2 are plotted in Figures 9-22 and 9-22a, respectively. The data indicates an absence of the pressure fluctuations described i;n Subsecti.on 9. 4.1.3 during the phase of constant mass -flow, lasting from approximately' sec= to.18 sec after test start. The segments before and after that, which correspond to the transient steam-mass-flow variations, exhibit the pressure oscillations described in Subsection 9.4.1.3 The static pressures in the two chambers at the beginning and at the end of the RCL breaks are indicated in Table 9-5. The final pressures in the air =reqion of the suppression chamber are 2.8 to

3. 0 bar.

1 4. QgL Break - temperature Variations in the Suppression Pool For the MSL breaks, three representative measuring points were selected for the presentation of the temperature variations in the water region of the suppression chamber: T 6 1 for the upper region, T 6.3 for the middle region and T 6 8 for the lower reqion of the pool (see Figure 9-5) .

As examples, Figure 9-23 shows the transient heating and mixing of the pool for the different MSL breaks, Tests No. 5, The initial pool temperature was approximately 33~C.

ll and 15.

Zxamples of temperature variations for the three different initial pool temperatures (24<C, 32oC and 55oC) are illustrated in Fiqure 9-24 for the full MSL break (Tests No. 3, 5 and 9) and Rev. 9, 07/85 9P-23

PBOPHIZTABY in Fiquze 9-25 for the 1/6 HSL bzeak (Tests No. 14, 15 and 20)'.

All other initial parameters are the same for each test.

9.4.1.4.1 RCL Break Temperature Variations in the Suppression Pool Fiqure 9-26 shows the temperature vaziations in the water region of the suppression chamber for Tests No. 1 and 2. Here again,,

representative measurinq points were selected: T6. 1 for the upper region, T 6. 3 foz the middle region and T 6.8 for the lower reqion of the pool (see Figure 9-5).

Figure 9-27 is a comparison of the variation of the temperature for the high initial pool temperature of 540C for Test No. 33 and the low initial pool temperature of 34~C for Test No. l.

Pzactically no difference exists Fiqure 9-27a shows a comparison of the pool temperature versus steam mass flux for various break sizes. Also marked on the fiqure are the points at which the air content passes through the values 10%, 1",o and O.lÃ.

9,4 2 Qyngmic Pressure Loads at the Pool Boundary This subsection presents the positive and negative dynamic pressure amplitudes Eased on statistical evaluations with a differentiated quantifiation of the parametric influences of break..size, initial pool temperature and inital air content in the drywell. As an introduction to that, a characteristic insiqht into the observed event and oscillation f requencies is i

given. F in a 1 ly, a vi sua 1 mpre ssi on o f the cond ensa tion processes is provided by a few excerpts f rom high-speed film record ings.

9. 4. 2. 1 C'eneral Impressions As a result of the original 4T tests, a LOCA load specification has been defined for the containment boundary loads (see Subsection 4.2.2). This load specification is classified into two distinct regions, consistent with the two types of pressure time histories observed during the 4T tests. The first region occurrinq during the early portion of the blowdcwn has been termed "condensation oscillations>> and the pressure fluctuations were observed to be sinusoidal in nature. The second region occurrinq later in the blowdown has been termed >>chugging>> and was characterized by periodically occurring, higher amplitude pressure fluctuations.

However, the GKN ZI-N test results indicate an absence of the two distinct condensation reqions observed durinq the 4T tests. A closer examination of the test data showed no clear-cut, consistent method for defining the >>condensation oscillation>> or

>>chuqqinq>> regions. Thus, in the discussions that .follow the Bev. 9, 07/85 9P-24

PROPRIETARY steam condensation phenomenon vill not be defined in terms of two distinct zeqions, but rather the pulsating steam condensations will be arbitrarily termed chuqqinq. This chuqqing for the purpose of load specification for SSES will cover the entire blowdown history previously broken into two distinct regions.

An insiqht into the strenqth of the measured pressure oscillations is provided by the survey diagrams in Figures 9-28 thru 9-36. In pressure-time histories hiqhly compressed in time, these presentations show the variation of the pressure loads in the water pool (measurinq points P 6.1 .. P 6.8, see Figure 9-5) for Tests 3, 5, 9, 11, 14, 15, 20, 1 and 33 throughout the entire test period. The selected examples are representative of the followinq parame ters:

Break size: Tests No. 5, 11, 15 and 1 (Figures 9-29, 9-31, 9-33 and 9 35); Full, 1/3, 1/6 HSZ breaks and RCL b reak s.

Initial water temperature: Tests No. 3, 5 and 9 (Fiquze 9-28, 9-29, and 9-30; Full MSL break) .

Initial water temperature: Tests No. 14, 15 and 20 9-33 and 9-34; 1/6 MSL break) .

(Fiquzes 9-32, Initial water tempezature: Tests No. 1 and 33 (Figures 9-35 and 9-36; RCL break) .

In most instances the pressure fluctuations-are-recognized as- ".-

havinq the character of events. That is suddenly excited pressure amplitudes followed by periods of smaller amplitude damped oscillations. It can be seen that the process o f chugging is stochastic duzing the phase of air-poor and air-free steam flow throu qh the vent pipe into the water pool.

There are larqe differences in the amplitudes of consecutive events. This fact is'lso demonstrated .by the video and high-speed films: Two consecutive events dif fer considerably with respect to the shape and maximum size of the steam bubble (see Subsection 9. 4. 2..1. 4) .

Essentially independent of the break size and the initial pool temperature, the pressure-time histories in Figure 9-28 thru 9-36 exhibit hiqher amplitudes in the period arcund the middle of the test, but lower a mpli tudes at the beginn ing and at the .end of the test., As examples, Figure 9-37 thou'-54 show details from these pressure-time histories on a more expanded time scale. These are copies of the oriqinal Visicordez No. I traces from two segments each in Tests No. 3, 5, 9, ll, 14, 15, 20 (NSL breaks), 1 and 33 (RCL breaks). Fiqure 9-36a identifies the sensor and scale used for each trace in Figures 9-37 to 9-54.

Rev. 9, 07/85 9P-25

PROPRIETARY

9. 4. 2. l. 1 Frequen~c~of Oscilla tionsQ Analysis To evaluate the dynamic pressure loads in the pool, the frequency spectrum is used in addition to the pressure-time history. It is common to display the frequency spectrum in the form of a so-called power spectral densi~t (PSD). It is calculated by means of a Fourier transformation using the equation indicated in Fiqure 9-55.

The power spectral densi ty is represented as the sguaze of the root-mean-square value of the relevant physical variable x(t),

with respect to frequency, and has corresponding dimensional units, e q., bar~/Hz. It is a measure of the intensity of the frequencies occurring in the region under consideration.

Foz example, the following frequencies can be determined:

The f requency of the spacinq between events (chugging frequency) .

The fzequency of the acoustic osci 1 la tion of the water in the pool {>>pool acoustics<<).

The freguency of the acoustic oscillation of the steam column in the vent pipe (>>vent acoustics>>).

Possibly present tape noise frequencies and/or those frequencies at which the intensity of the process is relatively small and which may be able to serve as fundamental fzeguencies for a low-pass or high-pass filter inq.

9 4 2,1,1 1 Oscillation Freguencies analyses were performed with the I'he Fourier System HP 5451 A.

This system provided a .freguency resolution of +0.5 Hz at the required upper frequency cut-off of 500 Hz and the desized seqment length of one second for the analysis of individual events. All PSD~s were calculated by the Hanning function.

In principle, the powez spectra were determined throughout the

>>entire>> test period for the pressure measuring points P 6.4 and P 6.7 (tank wall, Figure 9-5) and P 6.8 (tank bottom, Figure 9-5) for all tests. The beginning and end of each evaluation segment are indicated in Table 9-6. As shown below, the determined PSD's exhibit frequency-dependent differences between the individual pressure measuring points. The spectra calculated from the three pressure transducers pzactically coincide in the range from 6 to 10 Hz. T he higher the frequency, the smaller is the agzeeme nt, i.e., the lcw-f requency pressure oscill'ations (air bubble oscillation under water and vent acoustics) act in the pool beneath the vent pipe with the same intensity. In contrast, for the hiqhez-f zequency pressure oscillations there are distinct Rev 9, 07/85 92-26

PROPRIETARY differences between measurement locations which are due to local effects.

Representative exampl'es of these PSD's for the NSL breaks can be found in Fiqures 9-56 to 9-58 The PSD's of the pressure measurinq point p 6.7 vere selected foz presentation. In particular:

1 Fiqure 9-56 with the PSD's of Tests No. 5, 11 and 15 shows. the influence of the break size.

Figure 9-57 with the PSD's of Tests No. 3, 5 and 9 shows the influence of the initial water temperature f or the f ull HSL breaks..

Figure 9-58 with the PSD's of Tests No. 14, 15 and 20 shows the influence of the initial water temperature for the 1/6 NSL breaks.

Corresponding examples f or the RCL bzeaks are shown in the top part of Figures 9-59 and 9-60, with Test No. 1 having a mean initial water temperature and with Test No. 33 for the high initial water temperature. In addition, the middle and bottom parts of the fiqures illustrate the PSD's that were determined separately foz the phase of steady-state and monotonically decreasing steam flow, respectively.

The frequencies corresponding to the maxima of the determined power densities for .the measuzing -point -P 6.7 in -the range from.0 to 100 Hz for all tests are compiled in Table 9-7. These values were sorted by their intensity within a test (density "lass),

i.e., the first of the indicated maxima corresponds to the dominant frequency in all instances.

For the evaluation of the illustrated results, the frequency resolution of 0 5 Hz is of special significance in the lower range of frequencies.'urthermore, limits are imposed on the evaluation of low power densities by the noise level of the measurement and evaluation chain. For all three pressure signals (P 6.4, P 6.7 and P. 6.8), that limit lies on the average at 10-~

bar~/Hz in the lower fz guency range and at 10-~ bar~/Hz in the upper frequency range. The latter is the main reason for the restriction to only 100 Hz, since only values below these power densities occurred at higher frequencies.

9,4-.2.1.1. 2- -Event Frequencies Reference was already made in Subsection 9 4.2.1 to the periodically occurrinq, pulsatinq pressure fluctuations in the Pool. This process, termed "chugging,>> was the dominant form of condensation observed in the GEM II-M tests The water pen etra tes more or le ss in ten sel y into t he ve nt pi pe.

Rev. 9, 07/85 9P-27

PROP RZETA RY This process can be" seen particularly well at the measuring point T 5.6 (tempezature at the end of the vent pipe} and at the so-called "level probes>> (see Figure 9-5) . As an example, Figure 9-61 shows the variation of the measuring points T 5.6, LP 5.1 LP 5.5, P 5. 4 and P 6.4 throughout the entire test period of Test No. 5.

The periodic wettinq of the level probes (especially LP 5.1 LP 5.3) is clearly recognizable in the second half of the test. At the sa me time, the measurin g po int T 5.6 exhibi ts distinct temperature chanqes between>>warm>> and <<cold>> {the water emerges from the vent pipe and enters into the pipe, respectively) . The Figures 9-62 thzu 9-79 show detailed examples of this phenomenon.

These are copies of the original Visicorder No. III traces for two segments each from Tests No 3, 5, 9, ll, 14, 15, 20 (MSL breaks), 1 and 33 {RCL breaks) . Figuze 9-6la identifies the sensor aad scale used for each trace in Figures 9-62 thru 9-79.

In the first half of each test one can still recognize a compazable pzessure fluctuation in the pipe (measuring point P 5.4), but no>>pronounced>> wetting of the level probes. The steam mass flow ra te is obviously large enouqh here in comparison to the condensation rate that the steam/water interface does not penetrate into the pipe.

Mhereas the time intervals between chuqs or events depends very strongly on the steam mass flow transients, it varies less conspicuously within a given test for mass fluxes below 40 kg/m~s. A statistical evaluation of the pressure-time histories in the pool was performed to determine the mean event interval foz the different transients. The results are presented below:

Mean Event Bgegk Size Mean Fvent Interval F rendu encg Full MSL break 1 sec 1 Hz 1/3 MSL'reak 1.5 sec 0.7 Hz 1/6 MSL break 2 sec 0.5 Hz

~RCL break 1 sec 1 Hz These 9-80.

results fit smoothly into the curve illustrated in Figure

'I

~For the phase of steadily decreasing steam mass flow zates.

Rev. 9, 07/85 9 P-28

PROP RXET AR Y 9 4. 2.1 2 Statistical Evaluation The influences of individual test parameters can be essentially determined from mean values, since those values are significantly more typical than the magnitude of individual rare maximum values. In general, these trends are supported by the maximum values.

Therefore, to quantify the effects 'of parameters, the tests vere evaluated statistically by means of digital computers. Two main requirements vere imposed on that evaluation:

of amplitude oscillation fCorrelation components and the requency, i.e, a>> freguenc~-oriented>> evaluation.

Determination of the maximum values for each event, i.e., an>>event-oriented>> evaluation.

To meet the first requirement, tvo frequency bands were specified, taking into consideration the frequency analyses described in Subsection 9.4.2.1.1.1:

a low-frequency band with a range from 0.5 to 13 Hz, a high-frequency band with a range from 10 to 100 Hz.

The low-frequency band was to include all low-frequency amplitude components up to the vent acoustics (approximately 9 Hz) with correct. amplitude. The high-frequency band- vas--limited at .100 Hz in order to eliminate the locally occurring so called pressure spikes (see Figures 9-37 to 9-54), which can unrealistically falsify the load assumed around the entire containment.

The second requirement imposed on the statistical evaluation was satisfied hy interzoqatinq 'the pressure-time histories stored on the maqnetic tape for the positive and negative maximum value in constant intervals of time. The constant time intervals were selected accordinq to the mean event-intervals indicated in Subsection 9.4. 2.1.1. 2.

The pressure-time histories vere evaluated for the measuring points P 6.4 and P 6.7 (vali pressures) and P 6.8 (bottom pressure; see Figure 9-5) .

9.4.2.1.2.1 Amplitude Components of the Low-Freauency and High Frequency Bands The time seqment of each test indicated in Table 9-6 was evaluated by computer in the manner described in Subsection 9.4.2.1.2. The beginning of the evaluation segments was selected so that the>>static>> pressure rise was completed and thus only the d.ynamic pressure components were considered Rev. 9, 07/'85 9P-29

PROPRIETARY The det erm in ed pressure a mpli tudes (lo w- frequency, high-f requency) were plotted versus time in graphs for the pressure tzansducer P 6.7, using the magnitude of the maximum amplitude in each interroqation interval. Tests with indentical initial conditions (Test/Repeat Test: 1/2, 3/4, 5/6, etc.; see Table 9-4) were each combined in one qraph. This gives an insight into the reproducibility ("ee Subsection 9. 4. 2.1. 5) .

The formation of mean values was accomplished by defining large test seqments within which the arithmetic mean of the amplitudes was calculated. These test segments have diffezent leng ths for the individual test groups and are indicatd below:

Time Interval Break- Size for the Averagi~n Full MS L break 5 sec 1/3 NSL break 15 sec.

1/6 MS L break 30 sec RCL b.reak 5 sec Tests that were run with identical initial conditions were combined for the averaqing. The calculated mean values were assiqned to the middle o f the interval and also plotted in the qraphs (stra igh t line s) .

Examples of the results of the analysis for different break sizes are shown in Figures 9 81 to 9 84. Pairs of tests that were performed with an initial water temperature of 320 C were sel ect ed:

Illustrated Qgeyg- Si ze Full MSL break '/6 Test Pair ~

Figure 9-81 in 1/3 MSL break 11/12 Figure 9-82 1/6 MSL break 15/1 6 Figure 9-83 RCL break l/2 Fiqure 9-84 The top ha lf of each figure shows the results f or the low-frequency ranqe and the bottom half shows the results for the hiqh-f requency range. The time>>Os<<corresponds to <<Test Start <<

To facilitate comparison, the mean value straight lines are illustrated together in Figure 9-85 for Tests 3-10 (Full MSL break), Fiquze 9-86 for Tests 11-20 (1/3 and 1/6 MSL breaks), and Fiqure 9-87 for Tests 1, 2, 33 and 34 (RCL break) . Prom the mean-value curves it can be determined that in all the tests, Rev. 9, 07/85 9P-30

PHOPBIETAH Y both in the low-frequency and also in the high-frequency range, the pressure amplitudes have small values at test start, increase to higher values as the test period progresses and drop back to low values again at the end of the test (see Subsection 9.4.2.1) .

The mean pressure amplitude reaches a value of up to approximately l. 3 bar in the lov-frequency band and up to approximately 1.9 bar in the high-frequency band. Furthermore, the maximum deviations of the pressure amplitudes are larger in the hiqh-f requency band than in the lower band (see Figures 9-81 thru 9-84) .

Test No. 13 vas used to investigate the correct selection of the lover frequency cut-off of the high frequency band. The lower frequency cut-off was varied by using different high-pass filter settings, so that the followinq frequency ranges could be c ompa red:

10 100 Hz 12 100 Hz 15 100 Hz The result is illustrated in Figure 9-88 in the form of mean-value straiqht-line curves. It can be seen that the curves lie very close to one another and that the curve for the range from 10 to 100 Hz practically envelopees the other two curves.

Accozdinqly, the stipulation of the lower frequency- cut-off at 10 Hz can be considered conservative. Furthermore, the falsification of the results for the high-frequency band by "residues>> of the low-frequency band is negligible.

$ ,4,$ .1..2.$ Fr~euency /of occurrence) Distributions To prepare amplitude histograms for the low-frequency band, the positive and negative pressure amplitudes determined by computer vere classified into frequency of occurrence classes with a uniform class-interval of 0.05 bar. The histograms were normalized uniformly to 100 readings so as to permit direct comparisons of the tests with one another. Here too, the tests with identical initial conditions were combined in one histogram.

It vas found that the parameters >>water temperature and air content in the dzywell at test start>> have no significant influence on the shape of the distribution. Accordingly, all tests that were run with the same break size were also combined.

Figures 9-91 to 9-99 present comparisons of the freguency of occurrences distributions of the positive and neqative dynamic pzessure amplitudes for measuring points P 6. 4, P 6 7 and P 6.8 for the different groups of tests corresponding to the four break Rev. 9, 07/85 9P-31

PROPRIETARY sizes. Table 9A qives an overview of the assiqnments for the individual plots:

Bev. 9, 07/85 9P-32

PROPRIETARY TABLZ-9A: Histoqrams of the dynamic pressure amplitudes in the low frequency band {0.5 to 13 Hz)

Fiq. Break Tests Trans-ducers D ynamic pr essure Histogram Max.

a mpl it Mean ude s Size pos. neg. value value bar bar 9-91 Pull MSL 3... 10 l. 46 0 46 1/3 MSL 11/1 2 P 6 4 1. 51 0 36

- . 1~6. MSL 13...20 1. 12 0- 31 9-9 2 Fall MSL 3 - 10 -1.41 -0. 31 1/3 MSL 11/12 P 6 4 x -0.89 -0. 29 1/6 MSj, - 13 ..20. -0 94 -0 26 9-93 Full MSL 3.--10 1 91 0. 56 1/3 MSL 11/12 P 6.7 1. 60 0 38

-~1 6 MSL 13 ..20 1 38 0. 34 9-94 Full MSL 3. 10 -1.24 -0 37 1/3 M SL

--1/6 NSL 11/1 2 13 . 20 P 6.7 x, -0 95

-0.99

-0 30

-0. 28 9-95 Full MSL 3...10 1.99 0. 58 1/3 M SL 11/12 P 6.8 1.59 0. 39 186 NSL ~

13..20 1. 44 0 34 9-96 Full MSL 3 . 10 -1 54 -0 39 1/3 MSL ll/12 P 6 8 x -0 98 -0. 31 1/6- NSL ~

13 20 ~ -1 00 -0 28 0'.60 0. 24 9-97 RCL 1,2,33,34 P 6.4 x -0. 60 -0. 20 1.23 0. 39 9-98 RCL 1,2,33,34 P 6.7 x -0 68 -0. 28 1 31 0. 42 9-99 R CL P 6 8 x -0. 68 -0. 29 indicated in the last two columns of Table 9A is the maximum value and the point mean value'or the individual test groups.

