ML20141L528

NEDO-21888, Rev 2, Mark I Containment Program Load Definition Report

ML20141L528
Person / Time
Issue date:	11/30/1981
From:	Ahsan Sallman NRC/NRR/DSS/SNSB
To:
Sallman,A., NRR/DSS, 301-415-2380
References
EPID L-2019-LLA-0197 NEDO-21888
Download: ML20141L528 (439)

Text

NEDO 21888 81NED282 Class I November 1981 Revision 2 MARK I CONTAINMENT PROGP~AM LOAD DEFINITION REPORT This document prepared by personnel of the Nuclear Power Systems Engineering Department Reviewed: 2~.. t12. iOZ~zLh Approved:

S. A. Hucik, Aanager G. E. Wa Mark III Containment Engineering Containment Design NUCLEAR ENERGY ENGINEERING DIVISION

• GENERAL ELECTRIC COMPANY SAN JOSE, CALIFORNIA 95125 GENERAL ~SELECTRIC

NEDO21888 DISCLAIMER OF RESPONSIBILITY Neither the General Electric Company nor any of the contributors to this document makes any warranty or representation (express or implied) with respect to the accuracy., completeness~ or usefulness of the infor~nation contained in this docwnent or that the use of such information may not infringe privately owned rights; nor do they assume any responsibility for liability or domage of any kind which may result from the use of the infornation contained in this document.

ii Revision 2

NEDO-21888 TABLE OF CONTENTS Page ABSTRACT iii LIST OF ACRONYMS xvii

1.0 INTRODUCTION

11 1.1 Description of the Mark I Containment 11 1.2 Mark I Containment Program 12 1.3 Load Definition Report Objective 15 2.0 REVIEW OF PHENOMENA 21 2.1 Design Basis Accident 21 2.2 Intermediate Break Accident 24 2.3 Small Break Accident 26 2.4 Safety/Relief Valve Actuation 27 2.5 Other Considerations 28 3.0 LOAD COMBINATIONS 31 4.0 LOCA RELATED LOADS 4.01 4.1 Containment System Temperature and Pressure Response 4.1.11 4.1.1 Design i~asis Accident 4.1.11 4.1.2 Intermediate Break Accident 4.1.21 4.1.3 Small Break Accident 4.1.31 4.2 Vent System Thrust Loads 4.21 4.2.1 Analytical Procedure 4.2.11 4.2.2 Assumptions 4.2.21 4.2.3 Analysis Results 4.2.3i 4.2.4 Application 4.2.41 4.3 Pool Swell Loads 4.31 4.3.1 Torus Net Vertical Load Histories 4.3.11 4.3.2 Torus Shell Pressure Histories 4.3.21 v Revision 2

NEDO21888 TABLE OF CONTENTS (Continued)

Page 4.3.3 Vent System Impact and Drag 4.3.31 4.3.4 Impact and Drag on Other Structures Above the Pool 4.3.41 4.3.5 Froth Impingement Loads 4.3.51 4.3.6 Pool Faliback Loads 4.3.61 4.3.7 LOCA Jet Load 4.3.71 4.3.8 LOCA Bubble Induced Drag Loads on Submerged Structures 4.3.81 4.3.9 Vent Header Deflector Loads 4.3. 91 4.4 Condensation Oscillation Loads 4.41 4.4.1 Torus Shell Loads 4.4.11 4.4.2 Loads on Submerged Structures Due to Main Vent Steam Condensation Oscillations 4.4.21 4.4.3 Downcomer Dynamic Load 4.4.31 4.4.4 Vent System Loads 4.4.41 4.5 Chugging Loads 4.51 4.5.1 Torus Shell Loads 4.5.13 4.5.2 Loads on Submerged Structures Due to Main Vent Chugging 4.5.21 4.5.3 Lateral Loads on Downcomers 4.5.31 4.5.4 Vent System Loads 4.5.41 5.0 SAFETY RELIEF VALVE DISCHARGE LOADS 5.01 5.1 Introduction 5.11 5.2 TQuencher Loads 5.2.11 5.2.1 S/RV Discharge Line Clearing Transient Loads 5.2.11 5.2.2 Torus Shell Pressure 5.2.21 5.2.3 S/RVDL Reflood Transient 5.2.31 5.2.4 TQuencher Water Jet Loads on Submerged Structures 5.2.41 5.2.5 TQuencher BubbleInduced Drag Loads on Submerged Structures 5.2.51 5.2.6 Thrust Loads on TQuencher Arms 5.2

• 61 5.2.7 Maximum S/RVDL and Discharge Device Pipe Wall Temperature 5.2.71 5.2.8 Fatigue Cycles 5.2.81 vi Revision 2