The joint mean value was obtained by first averaging all the maximum amplitudes per mean event interval for a given test to obtain the mean value for the transient and then averaging these mean values for the same test group (i.e., 3... 10, 11/12, etc.).

9. 4.g 1.2 3- Sta tistical Ch aracterist ics Table 9-8 qi ves an overv iew of the mean and maximum values of positive pressure amplitudes measured by transd ucers P 6.4, P 6.7 and P 6.8 in the low-f requency band for all 22 tests. Likewise, Table 9-9 presents the negative mean and maximum values for all 22 tests.

R ev. 9. 07/85 9P-33

PROPRIETARY The mean values were determined hy the grouped data procedure, using the following equation:

where: P = Mean value P=

3.

Mean value of each class n.= Frequen cy o 3.

f eac h class The maximum values at the different measuring points need not have occurred at the same time, although that is generally the case.

The sensitivity of the e valuation with respect to the interrogation interval was investigated by using the shakedown Test llo. Xl so as to he able to specify the procedure for further evaluation of the GKM IZ-M tests. For that test, which corresponds to a 1/6 MSI. break the statistical analysis was perf ozmed for an interrogation>>Maximum value every 2.0 sec>>

(mean 'event interval) as well as an interrogation>>Maximum value every l. 75 sec>> (most frequent event-interval) Table 9-10 indicates that the interrogation interval has only a slight effect on the results. The mean values f or measurinq points P 6.4, P 6.7 and P 6.8 differ from one another by only about 3'$ in the two interroqations. This investigation has shown that sufficiently exact results are obtained with the interrogation

>>maximum value eve~r mean. event-interval>>

magj,mum- erron- of the single value of approximately +0.1 bar can be indicated for the maximum pressure amplitudes listed in Tables 9-8 and 9-9. This indication is based on an error estimate which is composed of the systematic measurement error, recording error and evaluation error {see Subsection 9. 2 2.7)

Assuming that the errors of the dynamic components practically cancel out in the formation of mean values, an order of magnitude of approximately +0.02 bar remains for the maximum error of the mean value. This residual error is composed of the possible zero-point shift of the analog measurement signal at the computer input and the error band of the method used for classification and averaqinq.

Rev. 9, 07/85 9P-34

PROPRIETARX 9 4.2. l. 3. Pa came ter Influences ~

The evaluation of possible parameter influences discussed in the next three (3) subsections is based on the mean value of the pzessure amplitudes for the entire duration of the test (see Subsection 9.4. 2.1.2.3 and Table 9-8 6 9-9) .

9.4.2.1.3.1 Influence of the Break Cross-Sectional Area The most important goal of the GKM II-M tests, namely to determine the dependence of the steam condensation loads on the postulated break size, was achieved.

V Fiqures 9-100 and 9-101 show a plot of the positive and negative mean values vs. the simulated break cross-sectional area for the measurinq point P 6.7. The values foz all the MSL break tests are included in these plots. The result is an increase of the pressure values with increasing break size. The positive mean values increase f rom approximately 0.35 bar for the 1/6 MSI, break to approximatley 0.55 baz from the full MSL break. In contrast, the negative mean values increase only f rom approximately 0. 25 ba r to a pproxi ma tel y 0. 35 bar.

The positive and negative mean values of the measuring point P 6.7 as a function of the initial pool temperature are illustrated in Fiquze 9-102 and 9-103 for the full MSI. breaks, in Figures 9---.-

104 and 9-105 for -the 1/6 MSL- breaks and in -Figures 9-106 and 9.

107 for PCL breaks. With the exception of the positive pressures for the RCL breaks in Figuze 9-106, the figures show no influence or only a slight influence of the initial pool temeprature on the maqnitude cf the pressure amplitudes. The sliqhtly more distinct increase of the positive values for the RCL breaks is attributed primarily to the higher level of amplitudes measured during the phase of steady-state steam mass flow in Tests No. 33 and 34 with.

the hiqher initial pool temperature at test start.

~9. ~. 1.3+ . Influence of- the Initial Air Content in the Drgwell Subsection 9.4.1.2 indicates that a 15$ reduction in the initial drywell air content has no siqnificant influence on the beginning of. air-free condensation. In addition, Figures 9-102 to 9-105 reveal tha t the resultinq 0 4 bar lower back-pressure in the suppression chamber associated with the reduction in initial air content has no siqnificant influence on the dynamic pressure loads. Tests No. 7 and 8 (Full MSL break, Figures 9-102 and 9-103) and Tests Ho. 17 and 18 (1/6 MSL break, Figures 9-104 and 9-.

105), performed with a reduced air content at a mean initial pool temperature of 32oC, corn parison tests.

fit readily into the scatter zone of the Rev. 9, 07/85 9P-35

PROPRIETARY Therefore, no significant dependence of the pressure amplitudes can he found in the zanqe of suppression-chamber pressure having relevance for SSFS.

9.4.2.1 '4- Correlation of Film and Pressure Recordings As previously mentioned, Stanford Research Institute International (SR I) pzovided two LOCAM, M odel 51-0003, high-speed cameras for filming the process in the pool.

These two cameras were mounted in fzont of the bull'-eyes of the large manhole, approximatley 0.6 m below the end of the vent pipe (see F iqure 9-3) .

The correlation o'f the f ilm recordings wi th the data recozdings on Visicozders and magnetic tape was accomplished by recording the same time signal on these recording devices.

Test Ifo. 15 was used for the f ollowing assessment. The parame ters at test start are listed in Table 9-5.

Fiqures 9-108 to 9-112 show a sequence of 10 instantaneous photographs (called>>Pictures<< in the following) during the chuqqinq phase. The pressure-time histories recorded for this event at the pressure measuzinq points P 6. 4 and P 6.7 (Figure 9.5) aze illustrated below each picture. The time>>0 ms>>

corresponds to 269 sec after test start. The, steam mass flow rate at that instant is 65~C ll kg/m2s and the pool tempezatuze is 60 The 9-108 f rame rate to 9-112 was 200 pictures/'sec. A reference line in Figures provides the conn ection between picture and pressure trace The process of chugginq can be followed well in Figure 9-108 to 9-ll2. At the beginning of the chugging phase, a>>hemispherical>>

steam hubble develops,'t the end of the pipe.

constant, but rather it It does not remain alternately expands and then collapses again. The actual origin of the pressure events is the rather violent condensation of the steam when it contacts the subcooled water (steam bubble collapse) at the outlet of the vent pipe.

Due to the resultinq underpressure, there is a more or less intense movement of the water into the pipe and back out again

(<<chugqinq>>) . As described in Subsection 9.4.2..1.1.2 this proce s i" zepeated at almost identical intervals.

gow to the =

f ilm recordings:

The reference is Picture a in Fiquze 9-108 at the relative time 0 ms. A str iated f ron t at the en d o f the vent pipe indicates the beginninq o f a steam bubble f orma tion.

Rev. 9, 07/85 9P-36

PROP RI ETAH Y In Picture b in Fiqure 9-108 at the relative time 200 ms, the development of a steam bubble is already clearly discernibly.

Only extremely small pressure fluctuations are indicated in the pressure traces P 6. 4 and P 6. 7.

In Picture c in Fiqure 9-109 at the. relative time 300 ms, the steam bubble has enlarged and begins to enclose the end of the p ipe.

In Picture d in Figure 9-109 at the relative time 400 ms, the steam bubble has enlarged f urther and has completely enclosed the end of the pipe. The boundary surface between water and steam can he recognized.

In Picture e in Figure 9-110 at the relative time 500 ms, the steam bubble beqins to collapse and to pulsate. This pulsation and the incipient underpressure phase is clearly recognizable in the pressure traces P 6.4 and P 6.7.

In Picture f in Figure 9-110 at the relative time 550 ms, the steam bubble has diminished further. The end of the pipe is aqain visible.

At the relative time 600-ms in Picture q in Figure 9-111, the steam bubble has shrunk further and is visible only in a shadowy manner. The underpressure phase is clearly recognizable in the pressure traces.

In Picture h in Figure 9-111 at the relative -time-650 ms,- a small-part has pinched off from the .almost completely condensed steam bubble and has bequn to pulsate. This pulsation is visible in the pressure traces P 6.4 and P 6.7.

In Picture i in Figure 9-112 at the relative time 700 ms, the steam bubble has completely condensed except for the pinched-off part.

In Picture j 'in Fiqure 9-112 at the relative time 800 ms, the steam bubble has completely condensed and the water has entered into the pipe.

In addition, Fiqure 9-113 shows two instantaneous photographs of the vent-clearinq process for different break sizes. The top Picture represents the process at 0.37 sec after test start for a f ull llSL break (Test No. 3) . The bottom Picture shows the vent clearing at 1.08 sec after tests staxt for a 1/6 HSL break (Test No. 18) . These two pictures illustrate the expulsion of water and water-air mixture, respec tively.

Rev. 9, 07/85 9P-37

PROPRIETARY 4,2. 1,5 ReReat Tests aud.a~eroducibilit2 cf the Results To verify the reproducibility of the measurement results, the Test Matrix (Table 9-4) stipulated a repetition of each test resultina in 11 pairs of tests.

A comparison of the associated repeat tests with respect to the mean values of the positive dynamic pressure amplitudes at the measuring point P 6.7 (Table 9-8) gives an insight into the qua lity of the repro due ib ility.

The ~mximum deviations from the joint mean value of the pair of tests in each instance are as follows:

for the full MSL breaks +0.04 bar or +7~

for the 1/3 MSL breaks +0.02 bar or +6%

for the 1/6 NSL breaks +0.02 bar or +6%

for the RCL breaks +0.05 bar or +15%

Figures 9-81 to 9-84 provide an idea of the scatter of the individual values (see Subsection 9. 4. 2. l. 2) . The test results follow identical trends within the range of the scatter.

According ~ we can state:

If the initial conditions of the tests are adjusted in a controlled manner within the prescribed tolerances, then the test o

results are reproducible.

9.4.3 Loads in the Vent-Pipe Bracing, Dummy Quencher and X-Ream In this subsection, a discussion of the results of the measured loads on the vent-pipe bracinq, dummy quencher and. I-beam are presented. In addition, an evaluation of the influence of the test parameters on the load is made.

The measuring points provided for them are (see Figures 9-7 to 9-9):

SG 6.1 and 6.2: Longitudinal forces in the bracing SG 6.7 and 6.8: Bendinq moments at the dummy quencher LC 6.1: Bearing forces on the I-beam Strains were measured at all the measuring points. From the measured strains, the loads that produced those strains were calculated. Thus, in all cases we are dealing with pure Rev. 9, 07/85 9P-38

PROPRIETARY measurement values that are not directly transposable to plant conditions but rather still require an evaluation.

9.4.3.1. Statistical Evaluation Analogous to the statistical evaluation of the dynamic pressure loads in the pool, the loads on the internals'axe illustrated by exemplary plots of the forces and bending moments calculatd from the measured strains vs. time and the resulting freguency distributions. In addition, the distribution of directions of the resultant bracing forces and bending moments at the dummy quencher is presented by means of <~point correlations."

9. 4. 3.1. 1 Frequency Distributions of the Resultant Bracing Forces ~

To measure the longitudinal forces in the bracing, the strain qauqes SG 6.1 and SG 6.2 were mounted on the bracing pipes (see Figures 9-6 and 9-7). For symmetry reasons and in order to eliminate bending strains, two strain gauges displaced by 180~

were mounted alonq the bracinq pipe and connected to form a complete bridqe to measure the normal forces.

The strut forces P were calculated from the measured strains by usinq the followinq equation:

F = A where: A = Cross-sectonal--area-of strut pipe o = Normal stress Hooke's law is valid for the elastic range and gives:

a=E ~ c where: F. = No dulus o f clast icit y c =pm/m measured strain Combin,ing the above equations yields:

F =AE c The vent pipe bracing constants are:

A = 2478.7 mm~

E = 2.06 x 10~ kN/mm~

Inserting these factors gives:

F = 0.511 ' kN Rev 9, 07/85 9P-39

PROP RIETAHY To determine the resultant bracing forces, the signals of measurinq points SG 6.1 and SG 6.2 were digitized for identical times. Taki nq in to consi derat ion the direction of action (90 see Figure 9-6), the associated measurement values vere added vectorially and the resulting resultant forces were sorted by magnitude. As in the evaluation of the pressure loads, only the maximum values for each event were deter mined. The event interval for the different break sizes was not changed f rom the values indicated in Subsection 9.4.2.1.1.2. The evaluation was carried out throughout the entire test period in a frequency range from 0.5 to 200 Hz (see Subsection 9.4.3.3) .

The calculated resultant bracing forces were plotted in graphs vs. time, using the magnitude of the maximum resultant in each interrogation interval. Here too, the tests with identical initial conditions (see Table 9-4) were combined into one graph.

As examples, Fiquzes 9-114 to 9-117 show these gzaphs for the Qifferent break sizes. In each instance pairs cf tests that were run with an initial water temperature of 320C were selected:

Break Sipe Test Pair Illustrated in Pull MSL break 5/6 Fi gur e 9-114 1/3 MSL break 11/12 Figure 9-115

.1/6 MSL break 15/16 F iguz e 9-116 HCT. break 1/2 Figure 9-117 In addition, Figures 9-118 to 9-121 classifies the maximum resultant bracinq force for each event intezval into histograms.

The class interval is uniformly 5 kN. For. the reasons mentioned in Subsection 9.4. 3. no normalization was made, i.e., the examples in Fiqures 9-118 to 9-121 are absolute-value histogzams.

Therefore, each of those histoqrams contains only the results for one test (No. 5, 11, 15 and 1) 9.4.3.1.2 Distribution of Direction of the Resultant Bracing ZQZcQQ A point correlation method 'was used'o determine the direction distribution of the resultant bracing forces. At identical instants of time, instantaneous values were taken from the load-time functions and recorded. in a manner correlated with one another. The instants of time Vere determined by the reversal points of the guide trace. In two computation runs, the guide trace and correlation trace were interchanged. The results of the two computation runs are combined in one graph for each test.

The resulting maximum values can be distinguished from the maximum values by the method described in Subsection 9.4.3.1.1 within the evaluation accuracy of +3 kN.

Rev. 9, 07/85 9P-40

PROPRIETARY Examples of these correlations for Tests No. 5, ll, 15 and 1 are shown in Figures 9-122 to 9-125. These depictions provide no in forma tion about the fzequency (of occurr ence) distributions of the resultant fozces A point can also include frequencies of occurrence s > 1.

When analyzinq Figure 9-122 to 9-125, no significant preferred directions are discernible, but rather the "point accumulations>>

hint at an isotropic distribution. In addition, the summary of all MSL breaks (Tests No 3 to 20) in Figure 9-126, show no apparent preferred, direction for the higher values of the resultant bracinq forces. The positions of the braces are at 00 (bracing 1) and '900 (bracing 2) . In this illustration values 25 kN 'were omitted..

9;4.3.1.3 Frequency Distributions of the Resultant Bending

-. Moments at the Dummy~uencher In the GKM II-M condensation tests, the bending moments at the dummy quencher were measured in the horizcntal plane (parallel to the bottom of the te st tank) an d also in the vertical plane. For that purpose, two strain qauges were connected such that they only recorded the nonsymmetrical component of the normal stresses. The followinq strain gauges were mounted for that purpose (see Figure 9-8):

SG 6. 7: Moment in the vertical direction S G 6. 8: Mom en t in t he hor iz onta 1 -di zec tion.

Usinq the equation indicated in Subsection 8.5.2.3 3.1 M = 0.38 c kNm B

with c,in p m/m, the measured bending strains were converted into bendinq moments.

The evaluation of the bending moments relates to the resultant bendinq moment, i. e., the bending moment which actually loads the quencher azm. The maximum resultant bending moment .for each event was determined hy the same method as was used for the b racing f o rces (see Subsect ion 9.4. 3. l. 1) . Hez e too, t he evalaution was performed throughout the entire test period in the frequency range from 0.5 to 200 Hz (see Subsection 9.4.3.3)

Fiquzes 9-127 to 9-130 present the results as plots vs. time for the different break sizes. Once again, the tests with identical initial conditions are combined into one graph.

Rev. 9, 07/85 9P-41

PROPRIETAR Y The elected examples aze as follows:

Break S ice ~ Test Pair-Illustrated in Full HSL break 5/6 Figure 9-127 1/3 HS L break 11/12 Figure 9-128 1/6 HSL break 15/1 6 Figure 9-129 RCI, break 1/2 P igur e 9-13 0 Piqures 9-131 to 9-134 show the histograms of the resultant bendinq moments with the uniform class interval of 5 kNm. These are absolute-value historgrams for Tests No. 5, ll, 15 and l.

9.4.3.1.4 Direction Distribution of the Resulting Bending moments at the Duu~euencher Mhe n a pplied to the measured bending strains, the po int correlation method described in Subsection 9.4.3.1.2 indicates that, the maximum bending strains and bending moments occur prefezentially in the vertical direction. The contribution of the horizontal component SG 6.8 to the resultant can almost be neqlected, so that the bendinq moments measured at the measuring point SG 6.7 can practically be equated to the resultant moment.

As examples, Figures 9-135 to 9-138 show the point corzelations of the resultant bendinq moments for Tests No. 5, respectively. Here too, the maximum values can deviate from the ll, 15 and 1, results described in Subsection 9 4.3.1.3 within the evaluation accuracy of +6 kNm 9 Q g 1 Q Preguency Distributions of the Forces on the I-Beam Por symmetry reasons and in order to eliminate bending moments at

.the measurement bolt, two strain gauges displaced hy 180~ were mounted longitudinally in the middle of the bolt and connected to f f orm a complete .bridge or the measurement of normal forces (measuring points I.C 6.1 in Figure 9-9).

The statistical evaluation of the measurement signals was again accomplished using the method described in Subsection 9.4.3.1.1, but was performed separately for the positive vertical forces (directed upward) and the negative vertical fzoces (directed downwa rd) .

The positive and negative maximum values vs. time for each event no. 5/6, ll/12 and 15/16 (MSL breaks) .'n interval are plotted in Figure 9-139 to 9-141 for the test pairs addition, Figure 9-142 shows the corresponding .results for the Test No. 1 (RCL break) .

Rev. 9, 07/85 9P-42

PR OPRIZT ARY Figures 9-143 to 9-146 illustrate the absolute-value historgrams the positive and negative vertical forces on the I-beam for 'f the freguency range f rom 0. 5 to 200 Hz. The class interval is uniformily 5 kN. As for the bracing .forces and the bending moments at the dummy quencher, the histograms for Tests No. 5, ll, 15 and 1 were selected as examples. The results of this evaluation are static equivalent loads and again are not directly tran posable to plant conditions.

9. 4. Q. 1. 6- Sta tistica1 Ch ar acterist ics Table 9-11 provides an idea of the most important statistical characteristics of the loads on the internals in the water region of the test tank f or all 22 tests. They include: r maximum and mean values for the resultant bracing forces, maximum and me an values o f the resultant bending moments at the dummy quencher, maximum and mean .values of the vertical forces on the I-beam, for the fzeguency range from 0.5 to 200 Hz.

The mean values were determined by the single data proceduze, using the equation indicated below:

, le ZP 3.=1 1, P

n where: P = Mean value i

P. = Sinqle value n = Total number of values The absolute maximum measurement values of the zesultant bracing forces, resultant bending moments and vertical forces at the I-.

beam and the maximum mean values for these loads aze shown below:

gbsogute N aximum Vague of the Load

~

Heasured in Test Resu'ltant bracinq force 88 7 kN No. 17 Resultant bendi.ng moment at the dummy- quencher 79. 6 kNm No 11 Vertical force at the I-beam 32 0 kN No 4 R ev. 9, 0 7/85 9P-43

PROPRIETARY ga x.imu m Qe a n Va.lue of the Load

• Measured in Test Resultant bracinq force 22 5 kN No. 18 Resultant bendinq moment at the dummy quencher 16. 1 kNm No. 12 Vertical force at the I-beam 4 4 kN No. 12 9.4,3~2 Parameter Influences The influence of the .test parameters "Break size and water temperature at test start>> is illustrated in graphs in which the mean values of the load on the internals in the pool are plotted versus the above-cited parameters. A sepazate illustration is not necessary for the parameter>>Aiz content in the drywell at test start>>, since, as expected, this parameter has no signif ica nt inf1 uence.

9. 4. 3. 2,1 ~nQuence of the Break Cross-Sectional Area In the top half of each of Figures 9-147 to 9-149, the mean values of the resultant bracing forces and bend.ing moments at the dummy quencher and of the vertical forces on the I-beam are plo'tted vs. break size. Por the vertical forces, the mean value which was 1azqer in absolute magnitude was used for each test.

Figure 9-147 indicates a decreasinq trend in the magnitude of the bracinq forces with incresinq brea'ize. In contrast, Figures 9-148 and 9-149 indicate the loads on the dummy quencher and I-beam exhibit a maximum for the medium break size (1/3 HSL break) .

9,4,g,2 g. Influence of. the Initial Pool Te~merature The bottom half's of Figures 9-147 to 9-149 illustrate the mean values of the resultant bracinq forces, bending moments at the dummy quencher and o f the vertical forces on the I-beam as a function of the initial pool temperature. There is uniformly indicated here a more oz less significant decreasing trend foz the loads with increasing initial pool temperature.

~4 g g,). Influence of the Initial Air Content. in the T)r~well This parameter has no significant'nfluence on the magnitude of the loads on the internals in the pool. This can clearly be seen from the bottom half of each of the Figures 9-147 to 9-149. The mean values of Tests No. 7 and 8 (Pull MSL breaks) and Test No.

17 and 18 (1/6 MSL breaks) with approximately 85%%u initial aiz content in the drywell vary (as for the dynamic pressure loads; see Subsection 9.4.2.1.3.3) within the same scatter range as the comparable Tests No. 5, 6 and No. 15, 16, respectively, with 100$

air content at test start.

Rev. 9, 07/85 9P-44

PROPRIETARY

9. 4. 3 3 F reaue ncy g of Oscillation) Analysis Typical examples for the oscillation frequencies recorded in the bracinq of the vent pipe and at the dummy quencher during the GKM II-M tests are shown in Figures 9-150 and 9-151 for Test No. 13 (1/6 MSL break) .

In addition to the power spectral densities (PSD's) of the strains recorded in the individual braces (top graph in Figure 9-150), the cross power spectral densities (CPSD's) for measuring points SG 6.1/SG 6.2 were determined in magnitude and phase (middle qzaphs in Fiqure 9-150) . For two measurement signals to be correla ted, the CPSD is calculated according to the equation indicated in Fiqure 9-55 (see Subsection 9.4.2.1.1) . Besides the phase position, the magnitudes obtained are each a measure for the degree of correlation in comparison with the two individual power densities. That comparison is represented in the form of the coherence function (bottom qraph in Figure 9-150). Here, coherences > 0.5 are to be designated as significant.

Both the PSD's of the strains at the individual bzaces and also the CPSD exhibit power-density maxima at the same frequencies:

ll, 17, 31, 36, 56 and 63 Hz with clear emphasis of the frequency components at 17, 31 and (36) Hz.

The dominance of the loading on the dummy quencher in the vertical direction, as described in Subsection 9.4.3.1.4, becomes especially cleaz in the power spectra. As an example, Figure 9-

.151 shows the PSD's and the CPSD of the measuring points SG 6.7-(bendinq strain in the vertical direction) and SG 6.8 (bending .

strain in the horizontal direction) . For the stzain in the vertical direction there is one significant power-density maximum at 105 Hz. The frequency components of the horizontal bending strain are,neqliqible.

9. 4. 4 Summary ~

The following conclusions can be drawn from the experimental investigations of the pressure suppression system in a conservatively simulated sinqle cell of the Susquehanna SES, the observed phenomena and the quantitative determination of the structural loads produced by them. The first four conclusions follow from the mean value for the entire transient calculated for each of the tests (see Subsection 9.4.2.1.2.3) and plotted in Figures 9-100 to 9-107.

r The pressuze amplitudes during chugging aze slightly dependent on the brea'ize which determines the steam mass flow transient.

The initial pool temperature has no influence or only a slight influence on the magnitude of the pressuze am pli tudes.

Re v 9, 07/85 9P-45

PROPRIETAH Y In the r anqe o f suppr ess ion cha mbe r back- pressures o f relevance to the plant, there is a3so no significant in fluence on the pressure amplitudes.

The maximum mean values of the dynamic pressures in the pool, measured in the dominant frequency range from 0.5 to 13 Hz, are +0.62 bar and -0.41 bar (wall pressures) .

These maximum values are compared with absolute maximum values of +1.91 bar and -1.24 bar.

The measured loads on the pool boundaries and pool in te rn als (vent-p ip e bracing, du mmy q uencher an d I-bea m) are no t di rectl y appl ica hie to the plan t and still require further evaluation.

In conclusion, the measurement results provide an adequate data base in which to e valuate the conservatism of the DFFR C.O. and chuqqinq load specification (see S ubsection 4..2) .

Rev. 9, 07/85 9 P-46

PROPRIETARY 9 5 ~ DATA ANALYS IS- AND LOAD

~ SPECIFICATION

9. 5. 1- In trod uction ~

As described in Subsection 9 1, the GKH II-N test program was initiated to resolve the HRC's concerns with the DFPR LOCA load specification developed from the original 4T tests. The GKM II-N.

test facility, instrumentation, test parametezs and matrix aze described in Subsection 9.1 9. 2 and 9.3. The test results presented in Subsection 9.4 indicate that the tests covered the range of expected plant parameters and provides an adequate data base for verifying the conservatism of the DFFR specification.

This section provides an analysis of the GKN-IIN data and the specification of a new LOCA steam condensation load for comparison with our existing DFFR condensation oscillation and chugqinq load specification. Subsection 9.5. 2 gives a detailed description of the various mechanisms involved that determine the character of the pressure time histories at the pocl boundary, as well as a frequency evaluation of the pressure loads measured in the pool.

Subsection 9.5.3 qives an explanation of the SSES LOCA load

'pecification that, will he used to verify the conservatism of the oriqinal DFFB load. The load specification consists of the followinq key elements:

Similar to the generic Nark II'.program, the load is specified by the methodology =contained in- Reference 65:-

This consists of generation of volume sources and application of these sources to an acoustic model representing the SSES suppression pool The sources are developed fzom selected GKM II-M data.

For the impulsive type chugging, the sources represent mean value chuqs. Four (4) time seqmen ts -are selected as repzesen'tative of the mean value chugs. (see Subsection 9. 5. 3. 1)

For the condensation oscillation (CO) like pressure oscillations, one (1) time segment is selected as representative of the bound of the GKN II-N CO data base. {see Subsection 9.5.3.1).

The above time segments are sourced using the IMEGS model of the GKN II-N test tank (see Subsection 9.5 3.3) .

As the selected impulsive type chugging represent mean value events, they cannot be transposed unaltered to the SSES acoustic model. Such a direct transposition would lead to a high probability of exceedance. To account foz this, each chug source is increased by an.

Re v. 9. 07/85 9 P-47

'PROP BI ET ARY amplitude factor whose magnitude is based on a statistical analysis and a required exceedance criteria (see Subsection 9. 5. 3. 2)

Since the selected CO time segment bounds the CO observed in the plateau region of the RCL break tests, and beqinninq of the NSL tests, no amplitude factor is re aui red.

Application of the SSES LOCA load specification and the sources to the SSES multivent geometry to calculate pressure time histories for both the symmetric and asymmetric load case. In addition, the time scales foz the pzessuze time histories are contracted and expanded to compensate for the variation in the pressure oscillation frequencies observed during the GKN II-M Tests. (see Subsection 9.5.3.4)

Verification of the SSES LOCA load specification thzouqh comparison with the available multivent test data from JAERI. This involves using an acoustic model of the JAERI facility and the SSES LOCA load specification to calculate JAERI wall loads for comparison with the available JAERI data. (see Subsection 9.5. 3. 5)

9. 5. -2 Data Analysis.

9 5 2 1 Dynamic Pressures in the Pool and Their Physical interpretation In this subsection, a more detailed explanation of the physical processes that occur when steam is introduced into the water is given. The origin of the pressure events oz pulsations is the rather violent condensation of the steam whenever contact with subcooled watez it comes into As a result of the underpressure that is ozoduced, a number of mechanical degrees of freedom of the system are excited. They are classified as follows:

Notion of the water into and back out of the vent pipe.

Acoustic oscillation of the steam column in the vent pipe (vent acoustics) .

Acoustic oscillation of the water in the pool (pool acoustics) .

Oscillation of the tank (structural vibration) .

In addition, there are also other steam condensation phenomena which can only be explained by additional events.

Hev. 9, 07/85 9 P-48

PRO P'R IET AR Y Pinch-off and collapse of 'a bubble underneath the vent pipe, outside against the vent pipe or inside the vent pipe.

Pulsation of a bubble connected to the vent pipe Each of these events contributes to the pressure time histories at the containment boundary in a more'r less pronounced manner, dependinq on the system conditions at the time (i.e., steam flow, pool and steam air content, pool temperature, etc.). Foz the ma jority of the data recorded at GKM II-M, the vent and pool acoustic models are the principle mechanisms contzibutinj to the observed pzessure time histozies. This is especially true for the steam condensation or chugs associated with a low pool and steam air content. These chuqs produce an acoustic wave which travels up the vent pipe and through the pool causing the vent and pool to ring at their natural freguencies. The pressure oscillations measured at the pool boundaries are then a composite of the vent and pool rinqouts and other related oscillations associated with the steam bubbles.

The vent acoustic frequency of approximately 9 Hz observed at GKM II-M can be calculated by using an effective vent pipe length of

13. 5 m and a speed of sound of 483 m/sec in the steam. Figures 9-152 to 9-154 are typical examples of pressure time histories from Test Xl (shakedown tests for the 1/6 MSL break), which display this frequency in the vent whenever water is situated inside of it. The first harmonic of this vent acoustic frequency can be see n in Fig ure- 9-154.

Fiqures 9-152 to 9-154 also exhibit the pool acoustic frequencie . These frequencies can be derived from the traces for time intervals in which the water is situated in the vent The oscillation frequencies increase with time (16.4 and 32.2 Hz) due to the increasing sonic velocity associated with the decreasing pool air content.

Fiqure 9-155 gives an overview of a typical blowdown. This figure plots all the frequencies for Test Xl vs. time. The frequencies were determined by reading off the oscillation periods from pressure gaqe P6.4. The good agreement observed between the evaluation results and the calculated frequencies in the model (>>bubble oscillations>>, >> pool acoustics>>, "vent acoustics>>) supports the physical description of the events presented above. The frequencies of the pool acoustics in -the model (solid line in Figure 9-155) is calculated according to Fiqure 9-156, assuming the air content illustrated in Figure 9-157 For condensation oscillation, the pressure traces exhibit regular pressure oscillations without a clear association to the vent and pool acoustics. This phenomena is observed primarily at high mass flow densities and. for pool air contents of more than 1%.

Rev. 9, 07/85 9P-49

PROPR ZET ARY Fiqure 9-158 shows typical condensation oscillations during the plateau region of Test. No 33.

9 $ g.2 evaluation of the Chugging Freceu~enc at GKM II-M Fiqures 9-159, 9-160 and 9-161 summarize the typical chugging frequencies (i e., the frequency which corresponds to the time interval between two chugging events) for the full, 1/3 and 1/6 MSL breaks, respectively.

For the, full MSL breaks (see Figure 9-159) the chug frequency is approximately 1.7 + 0.2 Hz (Test No 4) or 1.4 + 0.2 Hz (Test No.

10) at test start, drops monotonically to about 0.6 Hz at 50 sec and then remains constant until test end. The lower chugging

.frequency in Test No. 10 (warm pool) compared to Test No 4 (cold pool) is explained by the poorer heat transfer at the end of the vent pipe in Test No. 10 than in Test No. 4, due to the lower pool subcooling. The bubble at the end of the vent pipe must then be larger and therefore has a longer period due to the larqer hydrodynamic coupled mass of water.

For the 1/3 MSL breaks (see Figure 9-160), the chugging frequency follows a similar pattern as the full MSL breaks, except that the monotonically decreasing chuqqing frequency begins at 50 sec with

0. 8 Hz, reaches 0.6 Hz at 80 sec and from thereafter remains constant for 100 sec until the end of the test.

The 1/6 MSL breaks exhibit the same chuggin g frequency of 0. 6 Hz

.for almost the entire duration of the test, except for a slight rise at test end (see Fiqure 9-161) .

These three fiqures indicate that a chugging frequency of 0.6 Hz prevails for mass flow densities below approximately 40 kg/m~ sec (see Fiqure 9-15). Conversely, above 40 Kg/m~ sec, the chugging frequency is approxima tely proportional to the mass flow density.

9.5.2.3 Evaluation of the Pressure Oscillation Frequencies

-.j n fQe. Pool-Fiqures 9-162, 9-163, 9-164 and 9-165 contain the dominant pressure oscillation frequencies for the RCL, f ull, 1/3 'and 1/6 MSL break tests, respectively. Fiqure 9-166 shows how these frequencies were determined from the period between oscillations for the hiqh.and low bandpass filtered traces (see Subsection 9.4.2.1 2) . Frequencies were read-off for both large amplitude oscillations as well as small amplitude oscillations extending over several oscillations.

An evaluation of these figures indicates that the trends found previously in Test Xl (see Figure 9-155) are quantitatively reproduced, in principle, for all the tests. The vent acoustic frequency of about 9 Hz is readily seen in all the MSL tests, Rev. 9, 07/85 9P-50

PROPRIETARY-startinq at 60 sec for the fu11 and 1/3 MSL breaks azd 100 sec for the 1/6 MSL breaks.

~9. 3. SSES LOCA Load Specification The previous subsections qive a thorough review of the GKM II-M test results and e valuation o f t hese results. This data oriqinates from the GKH II-M single vent facility, but can be applied to the SSES multi-vent suppression pool, if the f allowing conditions are met:

The sinqle vent system represents the so-called single cell of the SSHS plant, and the same initial conditions exist at the single vent system as at each vent of the multi-vent facility.

As described in Subsections 9.1, 9. 2 and 9.3, the test facility, parameters and matrix were designed to ensure that the above conditions were met. In addition, Subsection 9.4 showed that the measured GKM II-M steam fluxes for the RCL and full MSL transients very closely simulated the CESAR calculated blowdowns, thus quaranteeinq prototypical test conditicns and results throuqhout the transients. The 1/3 and 1/6 MSL breaks also provided transients of sufficient duration to quantify the high amplitude chugs associated with hiqh mass fluxes and low steam and. pool air contents. These tests resulted in' total data base of approximately 30'00 chugs of which over 800 chugs exceeded 0.5 bars. Thus, the GKM Il-M data "provides, a very prototypical -and conservative basis for specifying a SSZS load definition.

With this in mind, the problem mow is how to develop and specify a sufficiently conservative yet realistic LOCA load definition from the GKM II-M data base. To do this we make use of the fact that multi-vent pressures vill be distin'ctly smaller than the pressures measured at GKM II-M. This conclusion follows from the highly random nat'ure of the condensation events. This effect has been observed and verified numerous tines in a number of multi-vent test proqrams (see Reference 66) . This randomness of events was also observed at GKM II-M, where the measured pressure amplitudes of events occurrinq at close time intervals can be very different, even though the overall test conditions at those times, such as drywell pressure, steam flux, air content in the pool'and steam, and mean pool temperature are practically identical (see Figure 9-81 to 9-84) .

I The stochastic nature of chugging stems from the fact that the actual initiatinq event is the spontaneous condensation of steam at the steam/water interface. The violence of the condensation and, related to it, the thermodynamic and mechanical occurrences in the time sequence are sensitively influenced by the instantaneous size and shape of the bubble surface, the temperature distribution and turbulence in the pool. All Rev. 9, 07/85 9P-51

PROPRIETARY experience indicates that such conditions cannot be the same at every vent pipe of the SSES suppression pool during a postulated LOCA The second assumption is that the loads are further reduced by the desynchronized nature of chugging. Again, numerous multi-vent tests indicate that chugs do not occur simultaneously or in-phase at each vent pipe, but rather the chugs at different vents occur within a prescribed time window.

In view of these factors, the load specification assumes that the chugs occuring simultaneously at different vent pipes of SSES have different'ntensities and follow the same distribution of chug amplitudes in time as in the GKM II-M single-vent facility.

This means that the total pool bottom load at SSES would be an averaqe of the different chugs at each vent and the deviation from this mean value follows a probahilistic distribution. Thus, the random amplitude chuqs at different vents can he replaced with the same mean value chug at each vent.

Therefore, as briefly described in Subsection 9.5 1, the SSES LOCA load specification for chugging is oriented toward mean value chugs. The random chuqs at different vents aze replaced with the same chug at each vent whose maximum pressures correspond to the mean values at GKM II-M. Pour (4) GKM II-M time segments have been selected as representative mean value chugs in both amplitude and frequency. These time segments are sourced and applied with a 50 msec time window to the IMEGS/MARS acoustic model (see Subsection 9.5.3.4) to calculate pressure time histories at the containment boundary. In addition, since the deviation f rom the mean value at SSES follows a prohabilistic distribution, the mean value sources are increased by an amplitude factor to obtain the desired exceedance probability prior to transposition to SSES. The time scales for each set of pressure time histories are also contracted and expanded in time to cover the range of frequencies observed at GKM II-M. Both a symmetric and asymmetric load case are considered.