NEDO21888 TABLE OF CONTENTS (Continued)

Page 5.3 Ramshead Loads 5.31 5.3.1 S/RV Discharge Line Clearing Transient Loads 5.3.11 5.3.2 Ramshead Torus Shell Pressures 5.3.21 5.3.3 S/RVDL Ref lood Transient (Ramshead) 5.3.3I 5.3.4 Ramshead Water Jet Loads on Submerged Structures 5.3.41 5.3.5 Ramshead BubbleInduced Drag Load on Submerged Structures 5.3.51 5.3~6 Maximum S/RVDL and Discharge Device Wall Temperature 5.3.61 5.3.7 S/RVDL Fatigue 5.3.71 6.0 OTHER CONSIDERATIONS 6.01 6.1 Seismic Slosh Loads 6.2 Post Pool Swell Waves 6.21 6.3 Asymmetric Vent Performance 6.31 6.4 Downcomer Air Clearing Lateral Loads 6.41 6.5 Sonic Wave 6.51 6.6 Compressive Wave 6.61 6.7 Safety/Relief Valve Steam Condensation Loads 6.7.11 6.7.1 TQuencher Discharge 6.7.11 6.7.2 Ramshead Discharge 6.7.21 DISTRIBUTION 1 vii/viii Revision 2

NEDO21888 LIST OF ILLUSTRATIONS Figure Title Page 1.11 Typical Mark I Containment 18 1.12 Typical Composite Section Through Suppression Chamber 19 1.13 Typical Mark I Plant S/RV Discharge Line Configuration 110 1.14 Discharge Devices Employed in Mark I Plants 111 3.01 Loading Condition Combinations, Structures Affected:

Vent Header, Main Vents, Downcomers and Torus Shell, Accident Condition: DBA 36 3.02 Loading Condition Combinations, Structures Affected:

Vent Header, Main Vents, Downcomers, Torus Shell and Submerged Structures, Accident Condition: lEA 37 3.03 Loading Condition Combinations, Structures Affected:

Vent Header, Main Vents, Downcomers, Torus Shell and Submerged Structures, Accident Condition: SEA 38 3.04 Loading Condition Combinations, Structures Affected:

Small Structures Above Pool, Accident Condition: DBA 3.05 Loading Condition Combinations, Structures Affected:

Submerged Structures, Accident Condition: DBA 310 4.1.11 Typical Containment Temperature Response to Design Basis Accident (Recirculation Line Break) 4.1.111 4.1.12 Typical Containment Pressure Response to Design Basis Accident (Recirculation Line Break) 4.1.112 4.1.13 Drywell Pressure Response to Main Steam Line Break and Recirculation Line Break 4.1.113 4.1.21 Typical Containment Temperature Response to an Intermediate Break Accident 4.1.25 4.1.22 Typical Containment Pressure Response to an Intermediate Break Accident 4.1.26 4.1.31 Typical Containment Temperature Response to a Small Break Accident 4.1.35 4.1.32 Typical Containment Pressure Response to a Small Break Accident 4.1.36 4.21 Definition of Positive Thrust Loads 4.2.45 4.22 Single Main Vent Forces (05 sec) 4.2.46 4.23 Vent Header Forces Per Mitre Bend (05 sec) 4.2.47 4.24 Single Downcomer Forces (05 sec) 4.2.48 4.25 Total Vertical Forces, Net Vertical Force (05 sec) 4.2.49 4.26 Single Main Vent Forces (030 sec) 4.2.410 4.27 Vent Header Forces Per Mitre Bend (030 sec) 4.2.411 ix Revision 2

NEDO21888 LIST OF ILLUSTRATIONS (Continued)