As explained in Subsection 9.4.2.1, the SSES LOCA load specification does not distinguish between chugqing and CO as have the Mark II Owners. However, CO-like pressure time histories (low freguency and constant amplitude) were observed in the plateau region of the RCL break test and in the beginning of the MSL break tests. Thus, in order to cover this type of pressure oscillation, one (1) time segment has been selected foz the SSZS LOCA load definition as representative of the CO-like pressure time histories. However, separate load combination acceleration respose spectrum (ARS) curves for CO and chugging are not qenerated for design assessment Instead, the ARS curves calculated from the four (4) chuqging time segments and one (1)

CO like time seqment are combined to form one (1) envelopeing LOCA ARS curve at each containment node for combining with the remaininq loads for design assessment (see Subsection 9.6) .

Rev. 9, 07/85 9 P-52

PROPRIETARY Thus, the SSES LOCA load specificaton does not treat CO and chugging separately when used for load combination ARS curves, but only selects a separate CO time segment to depict the second category of time histories seen at GKM II-M.

As with the four (4) chuqqinq time segments, the one (1) CO time segment is sourced and .applied to IMEGS/MARS. However, application to the suppression pool acoustic model is made without-dephasinq to calculate containment loads. This source is not multiplied by an amplitude factor, since the magnitude of the CO is relatively constant compared to the highly random chugging amplitudes, and the selected trace bounds the CO at GKM II-M.

Aqain, the CO time seqment will be contracted and expanded in time to compensate for any uncertainty in the frequency content of the selected trace. Only a symmetric load case is considered.

The followinq subsections give a detailed description of the SSZS LOCA load specification. 'pecifically Subsections 9.5.3.1 and 9.5.3.2 explain the methodology for selecting the five {5) time seqments for sourcinq and the procedure for determining the amplitude factors, respectively. Subsections 9. 5.3.3 and 9.5.3.4 delineate the procedure for sourcing the selected time segments and the applicaiton procedure for calculating the SSES containment loads, respectively. F inally, Subsection 9.5. 3. 5, provides justification for the SSES LOCA load definition and the 50 msec time window.

9 5.3.1. Selection of the GKM II-M Time Segments to be Sourced As previously described, the SSES LOCA load specification selects five (5) pressure time histories for sourcing in IHEGS/MARS to calculate the SSES wall loads. For chugging, four (4) time histories have been chosen as representative of the maximum mean value chuqs at GKM II-M. For CO, one (1) time segment has been.

selected as boundinq of the CO at GKM II-M. This subsection describes the selection procedure, presents the finally selected time segments and provides verification of the specified traces by comparinq the PSD's of the selected, traces with the PSD's of the GKM II-M data.

9. >.$ .1.1-- Description of the Evaluation Procedure The evaluation of the GKM II-M data for selecting the time seqments to be'used in the SSES LOCA load definition consists of the followinq steps:

Selection of a representative pressure transducer.

Preparation of two bandpass filtered pressure traces: one from 0.5 to 13 Hz (low-frequency band) and one from 10 to 100 Hz fhiqh-frequency band).

Rev. 9, 07/85 9P-53

PROPRIETARY Determination o f events, maximum ampli tudes and oscillation frequencies of the events.

For the chuqqinq, time segments, form mean values and select chuq time seqments near the maximum mean value curves.

Foz the CO time segment, select a CO tzace which bounds the CO data at GKil II-.'l.

Verification of the selected traces.

Pressure transducer P6 7 (see Figure 9-5) was chosen as the representative tzansducer because it generally recorded pressures which bounded the zemaininq transducers (see Figures 9-36a to 9-54)

The bandpass filtering is intended to faciltate the reading of the pressure amplitudes and. to sharpen the criteria for the selection of events. Subsection 9.4.2.1.2 gives the criteria for stipulatinq the above high and low frequency .bands.

Subsections 9.4.2.1.2 and 9.3.2.3 provide explanation for reading the maximum amplitude pez mean event interval and the pressure oscillation frequencies from the bandpassed traces, respecti vel y.

Fiquzes 9-81 to 9-84 are examples of the maximum amplitude pez mean event interval determined from typical full, 1/3, 1/6 HSL and RCL break tests. In addition, typical oscillatory frequencies of pressure events for the RCL, Full, 1/3 and 1/6 HSL break tests are shown in Figures 9-162 to 9-165.

The mean values were determined from the arithmetic averaqes of the maximum amplitudes within the time intervals defined in Subsection 9.4.2.1.2.1. The averaging is always based on those tests that were run with identical initial conditions (i.e.,

Tests 5/6, 11/12, etc.). Figures 9-85 to 9-87 pzesent the results for the full, 1/3, 1/6 HSL and RCL break tests. Section 9.4 qive detailed results of the above statistical analysis.

9.5 3 1.2. Selection of. the Ch~u in Pressure Time Histories Based on the results of the statistical analysis cf the pressure time histories (see Fiqures 9-85 and 9-86), representative chugs can be evaluated which are oriented toward the mean value curves of the pressure amplitudes. 'For determining the chug time seqments for souzcing, all NSL break tests (Tests 3-20) were eval'uated. When selectinq events, an attempt was made 'to select chuqs lying very close to the mean value curves in both the low-frequency and high-f zequency ranges.

certain that most of the test duration In addition, was covered it was made

.by the chugs.

Under this assumption a pre-selection of 46 time segments was made. These chuqs adeguately represent the results from HSL Test Nos. 3-20. However, for practical application to the SSES Rev. 9, 07/85 9P-54

PROPRIETARY load specification, the number of pze-selected pressure time histories was too extensive and would require too much computation time. Therefore, the pre-selected chugs required further reduction.

To make this final selection, the pre-selected chugs from the same break size were grouped together (i.e., full, 1/3 and 1/6 MSL) and ordered accordinq to increasing initial start times.

Then one chug was picked out arbitrarily from every 10 sec interval and compared consecutively with its neighbors for shape of the siqnal, amplitude, dampinq and frequency. Foz each pair subject to compazison,. the one that appeared to be covered by the other was eliminated. In doing this, it was taken into consideration that the SSFS LOCA load specification also, assigns an amplitude factor and a time factor to the source determined

.fzom the f inally selected pressure time histories.

The result is a selection of four (4) time segments which represent the mean value chuqs at GKM II-M. Figures 9-167 to 9-170 present the four (4) selected time histories and the time multipliers for each trace. As shown in these figures, two (2) traces oriqinate from the full MSL tests, one (1) trace is, from the 1/3 MSL tests and the remaininq trace is from the 1/6 MSL tests. Figure 9-170a presents the PSD's of the specified chug time seqments. Figures 9-171, 9-172 and 9-173 compare the pressure amplitudes of the specified time segments with the mean values from the various MSL tests. These figures show that the selected traces compare very favorably with the maximum mean value curves from. the MSL tests. A time segment f rom the -maximum- -"

pressure amplitude region of the 1/6 MSL tests (approximately 100 sec after test start, see Figure 9-173) was not chosen, since the selected time segments from the 1/3 and Fall MSL tests (see Figures 9-171 and 9-172) are bounding. Thus, the finally selected chug time histories form a conservative and representative data base for sourcing in the SSES LOCA load specification.

Additional insight into the adeguacy of the selected time segments is given by Figures 9-174 to 9-176. These figures show the frequency range of the contracted and expanded specified time histories in relation to the pressure oscillation frequencies detezmined from the GKM II-M data (see Subsection 9.5.2.3) . They indicate that the selected traces cover the dominant freguencies at GKM II-M. However, since the oscillation frequencies were not determined'rom a PSD analysis, these figures do not provide informaton about the power at these dominant frequencies. Thus, to further verify the conservatism: of the selected chug traces, a comparison of the PSD's of the selected traces with the PSD's of the GKM II-M data has been performed. This comparison is presented in Subsecton 9.5.3.1.3.

Rev. 9, 07/8 5 9'P-55

PROPRIET ARY 9.Q. 3.1. 2.1 Selection of the CO Pressure Time History The CO observed during the plateau reqion of the RCL tests and the heqinninq of the NSL tests exhibit clear differences from the event like, high amplitude, 'damped chuqs in the previous subsection. The CO time histories are chazacterized by constant amplitude, low fzeque ncy-. (6 to 8 Hz) sin usoidal-like pressure oscillations. Since CO does not display random amplitude, event like pressure oscillations, the concept of a mean value event multiplied. by an amplitude factor to achieve the desired exceedance criteria cannot be utilized Thus, the evaluation procedure used in the previous subsection is not applicable for specifying the CO time seqment. Instead, the CO time segment was chosen to bound all the CO data measured at GKN II-M.

Figures 9-177 a 6 b present the finally selected CO trace from Test No. 2 and the time expansion and con traction factors. The time factors were selected to cover the range of frequencies at GKN II-M. In addition, Fiqure 9-177c shows the PSD for the specified CO time segment. Figure 9-178 shows the oscillation frequencies of the contracted and expanded specified CO time seqment compared with the oscillation frequencies determined from the RCL Test Nos. 1, 2 and 33. Again, this figure is not based on a PSD analysis and provides no information on the power at these dominant frequencies. f Thus, to urther veri f y the conservatism of the specified CO trace, a comparison of the PSD from the selected CO trace with the envelopeing PSD from the GKM II-N CO data has been performed. This comparison is presented in Subsection 9 5.3.1.3.

9. 5 3. l. 3 Verification of the S elect d T i me Se ments 9.5.3 1.R. 1- Chua in Time Seaments 4

Subsection 9.5.3.1.2 showed that the selected chug time segments are representative of the maximum mean valve curves at GKM II-N based on amplitude only (see Figures 9-171 to 9-173) . However, in ozder to further verify the conservatism of the selected traces, it must be shown that the selected traces represent the mean value chuqs at GKM II-N in both frequency and amplitude. To do this, a PSD analysis has been performed, which compares the envelopee of the PSD's of the selected traces (including the time multipliers) with the -envelopee of the PSD 's representing the chugs at GKN II-N. This comparison is shown in Figure 9-178a and indicates that the selected chugqing traces bound the frequency content of the chuqqing events observed .in the tests.

The envelopeing PSD of the selected chugging traces was obtained by first generating three PSD's for each selected time segment correspondinq to the time multipliers a = a a = 1 andaman.

are the limits of the The numbers inserted for cmin . and a max Rev. 9, 07/85 9P-56

PROPRIETOR Y intervals given in Figur es 9-167 through 9-170. Figures 9-178b to 9-178e illustra tes these PSD ~s. The resulting twelve (12)

PSD's were then combined to form the. envelopeing curve shown in Fiqure 9-178f.

The PSD representing the GKH II-N tests was generated from the PSD's of selected time intervals (10-20 sec) from Test/Repeat Test Nos. 3/4, 9/10 and ll/12. The selected time intervals were chosen to bound the maximum values of the mean pressure amplitude curves. Figures 9-178 g and h show the time intervals from which PSD's were produced. The PSD's for each pair of Test/Repeat Test for each time interval were then averaged to obtain a PSD which represents the average spectial density of the events occurring in the time interval under consideration. Fiqures 9-178i to 9-178q pre ent these averaqe PSD's for the seven (7) time intervals. These PSD's were'hen envelopeed to form the representative PSD for the tests as, shown in Figure 9-178r.

The above PSC's were normalized such that a chug is represented as having. a unique time duration of tref = 1 sec. This means that the original PSD's for the time segments were multiplied by a factor: A k = n - t where: tA = analyzed time seqment n -- -= -number of- ch This adjustment compensates for ugs within the- time-the>>dead time>>

i nte r val.

between chugs and th us en sures a valid PSD co m parison.

Re v. 9, 07/85 9 P-57

PROP RI ET ARY 9.5.3 1 3. 2 CO Time Segment For sourcinq and application to IQEGS/MARS, Subsection 9.5.3. 1.2. 1 selected PTH No.14 from Test No. 2 (see Figure 9-177aSb) as representative of the CO data at GKM II-M. The criteria for the selection of PTH No. 14 was that it boundinq in both amplitude and frequency of the CO data observed must be in the beqinning of the MSL tests and the plateau region of the RCL tests. However, Subsection 9.5.3.1.2.1 documented the selection of PTH No. 14 based only on a comparison of the CO from the RCL tests and failed to compare the CO from the MSL tests.

As a result, the followinq subsections present additional verification of the conservatism of PTH No. 14. This is based on a comparison of the CO amplitudes from all tests, as well as a comparison of the envelopeinq PSD of PTH No. 14 with the envelopeing PSD of the CO at GKM IZ-M.

Zn addition, the CO data from Test Nos. 33 and 34 were not considered in the selection and verif ication of PTH No. 14 since the hiqh initial pool temperature (130~F) for these tests is outside the bound of a realistic or credible LOCA initial suppression pool temperature at SSES. Thus, the data represents highly conservative CO and has not been included.

9 5.$ .1.3. 2.1 CO Evaluation 9.5.3;1.3.2.1.1 Comparison of the CO Amplitudes at GKM-ZI-M As previously stated, Subsection 9.5.3.1.2.1 only considered the.

RCL Test Nos. 162 in the selection of PTH No. 14, since the CO P6.7 amplitudes from the MSL tests are small compared to the CO amplitudes from the RCL tests. This is clearly shown in Figures 9-262a 8 b These two figures compare typical CO traces from pressure transducer P6.7 for the Full MSL, 1/3 MSL and RCL tests.

No CO were observed for the 1/6 MSL tests (mass flow density less than 30 kg/s) . These figures indicate that the CO pressure amplitudes from the MSL tests are small in comparison to the CO amplitudes from the RCL tests.

Table 9-12 and Fiqure 9-263 provide additional proof that the CO amplitudes, from the RCL tests are bounding. Figure 9-263 plots the mean pressure amplitudes for each break size vs average steam mass flux from Table 9-12. Figure 9-263 implies a monotonic increase in the CO pressure amplitudes with steam mass flux and that the RCL break tests bound the MSL tests.

Thus, further verification of PTH No. 14 can be restricted to the RCL tests.

Rev. 9, 07/85 9 P-58

PROP BEET ARY Table 9-12 shows the time span in which CO was observed for each test, the average steam mass flux during CO and the mean CO pressure amplitude for each test break size.

The time spans classified as CO (see Table 9-12) were selected from the traces based on the criteria:

o Sinusoidal pressure amplitudes, which do not vary much on the averaqe, with amplitudes higher than 0.05 bar (peak-to-peak reading divided by two (2); pressure transducer P6.7) o D ura tion longer tha n two (2) sec.

o Cleared vent pipe (level probes LP1 to I.P5 and temperature transducer T5m6 not wetted).

The averaae steam mass flux for each break size was obtained as follows: 1.) For each test in Table 9-12, the steam mass flux at the beqinning of the CO time span and the steam flux at the end of the CO time space were averaqed. ,2.) The average steam mass flux from step 1 for the same break size were then averaged.

The mean CO pressure amplitude for each break size was obtained as follows: 1.) For each test in Table 9-12, the sliding mean value was obtained by averaging the amplitudes of seventeen (17) successive pressure oscillations (peak-to-peak amplitudes divided by 2). This gives a mean value curve as a function time into the blovdown analoqou to the mean value curves for chugging (see Figures 9-81 to 9-83) . 2.) The maximum sliding- mean -value from step 1 for the same break size were then averaged. The mean CO amplitude for the BCL tests includes Tests 33 and 34, but the amplitude would not be much lower considered.

if only Tests 1 6 2 were Figure 9-16a provides further verification of the conservatism of PTH No. 14. This figure shows that the test transients at the end of the plateau region bound the theoretical RCL transient, thus ca using the test steam mass flow density to be higher than in the plant .for the same steam air content. This results in larger pressure amplitudes at GKM,XI-M relative to the plant since t he amplitudes increase with decreasing steam air contents.

9-5.2.1.5.2.1.2 999 Comparison To f provide uzther verification of PTH No. 14 a PSD analysis has been performed compazinq the envelopeinq PSD of PTH No. 14 with the envelopeinq PSD of the CO from RCL Test Nos. 1 6 2. HCL tests 33 6 34 vere not analyzed for the reason stated in S ubs ec t ion 9.5. 3.1. 3. 2.

Rev. 9, 07/85 9P-59

PROPRIETARY Figure 9-264 shows this comparison and indicates that PTH Ho. 14 bounds the frequency content of the CO .from the RCL'ests 1 6 2 except for a small portion in the low frequency region. However, the envelopeinq PSD of the four (4) selected chug time segments envelopes the CO data at this low frequency (see Figure 9-178a) .

The enveloping PSD of PTH No. 14 was obtained by first generating three (3) PSD's corresponding to the time multipliers a = a

= 1 and a = e (see Figure 9-265). The numbers inserted%ox'o in and e,a are%Ye limits of the intervals given in Figures 9-177a 6 b. Tfie resultinq three (3) PSD's were then envelopeed to form the curve shown in Figure 9-264.

The envelopeinq PSD of the CO from RCL Tests 1 6 2 (see Figure 9-264) were obtained as follows:

o RCL Tests 1 6 2 were broken into two (2) sec CO time intervals usinq the criteria in Subsection 9 5 3 1.3.2.1.1.

These time intervals are shown in Table 9-13.

PSD's were then generated for each of these two (2) sec CO time intervals from pressure transducer P6.7.

The PSD's of the first four (4) time intervals of Test 1 (8-16 sec) were then eliminated because a visual inspection indicated that they were covered by the PSD's of the last two (2) time intervals (16-20 sec) . Similarly, the PSD's from the first five (5) time intervals of Test 2 (6-16 sec) were eliminated leaving only PSD's from the last three (3) time intervals (16-22 sec).

The remaininq PSD's from each test were then envelopeed to form a PSD representing the time interval 16-20 sec for Test 1 and a PSD representing the time interval 16-22 sec for Test 2.

These two (2) PSD's {one from each test) were then averaged to form the mean of the PSD envelope of Test 1 and PSD f

envelope o Test 2.

Rev. 9, 07/85 9P-60

PROPRIETARY 9,5,3.2 Determiaation of the Amplitude Factors As previously stated, a key assumption foz the SSES LOCA load definition is that the chuqs occurzinq simultaneously at different vents of the SSES plant have stochastically varying amplitudes which follow the same distribution as the chug, amplitudes varying in time at GKM II-M. This means that the total vertica.l force, on the basemat of the SSES containment is an average o f the chuqs occurrinq at each vent, which according to the above assumption is equivalent to the mean pressure of successive GKM II-M chuqs. Thus, the random amplitude chugs at each vent can be replaced with the same mean value chug from GKM II-M at each vent. Zn addition, deviations 'from the mean pressure follow the laws of statistics.

Based on this, Subsection 9.5.3.1. 2 selected four (4) chug time segments for sourcinq ia IWEGS/MARS, that aze repzesentative of the ttaximum mean value curves at GKM II-M. Hovevez, since the deviations from the mean pressure at SSES follow a probabilistic distribution, usinq the unaltered design mean value sources in IHEGS/MARS vill give loads which have too high a probability of beinq exceeded. Thus, prior to applying the specified mean value sources to I!IFGS/MARS, they vill be multiplied by an amplitude factor whose magnitude depends on the desired exceedance criteria. The lower the exceedance probability the higher the required amplitude factor.

The SSES LOCA load definition applies a separate amplitude f actor for the s Vmmetric and asymmetric load case. -Conversely, the CO time seqmeat requires no amplitude factor.

Pith this ia mind,, this subsection documents the methodology for calculating the symmetzical and asymmetric amplitude factors.

Specifically, Subsection 9.5.3 2.1 explains the statistical analysis used to develop a gualified GKM II-M probability density distribution that forms the basis for determining the amplitude factors. Subsection 9.5.3.2.2 then applies this probability density distribution in a simple aaalytica1 model of the SSES contaiameat with a Monte Carlo procedure to calculate the amplitude factors. Subsection 9.5.3. 2. 2 also quantifies the exceedance criteria chosen for the SSES LOCA load definition and the corresponding amplitude factors.

9a5a3,$ ,1. renera1,.iOn Of the. GKN I~I- Prpbabi~lit DiStributipn In the aext subsection, the SSES amplitude factors will be determined probabilistically based on a qualified probability density distribution of the GKM II-M pressure amplitudes.

However, the comparatively small number of tests with identical initial conditions, as well as the steady chanqe of teststhe overall makes it parameters durinq the duration of one and the same difficult to generate a sufficiently large statistical population for a GKM II-M probability density distribution. If the time Re v. 9, 0 7/85 9P-61

PROP HI ETARY intervals of 5, 15 and 30 sec are used for averaging (see Figures 9-81 to 9-83), then there vould be too small a number of pressure amplitudes to be able to make a credible statistical assertion about the loads.

In order to expand the statistical base, the pressure amplitudes for each test vere first normalized to the sliding mean value.

This vas accomplished by dividinq the maximum amplitude per mean event interval (see Subsection 9.4.2.1.2) by the mean value for the time interval vithin which the maximum amplitude occurred.

Examples of this are shown in Figures 9-179 and 9-180. For each of the HSL Breaks (Test Nos. 3-20), a ranqe vas then defined around the maximum of the mean curve in which the mean values exceed 0.5 bars. The normalized pressure amplitudes with mean values greater than 0.5 bars vere then grouped according to break size {i.e., full MSL, 1/3 MSL and 1/6 HSL) to form the basis for a Zrequenc y (of occurrence) distribution.

The results are illustrated in Piguzes 9-181 to 9-183. For the lower frequency band {0.5-13 Hz), the normalized amplitudes are nearly symmetrical around the normalized value 1. The distributions for the full HSL and 1/3 HSL breaks differ only sliqhtly from one another, whereas the amplitudes of the 1/6- MSL breaks are visibly less scattered. The normalized amplitudes of the upper frequency range (10-100 Hz) are clearly more scattered in comparison to the lower bandpass.

Thus, a data base has been created fzom the test results which allows performinq a probabilistic analysis for SSES to determine the required amplitude factors. The next subsection documents this analysis.

Hev. 9, 07/85 9P-62

PROP RIET AHY 9.5.3.2. 2 Calcu.lation of a Symmetric and Asymmetric Amplitude= Factors To calculate the symmetric and asymmetric amplitude factors, a qlobal symmetric load and a global asymmetric load are first defined as follows:

(I) Symmetric Load: That spatially constant bottom pressure which yields the same total vertical force as the actual local bottom pressure distribution.

2~

P sym (t) = 2

%(r a r 2i )

. BOTTOM SSES where:

s PSSES (9,t)d 2 r = total vertical force BOTTOM r i = outside radius of

. pedestal ra = inside radius of containment wall d~r = sur face elemen t (II) Asymmetric T.oad: That sinusoidally distributed circumferential bottom pressure, superimposed with the symmetric load, which yields the same maximum overturning moment as the actual bottom pressure distribution.

P as~ (t)

= P

~ (t) +

3 3 P (r,t) rxndr (2)

~ g3( a i) BOFJXM Rev. 9, 07/'85 9P-63

PROP RI ET AR Y 2~

whe re: PS~(r,t) r x n d r = maxizmm overturing rxzrent of BOT3XN actual pressure distribution.

r = outside radius of pedestal ra = inside radius of containment wall der = surface element n

a

= norma1 vector of pool boundary azimuth angle Figure 9-184 illustrates the derivation of the maximum overturning moment.

The next step is to transform the above symmetric and asymmetric definitions into analytical equations that can be used in a probabilistic analysis. To do this, the SSZS suppression pool is subdivided into 87 single cells as shown in Figure 9-185. Then a sinusqidally varinq bottom pressure of amplitude A v is assumed in the v~ cell:

P V

(t) = A V

sin (o(t-tV )

where: A = pressure amplitude in v th single cell subscript of the single ce11 in the SSES pool angular frequency deph asing between sinq le cells Thus, for the symmetric load, the definition .(l) is used to obtain:

R ev. 9, 07/85 9P-64

PBOP HIZTAHY 87

() =X v sin u (t-tv, )

v=1 with v  %(r 2 r2 )~

where: rvi . = inside radius of the vth cell r va = outside radius of the vth cell vl v2 angles which enclose the vth cell A = pressure amplitude for the v th cell v

The asymme tzic 3.oad is obtained accordinq to definition gl):

.P BUM

($ ,t) = P S~ (t) + cos ) [P (t)] [P (t)]

87 with P asym (t) =

~

A v=1 v'

X" . sin v(t-t )

3-(1)

-Av(r3 av riv (ms ~v 2 ms %1

%(r r)

Bev. 9, 07/85 9P-65

PROPRIETARY 87 v (2) sin to(t-t )

(9)

(2)

(r3 3 ri ) (sin $ 2 sin 4>> {10)

(ra3 ri) 3 A further simplification can be made by setting -the phases 0 for all vents and assiqninq the same pressure time history at each vent, P (t),

(selected mean chugs) . Thus, equation (3) becomes:

P., (t) = A P (t) (3a)

Then, the symmetric load (phase = 0, identical pressure time historic ) is:

= A (t)

P S~ (w,t) (w) P 87 A (12)

S~

where: A~ = symmetric amplitude factor.

And the asymmetric load (phase = 0, identical pressure time histories) is:

P a~ (),t) = P

~ (t) + A as~ cos) P (t)

Hev. 9, 07/85 9P-66

PROP RIET ARY (14)

Where: A~ = assymme tric amplitude factor.

The spacial profile of the loads from equations (ll) and (13) are shown in Fiqure 9-186.

The individual a mpli tude s, A~, in equat ions (12) and (14) are then ran dom var iables and are prescribed in the f orm of a probability density distribution. For the SSES load definition, the key assumption in Subsection 9. 5. 3.2 reguires this distribution to be the same as the distribution calculated in Subsection 9.5.3.2.1 (see Figures 9-181 to 9-183 ).

The probabilistic formulation nov consists of v ieving the factors A ~an d A s~ as fu nctions of the exceedance probability v the expectation value 2 of P~ into and'nserting equa tion (6) .

For the present analysis, it is assumed that the chug amplitudes in each sinqle cell are synchronized (phase = 0) . Thus, equation (11) becomes:

P (w,t) = A (w) P (t) (15) and equation (13) becomes:

p (w,t) = E(P S~ ) + A (w) cos$ P(t) (16)

II The probability calculation vill nov be used to determine the amplitude factors A and A~~

Rev. 9, 07/85 9P-67

PHOPRI ET AH Y The amplitudes Az (amplitude associated with each single cell) represent a set of N=87 random variables. Each of these random variables sa tisfies one and the same probability density distributions. As previously stated, this is identical to the GKH II-M distributions in Figures 9-181 to 9-183. The amplitude factors As~ and A as are also random variables. The y, in turn, also have probaPilit y density distributions from which the exceedance probabilities cited above are then obtained.

To calculate the A and A ~ probability density distributions the Honte Carlo statistical method is used as follows:

(1) Usinq the random principle (Honte Carlo procedure), pick out one cell from the SSES pool, perhaps the cell v (see Figure 9-185) .

(2) Usinq the random principle (M onte Carlo procedure), ass ig n a normalized pressure amplitude Az to that cell (i.e., from the distribut on in Fiqure 9-181).

(3) Repeat steps (1) and (2) for the remaininq cells (87 single cells) .

(4) Usin q equat ions (12) and (14), resp numerical value for the quantity a histogram.

ecti v ely, A~ and A

~ca lcu late a and plot in (5) Repeat steps (1) to (5) sufficiently often (i e., 4 x 10~

times for SSES).

(6) Normalize the histogram and calculate the exceedance pro babilit y.

The results are illustrated in Figures 9-187 to 9-189 The top part of Figure 9-187 shows the frequency (of occurrence) distributions. The bottom part of Figure 9-187 shows the associated exceedance probabilities for the amplitude factor As calculated for 1, 6 and 87 vent pipes and low bandpass data (O.V-13 Hz)

Figures 9-188 and 9-189 represent the numerical basis for the probabilistic load assessment for SSES. For a given exceedance probability, the associated amplitude factors can be read from the curves. Figures 9-188 to 9-189 were calculated using only the probability density distribution from the full MSL breaks (see Fiqure 9-181), since this data forms a more conservative basis for a probabilistic analysis than the lf'3 and 1/6 MSI breaks (see Figures 9-181 to 9-183) .

To determine the factors, A ~ and A as~, the SSES LOCA load specification assumes a load which is exceeded with a probability of 10-~ per chuqqinq event. This corresponds to an exceedan ce probability of 10-~ per LOCA, based on 100 chugqing events with a Rev. 9 ~ 07/85 9 P-68

PROPRIETARY mean value larqer t han 0. 5 bar. Thus, f rom Figure 9-188 and 9-189 and are:

the high frequency range gives the most conservative factors, li at 10- ~ exc eeda nce '.rob abi t y th e factor A s~ and A ~

A~

A

1 3 0 4 asym,

The above analysis assumes that the chugs occur synchornized at each vent. However, the SSES load definition assumes desynchronized events and it must be shown that the above calculated factors for synchronized events are conservative when applied to a desynchronized load. For the desynchronized case the analysis is more complicated, since the form of the pressure time histories varies. Therefore, to simplify the analysis for calculatinq the desynchronized amplitude factors, that identical events are uniformly distributed in a time window it is assumed 5t =, z with a probability density distribution:

0 t < -x/2 1/v -v/2 <t< v/2 {17) 0 t > 7/2 The probabilistic analysis for the desynchronized events remains the same as for the synchronzied events, except that only for the Fourier component sin +t. Thus, equation (15) and it holds (16) become:

symmetric T.oad { phase > 0)

= A P

S~ (w,v,t) S~ (w,t) P(t) (18) a~smmetric load (phase > 0)

= E(P (w,z,t)) (w,z) cosf P(t)

P as~ (w,z,t) + A as~ {19)

Rev. 9, 07/85 9P-69

PROP RIETARY where ~ si qn i fies the Fourier componen t. Thus, when ca lcula ting the amplitude factors with desynchronized events, the Pourier components are considered by themselves. In other words, the substitution A +A =A.V cos MtV (20) is made in equations (12) and (14). Thus, when using these two equations to calculate the factors with desynchronized events, the GKM XI-M amplitudes are Monte Carloed for each single cell, s well as the time window, t>, in equation (20) . The f actors

~

~ ~

and k are then determined as a function of exceedance pilhabili ty, time window a nd fre queacy.

~ ~

In addition, the expectation value E (mean value) of the pressure in equati on (19) pan be calculated by means o f a frequency dependent f actor Appease . This factor quantifies the reduced mean value at SSZS with desynchronized events relative to the mean value. at SSES with synchronized events. It can be shown that the factor P phase is

0) 2 S331 Q)'t/

(21) phase p if one infinite assumes the probability density ag equation (17) and an number of vents. The quantity Aphase as a function of frequency and time window is illustrated z.n Figure 9-190. The mplitude factors A and as obtained Prom the Monte Carlo calculations are found in Fiqures 9-191 and 9-192, respectively. Note that the time window reduces the symmetric loads, but increases the asymmetric load.s.

To conclude, the above results can be used to determine whether factors calculated with synchronized events are conservative for the vent acoustics between 8 and 10 Hz. The vent acoustic is chosen, because these frequencies are most significant in terms of containment structural loading. Thus, a time window of 50 msec and 10 Hz is selected. From Figures 9-190 and 9-192 Rev. 9, 07/85 9P-70

PROPRIETARY

( G)T=% )

(0 Aph e

= 0.63 A~ = 0.82 A~~ = 0. 23 The factors which assumes A

~that andthe A

~

mean relate to the synchronized events, value for successive chuqs at GKM II-M are the same as the mean value for spacially varing chugs at SSES. However, the mean amplitudes at SSZS for desynchronized events yre distinctively lower than for synchronized events. The

.factor K'pbase quantifies this reduction. Thul, for desynchronized events, the (actors A s~ and A a~ must be zenormalized by the factor A ~~~e . &erefore, for desvnchronized events the amgiztude factors are:

td hl Asym /A phase

= 1.3

6) 0)

Aasym/A p~e = 0-37 These are bounded by the factors A ~ = l. 3 and A a~ = 0. 4 Subsection 9.5.3.4 explains how these factors are applied to the m

sources in WEGS/MARS to calculate containment boundary loads.

9.5.3,3- Determination of the SSFS Unigue ~Chu ging/CO Sources Subsections 9.5.3.1.2 and 9.5.3.3..2.1 selected four (4) chug time seqments and one (1) CO time segment, respectively, for the SSFS I.OCA load definition. This subsection describes the methodology for convertinq the selected traces into souces for use in the I WEGS/MARS acoustic model of the SSZS suppression pool for calculating the containment boundary loads (see Subsection 9 5.3.4) .

To do this, the GKM EI-M pool was modeled based on the assumption that the pressure field due to the steam ccndensation event can be obtained from a solution to the linear acoustic wave equation for a specified source function. This acoustic model is presented in Section 4 of Reference 65. The IWEGS (Inhomogeneous Wave Equation Green Function Solution) computer code was developed to obtain the numerical results cf the analytical solution.

The fundamental equation evaluated by IWEGS in right circular qeometzv is 0

Rev. 9, 07/85 9P-71

PROPRIETARY f or the flexible-wall case. This equation reduces to equation (4.17) of Reference 65 for riqid walls. The symbols for equation .

(1) are a1so Ref ined in Ref erence 65.

In order to determine the source for each selected tra"e, the sonic speed, damping f actor and fluid-structure interaction at GKN II-M must be known.

The sonic speed selection for each of the selected pressure traces is synon ymous with identification of the GKM II-M acoustic zinqout f requency:

c/4 L where: c = sonic speed in pool L = water depth The acoustic frequency is estimated for each selected pressure trace (see Fiqures 9-167, 9-168, 9-169, 9-170 and 9-177 aGb) from the associated power spectial density (see Figures 9-170a and 9-177c) . Figure 9-155 indicates that for the GKM II-M tests, the pool acoustic frequency lies roughly in the range 15-30 Hz. The variation of the pool acoustic frequency is due to variation of air content in the water which greatly affects the sonic speed as shown by Figure 9-156. Thus, that frequency which has the qreatest pressure response between 15 Hz and 30 Hz was assumed to be the poo'1 acoustic c/4L.

Care must be exercised so not to confuse the vent acoustic fundamental frequency (-9 Hz) or the first harmonic of the vent fundamenta 1 (-27 Hz) wi th the pool acoustic frequency. The selected sonic speeds correspond to pool acoustic frequencies in the range 15.0 to 23. 6 Hz as shown below. Thus, this problem did not arise.

The sonic speeds selected are listed below KHU PTH No Source No. Sonic Speed KHU 303 66 1 m/s KHU 305 448.6 m/s KHU 306 512 m/s KHU 309 456.0 m/s KHU 314 420.0 m/s The se speeds reflect the GKM II-M fluid-s tructure interaction (FSI) and SSZS prototypical air con tent.

Dampinq o f t he GKM II-M pressure response is the result of FSI and the presence of air in the watez. 1he ratio of the air dampinq contribution to the FS I contribution is greater than about 3 as shown in the Mark II Load Definition Report. Si.nce Rev. 9, 07/85 9P-72

PROP RIETAR Y the pool air content in GEM II-M was not measured, a nominal value of r, = 0 12 was used for all traces which is consistent with Mark II values.

The treatment of fluid-structure interaction (FSI) at GEM II M is described in Section 5 of Reference 65.

The above information may now be used to determine the sources usinq the method described below.

The Fourier Transform of the pressure field at a location specified by the vector r resulting from the gction of a single source at a location specified by the vector r is given =by {Eq.

(5. 03) of Reference 65)

( ) = H ( /,) S(,) (2) where H (r/ro ) = pG < {r/r o )

is the acoustic transfer function:

2 ~ ~ ' v22 4~ 22-1/2 e ifN H~(rlro ~ ~ ~N ~'r~Nur' g+ 4 2 2

) 4]

with 2 2 2-1 ]

\

g = tarl [2M'G)N + X M )

and VN (e, r) are the flexible wall eigenfunctions (for a complete description of the acoustic transfer function see Section -5 of Reference 65). The above symbols are also defined in Reference

65. S~ (z~) in equation (2) is the Fourier Transf orm of the source function S(t/Po) . Once the tank radius, water depth, sonic speed and damping have been specified, the transfer function is completely known If p (r) is known + (ro) and thus S (t/ro) can be determined via a complex division. Hen~ca a source time history for each spec idied trace is determined bg first taking the Fouriez Transform of the specif ied pressure p (r, t) to obtain ~ (r) . Next the GKM II-M geometry along with the proper sonic speed corresponding to the particular trace and the damping factor are specified and the H>(r/r0 ) is known. S~(r) is then obtained via division and the inverse Fourier Transformer effected to obtain the source function S (t/ro) .

Pressure transducer P6.8 was evaluated by the above methodology for each time segment to yield a set of five {5) source time histories. Figures 9-193 to 9-197 present the five (5) sources.

An indication of the accuracy of the above sourcing procedure is given by F iqures 9-198 to 9-207. These f iqures compare the selected time segments (P6.8) with the pressure time histories calculated with the above sources in IWEGS. They also show a R ev. 9, 07/85 9 P-73

PROP BI ET AR Y comparison of the PSD's of the selected traces (P6.8) with the PSD's of the pressure time histories determined with the sources and IWEGS. As shown in these figures, the final3.y determined sources faithfully reproduce the selected traces and are acceptable for use in the SSES acoustic model.

The next subsec'tion describes the proceduze for calculating the containment boundary loads.

9.5.3.4 Application Procedure foz Calculating the Susquehanna

.-SES-Boundaz oads The previous subsections described the selection of the" GKM II-M time seqments foz sourcing, the methodology for calculating the amplitude factors and the procedure for determining the sources from the five (5) selected traces. Subsection 9 5.3.4.1 and 9.5. 3. 4. 2 explain how this information is combined to calculate the SSES wall loads foz the symmetric and a symmetric load cases, respectively.

9 5 3. 4 1 Somme tric Load. Case Usinq acoustic theory, the SSES suppressicn pool is modeled by IWEGS/MARS -for flexible pool boundaries to calculate the presure as a function of containment boundary location and time using the 4 p ~ Z+~ ~ N{~)/ S {t ) oS~ 5'~(

where Qg(r) and Q (r ) az% given by Eq. (4.33) and Eq. (4.34) of

~

Reference 65 For rigid boundaries, this equation reduces to Eq.

(4.32) of Reference 65. The symbols aze defined in Reference 65.

A cylindrical coordinate system is used for the SSES geometry as shown in Figure 9-208.

The GKM II-M damping factors and sonic speeds (see Subsection 9.5.3.3) for each source (i.e., Source 303, 305, 306, 309 and 314) vere used to cal'culate the pool boundary pressures, since by desiqn the GKM II-N tank has the approximate FSI as the SSES containmen t.

As previously stated, the SSES LOCA load definition employs a 50 msec desynchzonization time window for the four (4) chug time seqments, but does not use desynchronizaticn for the C3 Source 314.