Figure Title Page 4.28 Single Downcomer Forces (030 sec) 4.2.412 4.29 Total Vertical Forces, Net Vertical Force (030 sec) 4.2.413 4.210 Pressure Time Histories (05 sec) 4.2.414 4.211 Pressure Time Histories (030 sec) 4.2.415 4.212 Application of Thrust Force on Main Vent End Cap and Main Vent Mitre Bend 4.2.416 4.213 Application of Vent Header Forces 4.2.417 4.214 Application of Downcomer Forces 4.2.418 4.215 Resolution of Horizontal Forces 4.2.419 4.3.11 Typical Net Torus Vertical Loading History 4.3.19 4.3.12 Comparison of Typical Analytical and Test Drywell Pressure Histories 4. 3. 110 4.3.13 Enthalpy Flow Comparison 4.3.111 4.3.21 Typical Average Submerged Torus Pressure History 4.3.26 4.3.22 Typical Torus Airspace Pressure History 4.3.27 4.3.23 Submerged Location on Typical Torus Shell 4.3.28 4.3.31 Vent System Coordinates 4.3.38 4.3.32 Downcomer Impact and Drag Pressure Transient 4.3.39 4.3.33 Application of Impact/Drag Pressure Transient to Downc omer 4.3.310 4.3.34 Vent Header Local Impact Pressure Transient 4.3.311 4.3.3S Schematic Diagram Illustrating the Methodology for Main Vent Impact and Drag 4.3. 312 4.3.41 Typical Mark I Torus Sector 4.3.412 4.3.42 Typical Pool Surface Displacement Longitudinal Distribution 4.3.413 4.3.43 Typical Plant Pool Surface Displacement in XY Plane 4.3.414 4.3.44 Typical Pool Surface Velocity Longitudinal Distribution 4.3.415 4.3.45 Typical Plant Velocity Transient in XY Plane 4.3.416 4.3.46 Calculation of Distance Over Which Impulse Acts 4.3.417 4.3.47 Impulsive Impact Loading Region for Structures with a Natural Frequency Less than 30 Hz 4.3.418 4.3.48 Effect of Drag Following a Parabolic Impact Pulse 4.3.418A 4.3.49 Pulse Shape for Water Impact on Cylindrical Targets 4.3.418B 4.3.410 Drag Coefficient for Cylinders Following Impact 4.3.418C 4.3.411 Pulse Shape for Water Impact on Flat Targets 4.3.418D x Revision 2

NEDO21888 LIST OF ILLUSTRATIONS (Continued)

Figure Title Page 4.3.412 4.3.51 Pressure Drop Due to Flow Across Gratings Definition of Froth Impingement Region I

4. 3.418E 4.3.57 I

4.3.52 Definition of Froth Impingement Region II 4.3.58 4.3.53 Froth Loading History Region I 4.3.59 4.3.54 Froth Loading History Region II 4.3.510 4.3.55 Froth Impingement Region II Possible Directions of Load Application 4.3. 511 4.3.56 Possible Directions of Froth Failback Load Application 4.3. 512 4.3.61 Positive Directions of Fallback Load Application 4.3. 63 4.3.81 Torus and Typical Submerged Structure Geometry 4.3. 85 4.3.82 Sample History of Total XForces on Vent Header Support Column 4.3.86 4.3.83 Sample Time History of Total ZForces on Vent Header Support Column 4.3.87 4.3.91 Typical Vent Header Deflector 4.3.96 4.3..92 Typical Deflector Designs 4.3.97 4.3.93 Definition of Space Variables 4.3.98 4.3.94 Variable Definition 4.3. 99 4.3.95 Static Pressure Immersion Depth 4.3.99 4.3.96 Impact and Steady Drag Force Correlation for Type I Deflector 4.3. 910 4.3.97 (CD) for Type 1 Deflector 4.3.911 4.3.98 Impact and Steady Drag Force Correlation for Type 2 Deflector 4.3.912 4.3.99 Impact and Steady Drag Force Correlation for Type 3 Deflector 4.3.913 4.3.910 Impact and Steady Drag Force Correlation for Type 4 Deflector 4.3. 914 4.3.911 Longitudinal Multiplier for Fluid Displacement Velocity and Acceleration 4.3.915 4.3.912 Impact Force Transient for Addition to the Empirical Data for Deflector Types 1 3 4.3.916

4. 3. 913 Calculated Load Histories 4.3. 917 4.3.914 QSTF Measured Load with Assumed Analytical Spike 4.3. 917 4.3.915 Time Correspondence Between QSTF Calculated Force and Calculated Forces at Other ~/2.Stations 4.3.918 xi Revision 2

NEDO21888 LIST OF ILLUSTRATIONS (Continued)