For the desynchzonized sources (Sources 303, 305, 306 and 309) the start time are determined in the same manner as the generic Mark II Program. This involve obtaining a set of source start times from an equal probability distribution over a 50 msec time window by repeated Monte Carlo selections. This is repeated for 1000 Monte Carlo trials one trial being defined as a selection of a start time set (87 start times) . The sum of the deviations Rev. 9, 07/85 9P-74

PROPRIETARY of the start times within a given set is a measure of how coincident the start times are because, for N sources, N

D= S S where D is the sum of the deviations, t s (i) the start time for the ith source, and ts the mean start time. For D = 0 all vents are chuqqinq synchronously. Thus, for the SSES LOCA load definition the set of chuq start times with the minimum variance based on the set of 1000 Monte Carlo selections is used for calculatinq the containment boundary loads, since they yield the highest global pressure response.

The above information is used to calculate the containment boundary loads for the symmetric case as follows:

(1) One of the sources is selected and its magnitude multiplied by the symmetric amplitude factor (see Subsection 9.5.3. 2). This same source is then assigned to each vent exit location (87 vents) in the XWEGS/MARS suppression pool model. No amplitude factor is used with the CO Source 314.

(2) For the chuq sources, the chug start times at each vent are determined as described above. For the CO source, a 11 vents are assum ed inp hase.

(3) ZMEGS/MARS is then used to calculate a set of pressure time histories at the containment boundary locations required by the ANSYS structural model.

This procedure is repeated for each of the five (5) sources. ln addition, the time scales for each set of pressure time histories are expanded and contracted by the factors a+a> and amiz. The values of ~in and mrna~ are the limits of the intervals in Figures 9-167, 9-168, 9-169, 9-170 and 9-177 aGb. This gives if f teen (15) se ts of pressure time histories for containment ana lys is.

9. 5. 3. 4. 2. Asymmetric Load Case The procedure for calculating the asymmetric wall loads is identical as the procedure for the symmetric load case (i.e. same sonic speed, dampinq, governinq equations, etc.), except that the asymmetric amplitude factor is applied to the sources in a different manner. Xn addition, the CO Source 314 is not considered for the asymmetric load case.

The asymmetric load is calculated as follows:

He v. 9, 0 7/85 9P-75

PROPRIETARY (1) One of the sources is selected and its magnitude multiplied by the asymmetric factor (see Subs ection

9. 5..3. 2)

A 'os dependinq on the azimuth location, ) of the vent pipe at which the source vill be located. These sources with varying amplitudes (cosine distribution) are then assigned to their respective vent exit in the IREGS/MA.RS suppression pool model (2) For the chug sources, t he start times at each vent are determined as in Subsection 9 5.3.4 1.

(3) I8ZGS/MARS is then used to calculate a set of pressure time histories at the containment locations required by the ANSY S structural model.

This pro"edure is repeated for each of the four (0) chug sources.

Aqain, the time scales for each set of pressure time histories are expanded and contracted by the factors a+a> and a+in. The values of a+in and a~a~ are the limits of the intervals in Fiqures 9-167 to 9-170. 'his yields twelve (12) sets of pressure time histories f or containment analysis.

Rev. 9, 07/85 9P-76

PROPRIET AHY

9. 5. 3. 5- Verification of the SSES LOCA Load Def inition

9. ~Q. 5; 1 J AEH I: Comma ri son 9 5 3 g.l.l Introduction-To demonstrate the conservatism of the SSES chugging load definition and application methodology, a comparison between the SSES chuqqinq load definition methodology and the JAERI data was performed. This involves applying the sources developed from the selected chug time segments in an acoustic model of the JAERI facility to calculate wall loads foz comparison with the eight large chugs observed in JAERI T'ests 0002. The procedure used for this comparison as well as the results of the comparison, are described in the followinq subsections.

$ ,5.3,5,1 2 ge scription- of Comparison Method This section describes the comparison methodology used for comparinq the predicted pressure response using the SSES chugging load definition against the JAERI chugqing data. The JAEHI data used for the comparison are discussed in Subsection 9.5.3.5.1.2.1. The single vent sources used are described in Subsection 9.5.3.5.1 2.2, and the JAERI acoustic model in which these sources were applied is discussed in Subsection 9.5.3.5.1.2 3 Finally, the comparison basis is described in Subsection 9.5.3.5.1.2.4.

9~ 3.5.1.2.1 The JAFRI.Test Data.

The data used for the data comparison are the eight largest chugs from JAFRZ Test 0002. These eight chugs include the four stronqest chuqs from the six tests selected by JAFRI for spectral analysis. All eight are among the 13 largest chugs from the tests analyzed hy JAERI. Figure 9-266 shows the average pool bottom HMS pressure for the various high amplitude chugs observed in the JAERI tests to,'ate. Although the pool bottom readings are believed to be somewhat distorted by the sensor mounting arrangements, it is reasonable to assume that they provide an adequate ranking of chug strength. Based on Figure 9-266 it is seen that the eight chuqs selected from Test 0002 are indeed among the larqest chugs seen in the JAERI tests to date.

Due to the problem with the bottom sensor as mentioned above, the actual comparisons are made against pressure measurements at the 1800mm and 3600mm elevations in JAEHI. The 3600mm elevation corresponds to that at the vent exit and the 1800mm corresponds to the elevation at mid-clearance. Pressure readings from three sensors on the wall at 1800mm (HRPF-201, 401, 501) and from six wall sensors at 3600mm (HHPF-202 to 207) were made available by JAEHI. The pressure time history from the sensors at a given elevation (3600mm or 1800mm) were first spatially averaged. That is, the pressure time histories were averaqed time step by time Rev. 9, 07/'85 9P-77

PROP RI FTARY step at each elevation. This is appropriate for this data since the spatially averaged pressure represents the global effects of the time varying wall pressure on the containment structure.

This avezaqing tends to preserve frequency components which are spatially in phase and reduce those which are out of phase. The components that are spatia11y in phase are from the pool normal mode response whereas those which aze spatially out of phase are the signals that are transducer specific. I For each o f the eight chugs, the spatially averaged pressure history was computed at two elevations. PSDs of the average pressure histories were computed and the maximu envelopee over m

the eight chugs was constructed. The resulting envelopee PSDs at 1800mm and 3600mm elevations were compared with the predicted pressure responses in the manner described in Subsections 9 5. 3. 5-1. 2. 4.

9.5.3 5 1.2.2 Chug Sources The four (4) pressure time histories (see Figures 9-167 to 9-170) selected for the SSES chuqginq load definition were converted by Bechtel into source time histories as described in Subsection 9.5.3 3.

The sources developed for the SSES load definition are different from those developed for the Mark II Generic Program in that SSES sources are actual souzce time histories as opposed to a combination of triangular impulse and sinusoids used by the Mark II Generic Program. The source time history description provides a much more detailed description of the source and duplicates all the frequency content observed in the single vent pool wall pressure traces.

To verify the integration procedure used with these source time histories, each in dividual source was applied to the GKM II-M single vent qeometzy, and the resulting pressure time history and PSDs were compared aq'ainst those provided by Bechtel. The Creare model was able to reproduce the Bechtel results to within 5% f oz the peak overpressure, within 7% for the peak underpzessure, and within 11% for the total mean square power.

Since the JAERI data represents only chugging, the CO source f rom PTH No. 14 (see Subsection 9.5.3.1 2.1) was excluded from the comparison.

9.5.3.5 1 2 3 - The JA1?RI Acoustic Model The JAERI test facility is a circular tank with concrete fillers to simulate a 200 annular wedge of the Mark II pool. The acoustic model developed at Creare represents the JAERI geometry as a 20o annular wedqe. A comparison between the actual JAERI qeometry and the qeometry used iz the acoustic model is shown in Fiqure 9-267'. The acoustic model of the JAERI facility closely Rev 9, 07/85 9P-78

PROPRIETARY'ollows the acoustic model developed in the Task A.16 Generic Nark II Acoustic Nethodology (see Reference 65). Briefly, the inhomogeneous wave equation is solved for the annular 20~ wedge by means of the Green's function method.

A JAERI acoustic run involves applying the selected source at each of the 7 vents one, at a time. The pressure histories due to the sources applied at each individual vents at each of the sensor locations aze then computed. A set of start times of a particular time trial is selected from a 50msec uniform probability desynchzonization window. Having obtained the start times for individual vents, the pressure time histozy at each sensor location is then synthesized with the appropriate start times. As described in the next subsection, a total of 160 time trials were run foz each of the four chug souces which constitute t he chugging load de fini tion.

9 S. 3..5.1. 2 4 Comparison Basis-As described in Subsection 9. 5.3.5.1.2.1, the comparison is made with eiqht chuqs from JAERI Tests 0002. To make the comparison on a statistically meaningful basis, a simulation was made in the same manner as that used for obtaining the envelopee of the actual eiqht JAERI chugs. That is, eight time trials or "chugs" were run in the JAERI acoustic model and a PSD~ envelopee of the eiqht trials was constructed. To make the envelopee of the computed pressure'ime'istories insensitive to the rand,om stat'istics of tne eiqht time 'trials selected, this procedure was repeated 20 times (i. e., 8x20=160 time trials) ., This results in, 20 PSD envelopees (one PSD envelopee for a set of eight time trials) . These PSD envelopees were then averaged to obtain an average PSD envelopee for a particular source. Such an average envelopee was then constructed for each of the f cuz sources.

Foz the four souzces, the average envelopees for each source were envelopeed and this envelopee is then compared against the envelopee of the eight chugs in the JAERI tests. In addition, the symmetric amplitude factor (see Subsection 9.5.3.2) was not applied to the vali loads calculated by the SSES load definition for comparison with JAERI. Thus, the comparison represents mean value chugs at GKN II-M.

However, the PSD's qenerated by the SSES chugging load definition reflect the use of the time expansion and contraction factors (see Figures 9-167 to 9-170) .

P The comparison basis described above is identical to that used by the Nark II Generic Program.

~The PSD's generated by the SSES sources were nozmalized to account for their difference in time duration when compared to the time duration of the JAERI chuqs.

Rev. 9, 07/85 9 P-79

PROP RIETAR It'.

5.3.5. 1.3 Results and Discussion.

The results of the comparison are shown in Figures 9-268 and 9-269 at the two elevations in JAERl. As described earlier in Subsection 9.5. 3.5.2.4, the envelopee foz the sources excludes the souzce for the CO period in the GKM II tests as well as the symmetric a mpli tude factor.

Fiqures 9-268 and 9-269 show the comparisons at the 1800mm and the 3600mm elevations, respectively. From these figures seen that the SSFS load definition bounds the JAERI data by a it is substantial marqin over the entire range from 0-100 Hz.

In conclusion, a comparison between our chugging load definition and the eiqht larqe chuqs observed in JAERI Test 0002 has been presented. The comparison has been made on a basis which is statistically meaningful and shows that the current SSES chugging load definition developed from GKM II-M is very conservative.

9. 5.3 5. 2 Verifications of the 50 msec Time Rindow Verification of the 50 msec desynchzonization time window will be documented in the Generic Mark .II Load Definition Report i

scheduled fo r submit ta 1 n A pri 1 of 1981.

9.5.3.5.3 Response to NRC Concerns Regarding the SSES C hu~ing Load Specification This Subsection responds to a zequest from the NRC staff (Reference 81) to document arguments demonstrating the adequacy of the SSES chuqqinq load specification. Essentially all of the information contained in this subsection was presented at a meetinq held in San Jose, Cali fornia on April 8, 1982 between zepresentatives of the Mark members of the HRC staff.

lI Owners Group, SSES personnel and Feference 81 specifies two elements required to address the staff 's concerns:

<<1. Formally document the material which was used to conclude that the asymmetric chugging load specification is not a major contributor to the structural response. Although not specifically discussed durinq the meeting, the response should address the validity of the rationale over the entire frequency ranqe of interest (0 to 50 Hz).

2. Extend the JAERI comparisons to 20 different sets of start times to be generated in the same manner as outlined in the qeneric methodology. For each set of start times evaluate the response of applying the chuqqinq sources to the JAERI facility and develop an Rev. 9, 07/85 9P-80

PROP RI ETAR Y envelope of the minimum of the twenty trials.

Comparison of that envelope to experimental data obtained from .'JAERI would then be provided. If frequency <<poke through>> exists, justification would be provided as to why the current specification is adeq ua.te. "

9. 5. 3.5. 3. 1 Contribution of the Asymmetric Chugging Load

~ - . Specification to Plant Structural Response g 5.3.5.3.-1 1 Comments on the A~smmetric Chuggi~n Load Case This section provides comments on the effect of vent stazt times on the asymmetric pool wall loads generated during chugging. As with symme tric loads, a unique vent start time set (STS) will produce a unique asymmetzic pool response. However, analysis of the complete pool transfer function shows that the primary asymmetric pool response in terms of a net <<overturninq<<moment is generated only by the first asymmetric mode in the frequencynet range between 15-100 Hz. No other asymmetric modes produce a overturning moment.

Piquxe 9-271 shows a schematic of the first asymmetric mode in the SSES pool qeometry. The mode shape for the pressure qenerated by a single vent is a cosine function, therefore, the

<<area of influence<< for that vent is quite large. Due to the

<<large area of influence<<, considerable overlapping cccurs between adjacent vents.

Puzther, foz the first asymmetric mode, the pressure variation in the radial direction is small. This means that vents located on a given radial ray but at, different radii have roughly the same effective <<moment arm<<.

Finally, in a multivent qeometry where events are randomly desynchronized tsuch as in the SSES case), the asymmetric mode is not stationary in space It rotates; hence there is no fixed axis in time about which the asymmetric mode produces a full amplitude <<overturning moment<<.

Due to these considerations, the effect of a selected STS on the "overturning moment" is much less than that predicted by simple analyses where the <<area of influence>'s taken to be a small area under each vent and the moment arm as the orthogonal distance between the vent and a fixed axis in time. It is expected, therefore, that in reality, the effect of vent start times on the asymmetric pool responses will be no more severe than those for the symmetric response.

9 5,$ ,5,3 1,g Generic Position on QNP-2.Suhmittal The Mark Part A II utilities of the and their respective A/Es have reviewed HNP-2 report (Reference 82) and compared it with Rev. 9, 07/85 9P-81

PROPRI ETAR Y t heir ow n con tainmen t analyses. All utilities have concluded that Mark II plants exhibit to those identified overall characteristics in the report:

and trends similar 0 The re "rock" is no evidence t hat the Mark due to hydrodynamic loads.

II con tainments o In a concrete containment, the plant response to chugging loads decreases as one moves away f rom the wetwell in the primary containment. Once outside the primary containment, the response is significantly red uced.

When reviewinq separate responses for the symmetric and asymmetric chugging load, very little difference is observed.

This result has been confirmed on the HHP-2 plant with a subsequent submittal to the NRC. (Reference 83) .

As an additional example of the structural response, this time for the SSES concrete containment, selected acceleration response spectra (ARS) curves generated with our chugging load definition are presented in Fiqure 9-273 through 9-280. Each figure compares the ARS curves for the symmetzic and asymmetric load cases at the node points in the ANSYS model identified in Figure 9-272. For most containment locations, the symmetric and asymmetric ARS curves are guite similar in both frequency content and ampli tude. Also, observe the decrease in amplitude of the responses at increasing elevation above the containment basemat.

The curves represent the en velope- o f all design sources.

However, in some containment locations, at certain frequencies, the symmetric ARS curves exceed the asymmetzic ARS curves, and vice versa, by more than 10%%uo. Cur explanation is as follows:

Fiquzes 9-273 to 9-276: These fiqures show the responses in the vertical direction at elevations above the diaphragm slab. These figures indicate that the symmetric response always exceeds the asymmetric, and at some frequencies by more than 10%. This is expected, since the symmetric souzce distribution was specified to maximize the vertical pressures on the containment .basemat These pressures ~ould excite the vertical modes of the containment in a more significant manner, than the asymmetric case. This is especially true in the low frequencies, where desynchronization does .not siqnificantly effect the pressure response, and the symmetric source distribution leads to a qreater source strength than the asymmetric load case.

This greater source strength in low frequencies, also excites the hiqhez frequencies as explained in the JAERI PRS comparison (Subsection 9.5.3.5.3.2.3.2) .

Re v 9, 07/85 9P-82

PROP R IETAH Y o Figure. 9-277: Please note that the ARS curves for Node 841, Qirection X, presented at the April 8, 1982 NHC/SSES meeting and documented in PLA-1063 weze labeled incorrectly. The curves should be reversed, such that the asymmetric ARS curve now exceeds the symmetric curve in the low frequencies. Figure 9-277 shows the correct labeling of the curves.

At all frequencies, except between 4 and 8 Hz, the two load cases compare quite well. As a result, we zequesteQ that Bechtel examine the natural fzeguencies and participation factors for Node 841. They indicated that in these low frequencies, there are significant modes in the horizontal direction. Thus, one would expect the asymmetric response to exceed the symmetric response for Node 841 at these frequencies, since desynchronization does not significantly effect the pressure distribution at these low frequencies and the asymmetric source distribution was specified to maximize the asymmetry in the containment.

o Figure 9-278: This figure exhibits a similar comparison as Fiqure 9-277. Aqain, at these low frequencies, Nod@ 411 contains significant lateral modes. Thus, as for Figure 9-277, one would expect the asymmetric response to exceed the symmetric response.

o Figure 9-279: 'his figure indicates that above 30 Hz the symmetric response exceeds the asymmetric response by a large margin. At 'this containment azimuth .

location, we determined that the Mean Sguare Power (MSP) for the pressure at the intersection of the containment wall and basemat, varies for the symmetric and asymmetric load case. Our analysis indicated that the symmetric MSP exceeds the asymmetric MSP at this azimuth location; probably due to the effects of the chug start .times for downcomezs in close proximity to this azimuth angle. Because of this, one would expect t he symmetric case to exceed t he asymmetric case.

Fiqure 9-280: For all frequencies, the symmetric and asymmetric load cases exhibit a similar response.

Thus, even with the source strength unbalance imposed by the asymmetric chugging load definition, the structural response is not noticeably different from that of the symmetric load case.

It can be concluded from the overturning moment associated preceding discussion that the with asymmetric chugging need not be used as a criterion in assessing the adequacy of the load definition. This is true for the reasons outlined in Subsection 9.5.3.5. 3. 1. 1 and because the asymmetric pressure distribution can also be found in the symmetric load specification due to Rev. 9, 07/85 9P-83

PROP RI ETAR 7 source desynchzonization. Thus, the adequacy of the SSES chuqqinq load definition can be demonstrated through a comparison.

of the symmetric load specification with the multivent JAERI, data.

9. 5.3.5.3 2 Comparison of Minimum Variance Trials Using the SS r SQGKH=- II-M Sources- with the JAEHI Data 9.$ . 3. 6. 3. 2. 1 Introduction-The SSES Chuqqing Loads Methodology specifies the use of the start time set fSTS) that has the minimum variance in 1000 STSs drawn randomly from a 50 msec desynchzonization time vindow.

This minimum variance STS is then used for the application of all the SSES/GKM II-M Chugging Sources.

qiven set of start times produces a unique pressure time history at the pool wall boundary, with a signal attentuation (referred to as frequency holes) occurring at some frequencies and amplification at others, due to the vent-to-vent phasing implied by the selected set of start times. A concern vas therefoze raised that the use of a single STS in evaluating the SSFS chuqqinq loads miqht result in non-conservatisms in frequencies vhere signal attentuation or>>frequency holes>>

occurred.