Figure Title Page 4.3.916 Typical Vent Header Deflector Load 4. 3. 919 4.4.11 Condensation Oscillation Baseline Rigid Wall Pressure Amplitude on Torus Shell Bottom Dead Center 4.4.110 4.4.12 Mark I Condensation Oscillation Torus Vertical Cross Sectional Distribution for Pressure Oscillation Amplitude 4. 4. 111 4.4.13 Mark I Condensation Oscillation Multiplication Factor Versus PooltoVent Area Ratio for Plant Unique Load Determination 4.4.112 4.4.21 Sample Predicted Time History of Total XForce on S/RV Line 4.4.25 4.4.22 Sample Predicted Time History of Total Z~-Forces on S/RV Line 4.4.26 4.4.31 Downcomer Dynamic Load 4.4.37 4.4.32 Downcomer Pair Internal Pressure Loading for DBA CO 4.4.38 4.4.33 Downcomer Dynamic Differential Pressure Loading for DBA CO 4.4.39 4.4.34 Downcomer CO Dynamic Load Application 4.4.310 4.4.35 Downcomer Internal Pressure Loading for lEA CO 4.4.311 4.5.11 A Typical Chug Average Pressure Trace on the Torus Shell 4.5.113 4.5.12 Mark I ChuggingTorus Vertical Cross Sectional Distribution for Pressure Amplitude 4.5.114 4.5.13 Mark I Chugging Torus Asymmetric Circumferential Distribution for Pressure Amplitude 4.5.115 4.5.14 PostChug Rigid Wall Pressure Amplitudes on Torus Shell Bottom Dead Center 4.5.116 4.5.21 Sample Predicted Time History of Total XForces on S/RV Line 4.5.23 4.5.22 Sample Predicted Time History of Total ZForces on S/RV Line 4.5.24 4.5.31 Deleted 4.5.39 4.5.32 Distribution of Chugging RSEL Reversals for a Typical 4.5.310 Sector 4.5.33 Notation Used for Transforming RSEL Reversals into Stress 4.5.311 Reversals at a Fatigue Evaluation Location A 4.5.41 Chugging Waveform for Grass Vent System Pressure Oscillation Load 4.5.46 xii Revision 2

NEDO21888 LIST OF ILLUSTRATIONS (Continued)

Figure Title Page 5.11 TQuencher Load Definition Scheme 5.13 5.12 Ramshead Load Definition Scheme 5.14 5.2.11 Sample Prediction of S/RVDL Internal Pressure Transient 5.2.17 5.2.12 Sample Prediction of TQuencher Internal Pressure Transient 5.2.18 5.2.13 Sample Prediction of Thrust Loading on an S/RV Pipe Segment Initially Filled with Gi~ 5.2.19 5.2.14 Sample Prediction of ihrust on S/RV Pipe Run Between the Discharge Device and the First Upstream Elbow (Pipe Run Initially Filled with Water) 5.2.110 5.2.15 Sense of Thrust Loading 5.2.111 5.2.16 Sample Prediction of Mass Flow Rate of Water Exiting TQuencher 5.2.112 5.2.17 Sample Prediction of Water Mass Acceleration 5.2.113 5.2.21 Sample Prediction of Torus Shell Pressure Loading Transient 5.2. 25 5.2.22 Sample Prediction of Torus Shell Longitudinal Pressure bistribution 5.2.26 5.2.23 Sample Prediction of Torus Shell Radial Pressure Distribution at Section AA in Figure 5.2.22 5.2.27 5.2.31 Sample Prediction of S/RVDL Reflood Transient 5.2.35 5.2.41 Phases of Quencher of Jet Formation and Decay 5.2.45 5.2.42 Jet Sections Along the Quencher Arm on Torus Plan View 5.2.46 5.2.51 Sample Predicted Time History of Total XForces on Downcomer 5.2.55 5.2.52 Sample Predicted Time History of Total 2Forces on Downcomer 5.2.56 5.2.61 Thrust Loadz on Arm Endcaps 5.2.67 5.2.62 GeneralizedShape of Thrust Loading Transient and Application Point With Time 5.2.68 5.2.71 Example of Predicted S/RVDL and Discharge Device Temperature Distribution 5.2. 73 5.3.11 Sample Prediction of the S/RVDL Internal Pressure Transient 5.3.15 5.3.12 Sample Prediction of Thrust Load Transient at a Typical S/RVDL Segment 5.3.16 5.3.13 Sense of Thrust Loading 5.3.17 xiii Revision 2

NEDO21888 LIST OF ILLUSTRATIONS (Continued)

Figure Title Page 5.3.14 Sample Prediction of Water Clearing Thrust Load on the Last Pipe Segment of the S/RVDL 5.3.18 5.3.15 Sample Prediction of Mass Flow Rate of Water Exiting Ramshead 5.3.19 5.3.16 Sample Prediction of Water Mass Acceleration During