Althouqh it is true that "frequency holes" vill be produced with any one set of start times, the overall conservatism in the strengths of the SSES/GKM II-M Sources will still produce a conservative response of the SSZS containment over the entire frequency range used in the plant structural evaluation. To demonstrate this, the SSES Chugging Loads calculation procedure was applied in the JAERI geometry and the results weze compared against the bounding JAERI data. The following describes the comparison and presents the results obtained. The results confirm the consezvatism in the SSES Chugging Load Definition.

First, one thousand STSs were drawn randomly from the 50 msec desynchronization time window. Note that each STS consists of seven start times--one for each of the seven vents in the JAERI qeometry. The STS having the minimum variance was then identified. This procedure vas repeated 20 times and 20 such minimum variances STSs vere obtained.

Usinq each minimum variance STS, the SSES/GKM II-M chug sources were applied one at a time in an acoustic model of the JAERI test facility. Spatially averaged pressure time histories (one for each source) vere then generated at the 1. 8 meter and 3.6 meter elevations. The power spectral densities (PSD) vere then obtained for these pressure time histories. From these, PSD envelopes over the SSES/GKM II-M sources were constructed at the Rev. 9 07/85 9P-84

PROPRIETAR Y 1.8 meter and 3.6 meter elevations. Similarly, PRS envelopes were obtained at the 1.8 meter elevation. This procedure was performed usinq each of the 20 minimum variance STSs. Thus, at each of the two elevations, 20 PSD envelopes resulted--one for each of the 20 minimum variance STSs, and 20 PRS envelopes were obtained at, the 1. 8 meter elevation.

Finally, minimum PSD envelopes for the 20 minimum variance STS vere constructed at the 1.8 meter and 3. 6 meter elevations; a minimum PRS envelope vas constructed at the 1.8 meter elevation.

These minimum PSD and PRS envelopes vere then compared with the corresponding maximum= envelopes of the eight largest chugs f or JAERI test 0002. Hote that the PHSs for the JAERI data were obtained by digitizing the JAERI pressure time histories to produce the respective PRSs. The comparison and conclusions are presented in the next section.

Before proceeding to discuss results, several important points require further clarification. First, the SSES/GKM II-M Sources used in the JAEHI acoustic model to calculate the JAERI pool wall pressure time histories used the same sonic speed as that given in the SSES Chuqqing load definition report. That is, the sonic speeds derived for the GKM II-M test facility were used with each of the SSES/GKM II-M sources. The sonic speeds for the SSES/GKM II-M ranqes between 661m/sec and 449 m/sec.

The sonic speeds in flexible wall facilities are related to the equivalent ziqid wall sonic speeds by the following relation:

C = C (1+pC 6) 2.

vhere is the watez density, C is the flexible wall sonic speed, C o is pthe ziqid wall sonic speed and 6 is volume distensibility+. Figure 9-281 shows the rela tionships between the riqid wall soni":speeds and flexible wall sonic speeds for GKM II- M an d JA E RI facilities.

Ideally, the GKM II-M ~'rigid wall~~ sonic speeds would be obtained first by usinq the above relation. The rigid wall sonic speeds would then be cozrec ted to obtain the equivalent flexible wall sonic speeds for JAERI'. These corrected sonic speeds would then be used with the corresponding sources in the JAEHI acoustic model for predicting the JAEHI wall pressures. However, this procedure was not used because the differences between the GKM II-M flexible wall sonic speeds and the corresponding JAERI flexible vali sonic speeds is less than 7% over the range of 6 = 139. 90 x 10 m/n for GKM II-M facility.

6 = 497.38 x 10 m/n for JAERI facility.

Hev. 9, 07/85 9P-85

PROP BIET ARY sonic speeds (661 m/sec to 449 m/sec) for the SSES/GKM II-M souzces.+~ This small difference did not warrant the added computational complexity and hence the GKM II-M sonic speeds were used in the JAERI wall pressure computations.

The second point that needs to he clarified is with respect to the dampinq values used in the calculation of the PRSs. The PRSs weze calculated usinq 4% and 7% damping. These are the damping values described in USAFC Reg. Guide 1.61 for reinforced concrete con tai amen ts.

The final point is regarding the amplitude factors used with the souzces. The SSES chugqing loads methodology calls for multiplyinq the SSES/GKM II-M sources .by an amplitude factor to achieve an exceedance probability of 10-~ For the 87 vent SSES containment geometry, an amplitude factor of 1.3 is derived.

Since the objective of the comparisons presented here is to compare the results of applying the SSES chuqging loads methodoloqy in JAZRI with the actual JAZBI data, one must compute the amplitude factor appropriate for the seven vent JAERI geometry. The amplitude factor for the JAERI seven vent geometry is 1.95 to obtain an exceedance probability of 10-~ used for the SSES plant evaluations. The comparisons presented in the next subsection aze made usinq the amplitude .factor of 1.3 corresponding to that used for the SSZS chugging loads evaluations, as well as the ampUtude factor of 1.95 which is the correct one for application of the SSES chugging loads

. methodology in the JAERI facility.

9 $ .g Q. 3.g.Q - Results and Discussion As described in the previous Subsection, two types of comparisons were perf ormed one on a PSD basis and one on a PRS basis. The PSD comparisons are presented in Subsection 9.5.3 5.3.2.3.1 followed by the PRS comparisons in Subsection 9.5.3.5.3.2 3.2.

9.5.3.5 3.2.3.1- PSD Comparisons Figure 9-282 shows the comparison of the minimum envelope of 20 minimum variance STSs with the JAEBI data at the 1.8 meter elevation. The amplitude factor used here was that for the SSZS containment and equal to 1.3. From this figure it is seen that the minimum envelope of the 20 minimum variance STSs hounds the JAFRI data by a substantial mazgin at frequencies below about 40 Hz. Above 40 Hz, the siqnal levels are quite small and so from this linear plot it is difficult to make many comparisons.

~>This is because most of the "flexibility" comes from the air in the pool rather than from the structural flexibility.

Rev., 9 07/85 9P-86

PROP RI ET AR Y Therefore, the same data plotted on log scale are shown in Figure 9-283. From this fiqure drops below the maximum it JAERI is seen that the minimum envelope data envelope hy a small amount in the frequency range between 40 and 80 Hz.

Fiqure 9-284 shows the comparison of the minimum envelope of 20 minimum variance STSs with the maximum JAZRI data envelope at the 3.6 meter elevation. From this linear plot, that the minimum envelope of the 20 minimum it is again seen variance STSs bounds the JAERI data envelope by substantial margins, and both the JAERI data envelope as well as the 20 minimum variance STSs envelope drop to very low levels beyond 30 Hz. Figure 9-285 shows the same data plotted on a log scale, and it is seen that in the frequency ranqe between 40 and 60 Hz, the JAERI data envelope "pokes through" the minimum envelope by small amounts.

As mentioned earlier in Subsection 9.5.3.5.3. 2. 2, the appropriate amplitude factor for the seven vent JAERI geometry is 1.95.

Comparisons are now made using the 1. 95 amplitude factor. Figure 9-286 shows the comparison of the minimum envelope of 20 minimum variance STSs with the maximum JAERZ data envelope at the 1. 8 meter elevation.

the It is seen that with this amplitude factor of variance envelope bounds the JKERI data by 1 95, minimum substantial marqins everywhere except for some small "poke throughs<< in the f requency range between 45 and 80 Hz. Figure 9-287 sho~s a similar comparison at the 3. 6 meter elevation and at this elevation, the minimum envelope of the 20 minimum variance STSs Mounds'the ZALRI data at all frequencies.

Prom the comparisons presented ahove, the following conclusions can be drawn:

1. Usinq an amplitude factor of 1.3 (the SSES value), the minimum envelope of 20 minimum variance trials bounds the maximum envelope of JAZRI data except for small

<<poke throughs<<between approximately 40 and 80 Hz.

2. Usinq an amplitude factor of 1.95, which is the correct factor for the seven vent JAERI geometry, the minimum envelope of the 20 minimum variance trials bounds the maximum envelope of JAERI data everywhere at the 3.6 meter elevation and there are very small "poke throuqhs<<between 40 and 80 Hz at the 1.8 meter e leva tion.

Comparisons on PRS basis are more appropriate than PSD comparisons in evaluating the effect of a given pressure time history on a multi-modal structure such as the SSES containment.

This is because the response of the multi-modal structure at a qiven modal frequency is the sum of the resonant response to forcinq function amplitude at the modal resonant freguency plus Rev. 9, 07/85 9P-87

PROP RIETAR Y the static or forced response due to forcing function amplitudes at other frequencies. A PSD only shows the component of the forcinq function at a qiven frequency and therefore, does not provide insight into the static or forced response of the structure. A PRS on the other hand takes into account both the resonant and the static or forced response and hence provides a more accurate picture of the response of a multi-modal system.

The comparison of the minimum PRS envelope for the 20 minimum variance STSs with the maximum PBS envelope of the JAERE data at the 1.8 meter elevation is shown in Figure 9-288 For the PRSs shown in this fiqure, the SSES amplitude factor of 1.3 was used with a damping value 47'. It is clear from this figure that the minimum envelope of the 20 minimum variance STSs bounds the maximum PRS envelope of the JAERI data throughout the frequency range between 0-100 Hz. Figure 9-289 shows the PRS comparisons for an amplitude of 1 95, and a 4% damping Again this figure shows that the minimum envelope of the 20 minimum variance trials bounds the maximum JAERI data envelope by significant margins across the entire fze quency range.

Finally, Fiqures 9-290 and 9-291 show the PRS envelope comparisons for the 7% damping case with amplitude factors of 1.3 and 1 95, respectively. From these figures, it is again seen that the minimum PRS envelope of the 20 minimum variance STSs bounds the maximum JAERI data PRS envelope by a substantial marqin throughout the frequency ranqe.

From the comparisons presented due to the 1arqe conservatisms above, it can be concluded that in the strength of the SSES/GKM II-M Sources, the response produced by the minimum variance STSs will produce conservative plant responses in spite of the

>>frequency holes~'reated by a particular STS selection.

Therefore, the SSES/GKM II-M chucging loads methodology is adequate Rev. 9, 07/85 9P-88

PROPRIETARY 9 6 VERIFICATION OF- THE DESIGN SPECIFICATION This section provides information verifying the conservatism of the DFFB steam condensation load definition (see Subsection

4. 2. 2) an d the MK II single vent lateral load definition (see Reference 47) .

9 6.1 Evaluation of the- DFPR CO and C~hu pic Load Specification Currently, all SSES plant design assessment for LOCA steam condensation loads employ the DPPR CO and chugging load specification developed from the original 4T test program (see Appendix A of Reference 21 and Reference 16) .

However, the NRC in NURPG 0487 expressed concern about the conservatism of the DPFR specification because of the non-

. prototypical vent lenqth used in the 4T test facility. As a result, PPBL initiated the GKM II-M test proqram to resolve the NRC's concerns. Subsection 9.4 documents the results of the GKM II-M tests and Subsection 9.5 presents the SSES LOCA load definition resulting from the GKM II-M data base.

In this subsection, a comparison of the DFPR LOCA .load and the SSES LOCA load definition is provided. This evaluation is accomplished by comparing the DPFB containment acceleration response spectra (ARS) curves with the containment ARS curves qenerated by the SSES specification. This comparison study is made f or LOCA (DPPR) vs. LOCA (KQU), as well as the combination LOCA (DPPR) + SSZ + SBV vs. LOCA (KNU) + SSE + SRV. Subsection 9.6.1.1 presents the results of the two comparison cases.

Subsection 9.6.1.2 provides an evaluation the ARS comparison.

Finally, Subsection 9.6.1.3 summazizes the LOCA loads comparison.

9.$ .1.1 . Containment Acceleration Re~sonse Spectra Comparison

9. 6. 1. 1. 1 LOCA (DPPB) vs. LOCA (KNU) Acceleration Response Sgectrg Co~m prison Piqures 9-209 thzough 9-218 compare the containment horizontal ARS curves due to the DFPR LOCA load definition (LOCA(DPFR) ) with the containment horizontal ARS curves generated by the SSZS LOCA load definition (LOCA (KHU) ) for 2% spectral damping. Figures 9-219 through 9-227 compare the vertical LOCA (DFPR) ARS curves with the vertical- LOCA (KNU) ARS curves for 2% spectral damping.

The LOCA (DFFB) curves represent envelope ing spectra and were qenerated as f ollows:

o The DPFR LOCA load consisting of chugging and CO (see Appendix A of Reference 21 and Reference 16), each of which contain three (3) frequencies, were inputed to the 3-D ANSYS structural model (see Fiqure B-1) to calculate ARS curves at the required node points. Both an asymmetric and symmetric Rev. 9, 07/85 9P-89

PROPRI ETAR Y load case were considered for chugging, vhile only a symmetric load case was considered for CO.

o These individual LOCA (DFFH) ARS curves were then envelopeed into one (1) AHS curve at each nodal point. This was done for both horizontal and vertical responses.

o The ARS curves .for nodal points at approximately the same elevation were then further envelopeed to give one (1) ARS curve for each required elevation. Table 9-14 gives the nodal points envelopeed at each elevation for the LOCA (DFFH) spectra and 3-D ANSYS model. Additionally, the peak frequencies of the spectra were broadened hy 15'o account for any uncertainties in the modeling techniques and material properties.

Similarly, the LOCA (KWU) curves represent envelopeing spectr a and vere obtained as follows:

o The SSES LOCA load def inition vas used to calculate fifteen (15) sets of symmetric vali loads a nd twel ve (12) sets of.

asymmetric wall loads (see Subsections 9.5.3.4.1 and 9.5. 3.4.2, respectively) for input to the ANSYS model. The ANSYS model then generated ARS curves at the required nodes for each set of pressure time histories.

o These individual LOCA (KWU) ARS curves were then envelopeed into one (1) ARS curve at each nodal point. This was done for both horizontal and vertical responses. ~ .

o The ARS curves for nodal points at approximately the same elevation were then envelopeed to form a representative curve for each elevation. Table 9-14 qives the nodal points envelopeed at each elevation for the LOCA (KWU) curves and 3-ANSYS model. The LOCA curves vere then broadened by 15't D (KWU) the peak frequencies.

9.6.1.1.2 SSE + SRV (ADS) + LOCA(DFFR) vs. SSE + SRV(ADS) + LOCA(KWU)

Acceleration Response Spectra Cc~marison Figure 9-228 through 9-237 compare the containment hori'zonta1 AHS curves for the combination SSE + SRV (ADS) + LOCA (DFFR) with the containment. horizontal- ARS curves for the combination SSE +

SRV (ADS) + LOCA (KWU) for 2% spectral damping Figures 9-238 throuqh 9-246 compare the vertical containment ARS curves for the same combination with 2,. spectral damping. These combination curves vere obtained by combining the individual spectra hy the absolute su m method (SSE + SHV + LOCA) .

The LOCA (DFFR) and LCCA (KWU) spectra combined with the remaining spectra vere generated as described in Subsection 9.6.1.1.1.

C Rev. 9, 07/85 9P-90

PROPRIETARY The SBV ARS curves combined with the other spectra were obtained as follows:

The three oriqinal K~iIU traces (see Subsection 4.1. 3. 5) were contracted and expanded in time to give five different traces between 55%%u and 110% of the frequency of the original trace. This qives f ifteen sets of pressure time histories for input to the 3-D ANSYS model.

The ADS load ca se is consi dered for combination with LOCA and SSE. However, the azimuth distribution on the containment boundary as discussed in Subsection 4.1. 3 indicates that the all valve case governs the ADS case for symmetric loading. Therefore, for the combination SSE + SRV

+ LOCA, the all valve load case (see Subsection 4.1.3.1) was used to calculate the SRV spectra.

The 3-D ANSYS model then produced fifteen sets of spectra at the required nodal points. This was done for the horizontal a nd ve r tie al spectra.

The fifteen spectra were then envelopeed to give one spectra for each node and then further envelopeed with the spectra f or nodes at approximately the sa me eleva tion. This yielded one ARS curve for each elevation. Table 9-14 indicates the node points envelopeed at each elevation for the 3-D ANSYS model and SHV loadinq. The envelopeed spectra were then broadened by 15%%u at the peak fzequencies.

The SHV spectra combined with the SSZ and LOCA(DFPR) spectra re fleet the use of a l. 5 peak pressure 4.1.3.5) applied to the SHV pressure time histories.

multiplier (see However, for this comparison, the SHV spectra combined with the LOCA(KHU) and SSE spectra zef lect the use of a 1.1 peak .

pressure multiplier. This is consistent with the NRC s acceptance of a peak pressure multiplier of 1.1 for all actuations documented in NUHEG 0487-Supplement No. 1. Again it must be emphasized that the 1.1 multiplier is for comparison only and that all SSES design assessment for SRV loads use the 1.5 peak pressure multiplier.

The SSE AHS curves combined with the other spectra were obtained usinq the 2-D seismic stick model (see SSES FSAR Figures 3.7b-7 and 3.7b-8) . The modal properties of the stick models are found in SSZS FSAR Table 3 7b-5 and Figures 3.7b-l4 throuqh 3.7b-19. SSES TSAR Section 3.7b gives a detailed. description of the methodology for determining the ARS curves due to SSE. Again, for the present comparison, the spectra for different nodes at approximately the same elevation were envelopeed to obtain-one spectra for each elevation. Table 9-14 gives the node points envelopeed at each elevation for the SSZ and 2-D stick model.

He v. 9, 0 7/'85 9P-91

PROPRIETARY 9 6 1.2 -Containment Response Spectra Evaluation A comparison of the containment acceleraticn response spectra

{ARS) generated by the SSES LOCA load specification and the DPPH chuqqinq and condensation oscillation load specifications shows the f ollowinq:-

{1) The SSFS LOCA load specif ication generally result in responses qreater than the DPBH LCCA load specification in two frequency ranges o 10 to 20 Hz o abo ve 40 Hz (2) From 20 Hz to 40 Hz the DFFH qenerally bounds the SSES LOCA load specif ication response.

To provide a more meaninqfu1 evaluation, we have also produced containmen t AHS curves for the lead combination which is generally design controlling:

SSE + SHV (ADS) + LOCA As described in Subsections 9. 6.1.1.1 and 9.6.1.,1.2 these response spectra curves utilize either the DPPH LOCA loads or the SSES LOCA load specification as the input LOCA load for generation of these combined response spectra curves. As also indicated in Subsection 9.6.1.1.2, the SBV load specification used for combination with the new SSES LOCA response spectra utilizes a 1.1 amplitude multiplier. The SHV loads used in combination with the DPFB LOCA response spectra utilize a 1. 5 amplit ude multi plie r.

An evaluation of these response spectra curves shows the following:

(1) The combined response spectra which includes the SSES LOCA loads generally result in responses greater than the DPFH combined response spectra for frequencies qreater than 40 Hz.

(2) For horizontal responses below 40 Hz the DFFR combined response spectra generally bounds the SSES combined response spectra.

(3) For vertical responses below 40 Hz, the SSES combined response spectra generally exceed the DPFR combined response spectra.

These evaluations indicate that in order to verify our design basis additional work will be required. Two general approaches can be utilized:

Rev .. 9, 07/85 9P-92

PROP RI ETAR y (1) Reduce the input load (2) Be-assess the affected components to determine whether suff icient design margin exists.

We vill discuss our plans for purusing both options.

9.6.1 g.l Load Reduction Assessment The degree of conservatism that exists in the SSES LOCA load definition can be easily seen in the comparison of this load specification and the available JAERI multivent test data given in Subsection 9.5.3.5.1. As indicated in Subsection 9;5.3.5.1, the SSES LOCA load definition without load amplitude factors (mean value load) bounds the available JAERX data by factors of 2 to 10. This indicates that substantial room for load reduction exists. The followinq subsections qive the areas of load reduction we will be pursuinq.

9 6.1.2.1.1 Fefuctiou of Loaf~Am litufe Factors As described in Subsection 9.5 3.2.2, symmetric and asymmetric amplitude factors were generated .for application to the selected chuqginq pressure time histories in our LOCA load specification.

The selection of these amplitude factors was based on a probability calculation. An exceedance probability of 10-~ per chuqqinq event was selected. If one assumed the probability of a LOCA occuring as 3.0-6, then the combined probability of having a LOCA and exceeding the specified load definition is 10- ~~. Me believe a high deqree of conservation will still be zetained in the load definition if a reasonable reduction in the specified exceedance probability is taken. As can be seen in Figures 9-190 6 9-191, the followinq amplitude factors result with the indicated change in exceedan'ce probability:

Symmetric Asymmetric Exceedance Probabilit'y Amplitude Amplitude (wv =0 ) Factor Factor 10-~ 1. 3 .37 10-~ 1 27 35 10-> 1 23 .31 10-> 1.17 25 1 g 1 2 'Re-Selection of Pressure Time Histories As described in Subsection 9.5.3.1. 2, the selected chugging pressure time histories were to be selected as representative of the mean value events. The selection basis was an evaluation of peak over pressure and observed oscillation frequency. As a check of the selected pressure time histories, a Power Spectral Density (PSD) evaluation was performed. This evaluation is described in Subsection 9 5.3.1.3.1. Figure 9-178a shows the Rev. 9, 07/85 9P-93

PROPRI ETARY final PSD comparison. As can be seen the selected mean value events substantially bound the mean value PSD for Tests 3/4, 9/10, 11/12, 13/14 and 19/20. This indicates that the selected chuqqinq events are actaully representative of events substantially above the mean.

application cf amplitude factors If this proves to be true, the would not be appropriate for these selected pressure time histories. A re-selection of new mean value events based on power will be performed.

9,$ ~ 1 3 . Adoption of. Mark II Ovne~rs Grou Load MethodolocCy Another option for potential load reduction would be use of the load methodology developed by GE for the Mk different selection II Owners Group.

for design This methodology uses a bases chuqs. These design chuqs are the average of the seven largest chuqs observed in the 4TCO data with their largest neighbor chug.

This methodoloqv could be used with the GKM IIM data base to develop an alternate load definition.

9.6.1.~1.9.~Develo ment of a Meu Chuggi~n Load Methodo~lo As a further option, the basis for a new chugging load methodology exists. This new methodology would consist of the development of a series of design sources from GKM IlM tests which represent bounding blowdown conditions. These sources would be developed from the actual chugs which occur during the selected time seqments for each test. These sources would then be applied randomly to an acoustic model of the Susquehanna suppression pool. In addition random source start times would be employed. This load definition would give the most realistic overall load input for plant assessment.

9. 6. l. 2.2 Plant Re-Assessment As a result of the comparison documented in Subsection 9.6.1 and Bechtel's opinion that a re-assessment of SSFS based on the GKM II-M load definition would not significantly impact our projected fuel load date, PPGL on April 1, 1981 decided to terminate the assessment of SSES based on the DFFR chugging and CO specification and re-evaluate SSES based on the GKM II-M load specif ica t ion.

Section 7. 0 pro vid es the results of this re-assessment.

9. 6. 1. 3 Summary The new SSES LOCA load definition results in containment responses which exceed, in many frequencies, the responses obtained from the DFFR chugging and condensation oscillation load.

Rev. 9, 07/85 9 P-94

N PROP RIET AH Y He are presently reviewing the degree of conservatism which exists in the SSES LOCA load. There are several azeas where load reduction could be obtained. In addition, several other options f

e xist or general 1oa d reduction.

On April 1, 1981, PPSL apted to terminate the re-evaluation of SSES with the DFPR load definition, and instead pezf ormed a re-evaluation of the plant based on the GKH II-H load specification.

Rev. 9, 07/85 9P-95

PROPRIETARY

9. 6.2 Verification of the Mk II Sinqle Specification Vent Dynamic Lateral Load The SSFS downcomer bracinq system has been analyzed using the dynamic multi-vent lateral load specification transmitted under Task A. 13. This specification applies the Mark II single vent dynamic. lateral load specification (see Reference 47) in con junction with a multi-vent multiplier whose magnitude depends on the number of vents analyzed and the desired exceedance probability. The single vent load specification was developed by Pretech Inc. usinq the oriqinal 4T data base. However, the HRC has expressed concern about the conservatism of the single vent load definition. He therefore measured lateral loads during the GKM IIM test and committed to compare those measurements with the load specification.

As a result, the following subsections compare the maximum resultant bracinq force measured at GKM II-M with the maximum resultant bracinq force at GKM II-M predicted by the Mk II load definition. Specifically, Subsection 9 6.2.1 describes the methcdoloqy f or applyinq the Mk II load to theoretically determine the maximum resultant bracing force. Subsection 9.6. 2. 2 presents the measured bracing force data and the maximum resultant bracing force for each break size measured at GKM II-M.

Finally, Subsection 9.6. 2.3 compares the theoretically determined bracinq force with the measured bracing force at GKM II-M.

9.6.2.1 Theoretical Determination of the Bracing Force at GKM II-M 9.6,2.1,1- Finite Element

. - Model Figure 9-247 is a schematic view of the finite element model The downcomer, inner .tank, and outer tank were modeled using beam type elements. The individual views of these components are shown in Fiqure 9-248.

The fluid elements are shown in Figure 9-249. There were 13 layers of fluid elements in the model. The solid elements used to represent the fluid had mid-side nodes which also matched up in elevation with the beam nodes of the structural model. A 90~

deqree sector of the fluid was used,, instead of the complete 360 deqree volume, in order to reduce the cost of the runs.

Appropriate structural properties were used to represent the 90 deqree sector model.

The brace, which is the component of primary interest in the structure, was modeled as a scalar spring element. The stiff bracinq connections between the downcomer and outer tank, and Rev- 9, 07/85 9P-96

PROPRIETARY between the inner and outer tank were modeled as "HEAR" rigid type elements within NASTHAN.

9,6 2,1.2- Model Ass~urn tions There were a number of important considerations that went into the development of the finite element model that was used in this analysis. They are:

(a) The dowrcomer is a long cylinder and at some distance beyond the bracinq level, the local effects of the bracing support die out. This allows for a variable spacing of nodal points alonq the down" orner, with a larger spacing away from the brace level.

(b) The outer tank is a tall cantilevered hearn. The downcomer is attached by very stiff bracing to the outer tank at its top. The inner tank is supported at two levels on the outez tank. It was deemed esse ntail to include the outer tank in the model, as the interact.ion between the downcomer and the outer tank could dramatically affect.the system response.

(c) Beam type elements were used to model the tanks and downcomez. This was deemed to be a ccnservative assumption in reqard to maqnitude of the bracinq force. The local shell effects at the bracinq connection points are essentially precluded by stiffening rings.

The General Electric Report (Reference 47), which developed the load function used herein, used similar assumptions in developinq a beam type element model of the 4T test facility.

(d) The brace can be represented as a scalar spring type element. This spring is attached between the downcomer and outer tank. The stiffer the spring the hiqher the expected reaction developed in the spring. The bracing stiffness used was 1400 K/in which represents the actual stiffness ia the GKN 'Il-H facility.

(e) The load under considaration is applied directly to the downcomer at its tip. The exact location and distribution of the load on the downcomer is not known and therefore using a beam type element for the downcomer, instead of t f j sh ell yp e ele me nts is ur ther ust ied.if The f act that the load is applied directly to the downcomer, and is not directly applied to the fluid means that the fluid effect can be handled with explicit fluid finite elements that condense into an added mass.

Rev. 9P-97

PROP RI ETAR Y The f ormulation used here corresponds to the equations shown in Appendix A of Reference 69. This theory is explained in more detail later.

9 6 2.l 3 Fluid Representation The fluid is accounted for as defined in References 68 and 69.

The fluid is defined as an acoustic medium which satisfies the equation:

~0 V P = P/C The combined structure plus fluid equations have the form:

K A J'A Q H 0 This set oX simultaneous equations can be solved by NASTRAN when some special procedures are invoked. For the incompressible fluid formulation, which was used in this model, the equations simpli fy to:

(M+ J'AH A

l AAT )U+ KU = f In this new equation the fluid effect has been reduced to an added mass term. Although the procedure is similar to an added lumped mass for the fluid, the theory and numerical values for the mass differ considerably from the simple lumped mass procedure.

9;6.2 1.4-- Structural Model In defininq the beam type elements that make up the structural model, care was taken to ensure a refined subdivision foz the structure. Thus the downcomer was divided into 34 elements along its lenqth. This permits accurate determination of the higher modes of the system. Due to the nature of the loading to be investiqated, namely the very short period of the load, the hiqher modes could possibly have a significant effect on the R ev. 9, 07/8 5 9P-98

PROPRIETARY response. A ref ined subdivision was also maintained on the inner and outer tanks. Shear effects were included in the analysis by specifying heaz coefficients for the beam type elements used to model the downcomer, and inner and outer tanks.

9.6. 2.1. 5 T.oading The loadinq applied to the model is d ef in ed as 2 possible case s.

was the Mk II Load. This load

l. A maximum msec.

amplitude of 30 ~ 000 pounds with a duration of 3

2. A maximum amplitude of 10,000 pounds with a duration of 6 msec.

The .loading is in the shape of a half-sine wave. Plots of the two loadinqs are shown in Fiquzes 4-'62GGH. These loadings were derived in Reference 47 as producing an upper bound on measured response of the downcomer.

Because of the fa't that a 90 degree sector or one f ourth of the structure was used in the model, one fourth of the total load was a p plied.

9 6. 2 1 6 A nalv sis- Results-The maximum resultant bracinq force developed in the bracing of the GKM II-M test facility, under application of the Mk II single vent lateral load definition {see Reference 47), is 22776 pounds.

The impulse with a maximum amplitude of 30,000 pounds and a duration of 3 msec gave the largest bracing force.

g.l- measurement of the Brac~in Forces Subsection 9,4.3.1..1 describes the instrumentation installed on the vent pipe bracing tsee Figure 9-S and 9-6), the eguation for convertinq the measured strains into'racing f orces and the procedure f or determining the resultant bracing forces.

Rev. 9, 07/85 9P-99

PROPRIETARY 9.6 2 2.2 Resultant Bracing Forces The resultang bracing force was determined as described in Subsection 9.4.3.l.l and is the equilibrium maintaining force which acts on the vent pipe horizontally at the height of the bracinq assemblage.

S] $ r~> ~ $)

where: Sl and S2 are the strut forces.

The resultant bracing forces as a function of steam mass flax and time into the blowdown for each MSL test are sho~n in Figures 9-250 to 9-258.

The resultant bracing forces for each test break size were then classified into the frequency distributions illustrated in Figures 9-259 to 9-261. This data is characterized by the followinq table:

Maximum value Mean Value kN kN Full MSL 'Breaks 70. 8 17 1/3 MSL Breaks 86.1 21 1/6 MSL Br ea ks 88 7 21.5 Thus, the maximum resultant bracing force measured at GKM II-M is 88.7 KN.

9.6. 2.3 Comparison of the Theoretica1 and Measured Maximum Resultant,BracincC Force According to Subsection 9 6.2.1.6 the maximum resultant bracing force usinq the Mk II load definition is 22776 pounds or 101.3 Re v. 9, 07/8 5 9P-100

PROP RIETAR Y kN. The maximum resultant bracing force measured at GKM II-H is 887 kN Therefore, the Hk II load definition (see Reference 47) conservati vely predicts the maximum measured resultant bracing force at GKN II-H.

9. 6. 3 Statistical Evaluation of the GKH II-M Resultant Bracing

-Porce Data ~ - ~

9,6 3,1 Introduct ion.

DAR Subsection 9.6. 2 compares the maximum calculated resultant bracing force at GKM II-N, using the Nark II single vent lateral load, with the maximum measured resultant bracing force at GKM ZI-H. This comparison reveals that the theoretical value bounds the measured value and indicates the conservatism of the Hark II single vent lateral load. However, the NRC performed a re-evaluation of the Hark II lateral tip load based on a statistical analysis of the original 4T bracing force data. They now conclude that the Mark XI impulse should be extrapolated to 65 Kips while preservinq the 3 msec impulse duration. This corresponds to a lateral load. which will be exceeded once in 10~

bracinq force events or once in ten LOCAs, chuqs at 100 vents per LOCA.

if one assumes 100 To provide additional confirmation of the conservatism of using an extrapolated Nark probability, a II lateral load at 10-5 exceedance statistical analysis of the GKM II-H bracing force data has been performed. This gave a relation for determining the re sul tan t bracin q force as a fu nction of th e exceedance probability. Prom this relation, the GKN II-N resultant bracing force required f or a 10-5 exceedance probability was then determined. The lateral load impulse which predicts this bracing force at GKM II-N was then determined and compared to. the revised Nark II tip impulse. The following subsections document this ef fort.

9.6.3.2 Derivation of a Probability Density Punction .from the

-. -. - -Measured- Resultant Bracin~Forces from the ~16 HSL Tests The mean and maximum resultant bracing f orce values for the 1/6 for the f ull and NSL tests enyelope the mean and maximum valves 1/3 HSL tests. Thus, to maximize the statistically determined resultant bracing force. the present'tatistical analysis is restricted to only the 1/6 MSL bracinq force data.

~6..3.p.l. Ge nera1 Consider ation To derive an expression for the probability density function, we assume the function follows an exponential decay. Thus, the Rev. 9, 07/85 9P-101

PROPRIETARY folloving probability density function is selected for the l/6 MSL resultant bracing force data:

f(u) = u - e-where: u = C ~ x C = constant to be determined x = resultant bracing force, kN Integratinq Eq. (1) and evaluating the interval yields:

CO f(u)du:= u e

-u du = e

-u du = e -u (-u-1) 0 0 Thus, this f unction satisfies the basic condition imposed on the probability density f unction; namely, that the total probability 1 %

To determine the constant, C, the mean value of the distribution, u, is defined as the first-order moment of the probability density function described by Fg. (1) . Thus

-u -u 2 u'-= u2 . e . du = e (-u -2u-2)

This is used to determine the constant, C, as:

2 C

and u f rom Fq. (1) as:

U

=2

= ~

X (2)

Z where: I is the mean value of the 1/6 MSL resultant bracing force data.

Rev. 9, 07/85 9P-'t 02

PHOPRIET AHY The exceedance probability based on the probability density f unc tion of Eq. (1) is:

P'u) = f (u)du = (1+u) e (3)

The probability that the resultant lies in the internal a < x < b is ~

P (a < u < b) = (a+1) e~ (b+1) e~ (4) 9,6,3. 2. 2- Application to the ~16 MSL Tests The ranqe o f mass fluxes to be used in evaluting the resulting bracinq forces from Tests 13 to 18 is:

ll <<

m A

33 (kg/m~s)

Tests 19 and 20 were omitted since they are bounded by the re maininq 1/6 M SL tes ts.

Table 9B shows the frequency distribution for Tests 13 to 18.

The numbers in parenthesis designate the number of occurrences per test.

TABLE 9B Bracinq Forces Number of Events kN 0-10 36 (6) 10 20 212 (35. 3) 20-30 237 (3 9. 5) 30 40 110 (18-3) 40 50 52 (8 6) 50 60 25 (4 16) 60 70 6 (1) 70 80 1 (1/6) 80 90 (1/6)

Listed in Table 9B are a total of 680 events, or 111.3 events per test R ev. 9, 07/8 5 9P-103

PROPRIETARY The me'an value of the frequency distri.bution is:

x = 25.6 kN Thus, Eq. (2) becomes u =

2 '

25.6

= 0.078'x {5)

Now Eq. (4) and (5) can be used to determine the interval probability. ' comparison o the theoretically determined interval probability (Eq. 4 and Hq. 5) with the relative frequencies obtained from Table 9B (test data) is shown in Table 9C.

TABLE 9C Bracing Forces Re.lative Frequency Interval kN (T es t Da ta) Pro babilit y 0- 10 0. 053 0-184 10 20 0.31 0. 2784 20 30 0. 35 0.216 30 40 0 16 0.14 40- 50 0 026 0-0826 50- 60 0. 037 0. 0465 60 70 8. 8 x 10-~ 0 025 70 80 1.5 x 10-~ 0-013 80 90 l. 5 x 10-> 6 8 x 10->

Figure 9-270 compares the theoretical with the measured rela tive f requencies. Thus, the function f (u) = ue- predicts a conserva ti ve distribution.

From Eq. (3) the exceedance probability is F'{u) = {1 + u) (6) wit h u = 0.078.x Thus, the bracing force x which is exceeded with a probability F can he determined.

Prom Eq (6) with an exceedance probability of 10-5 the resultant bracinq force x is:

182 kN = 40 9 Kips 6,3. 3.. 0eterminat ion od the Fxtra8olated Mark I~IIm ulee To determine the Mark- II impulse required to produce a bracing force of 182 kN at GKM II-M, we assume that the bracing force is Rev. 9, 07/85 9P-104

PROPRI ETHER Y linearly proportional to the impulse. This yields the following re la tion:

I (7)

F F m

where: Im impulse of present Nark II single vent lateral load definition bracing .force at GKN II-M produced by Im impulse required to produce bracing .force at GKM II-M correspondinq to a 10-~

exceedance probability F = statistical determined bracing force at GKH II-N for a 10-~ exceedance probability.

Subsection 9.6.2.1.6 calculates a maximum bracing force of 22.8 Kips with lateral load of 30,000 lbs and 3 msec impulse duration.

Thus, F = 22.8 Kips and the Hark II impulse, under tie half sine tion: wave impulse curve calculated I, is the area with the fol1owinq re la .

(8) where: F t

=

=

amplitude of Mark time duration of II impulse Hark II impulse Substitutinq F = 30,000 lbs. and t = 3 msec gives:

=

(2) (30,000) 57.3 0-sec ( 003) /

Substituting Fl = 182 kN = 40. 9 Kips, I = 57.3 0-sec and F 22.8 Kips into Bq. (7) qives:

57.3 Il 22 8 40.9 Il = 102 8 0sec To determine the extrapolated Nark IZ force required to give an impulse of 102.8 5-sec, we assume the impulse duration of 3 msec remains the same and solve for F in Eq. (8). Thus, for Il 102. 8 4-sec (102 8) (9 )

F =

(2) (-003)

Rev. 9~ 07/85 9P-105

PROPBIBTAH Y F = 54 8Kips Thus, an extrapolated half-sine wave impulse of 54 Kips with a 3 msec time duration produces a bracing force in GEM II-M corresponding to an exceedance probability of 10-~. This compares to the Mark II impluse maqnitude of 65 Kips.

For SSES, the Mark II impulse of 65 Kips with a 3 msec time duration, as required by the NRC, will be used for a single vent lateral load definition (see Subsection 4. 2.2. 3)

R ev. 9, 07/85 9P-106