- S/RV Discharge 5.3.110 5.3.21 Normalized Shell Pressure Transient for any Location on the Pool Wall Due to an S/RV Discharge for a Cold Pipe (Monticello Test Data) 5.3.27 5.3.22 Normalized Positive Pressure Distribution at the Torus CrossSection Coincident with the Ranshead ~j. Ramshead Located on Torus Centerline 5.3.28 5.3.23 Sample Prediction of Torus Shell Longitudinal Pressure Distribution Resulting from an S/RV Discharge Through a Ramshead Discharge Device 5.3.29 5.3.24 Procedural Flowchart 5.3.210 5.3.25 Normalized Shell Pressure Transient for any Location on the Pool Shell Due to an 51kV Discharge Leaky Valve (Monticello Test Data) 5.3.211 5.3.26 Normalized Shell Pressure Transient for any Location on the Pool Shell Due to a Subsequent Valve Actuation NWL (Monticello Test Data) 5.3.212 5.3.41 Sample Prediction of a Typical Structure Loading History 5.3.45 5.3.51 Sample Predicted Time History of Total XForces on Downcomer and Vent Header Support Column 5.3.55 5.3.52 Sample Predicted Time History of Total ZForces on Downcomer and Vent Header Support Column 5.3.56 6.11 Application of Seismic Slosh Drag Load 6.14 6.41 Sectors Used to Define Directions of Lateral Loads on Downcomers End 6.43 xiv Revision 2

NEDO21888 LIST OF TABLES Table Title Page 1.21 Mark I Utilities and Plants in the Mark I Owners Group 17 3.0-1 S/RV Discharge Load Cases for Mark I Structural Analysis 33 3.02 Event Timing Nomenclature 34 3.0-3 Structural Loading 4.1.11 Fuel Decay Heat and Sensible Energy for All Mark I Plants 4.1.17 4.1.12 Plant Vent System Loss Coefficients 4.1.1B 4.1.13 Typical Plant Conditions at Instant of DBA Pipe Break 4.1.19 4.1.14 Summary of Mark I Plant Break Areas 4.1.110 4.21 Nomenclature for Section 4.2 4.2.43 4.3.11 Mean Peak Net Torus Vertical Loads Operating Drywell!

Wetwell Pressure Differential (Al) 4.3.17 4.3.12 Single Test Peak Net Torus Vertical Peak Loads Zero Drywell/Wetwell Pressure Differential 4.3.18 4.3.21 Torus Shell Pressure History Multipliers Mz and Me 4.3.25 4.3.31 Internal Vent Header/Torus Airspace Pressure Differential at Time of Vent Header Impact 4.3.37 4.3.41 Hydrodynamic Mass and Acceleration Drag Volumes for TwoDimensional Structural Components 4~3*4...9 4.4.11 Condensation Oscillation Onset and Duration 4.4.17 4.4.12 Condensation Oscillation Baseline Rigid Wall Pressure Amplitudes on Torus Shell Bottom Dead Center 4.4.18 4.5.11 Chugging Onset and Durations 4.5.111 4.5.12 PostChug Rigid Wall Pressure Amplitudes on Torus Shell Bottom Dead Center 4.5.112 4.5.4I Vent System Load Amplitudes and Frequencies for Chugging 4.5.45 6.11 Mark I Containment Seismic Slosh Results 6.13 xv/xvi Revision 2

NEDO2 1888 ABSTRACT This document provides the met hodo logy and definition of the thermal-hydraulic loads produced on the pressure suppression containment system of GE Mark I containments during a postulated loss-of-coolant accident, safety/relief valve discharges, and related dyncm7ic events. Information and guidance has been pro-vided to assist in evaluating the design conditions for the various structures of the contait~nent system. This docwnent has been prepared for use by the Mark I Owners in performing plant unique structural evaluations.

iii/iv Revision 2

NEDO21888 LIST OF ACRONYMS A/E Architect/Engineer ABS Absolute Summation Method ADS Automatic Depressurization System BWR Boiling Water Reactor DBA Design Basis Accident DLF Dynamic Load Factor ECCS Emergency Core Cooling Systems EFRI Electric Power Research Institute FSAR Final Safety Analysis Report PSI Fluid Structure Interaction FSTF Full Scale Test Facility FWCI Feedwater Coolant Injection HEM Homogeneous Equilibrium Model HPCI High Pressure Coolant Injection lEA Intermediate Break Accident LDR Load Definition Report LOCA Loss of Coolant Accident LPCI Low Pressure Coolant Injection LTP Long Term Program MS IV Main Steam Isolation Valve MSLB Main Steam Line Break MWT Megawatt Thermal xvii Revision 2

NEDO21888 LIST OF ACRONYMS (Continued)

NOC Normal Operating Conditions NRC Nuclear Regulatory Commission NS SS Nuclear Steam Supply System NWL Normal Water Level PSD Power Spectral Density PULD Plant Unique Load Definitions QSTF Quarter Scale Test Facility RC IC Reactor Core Isolation Cooling RHR Residual Heat Removal RLB Recirculation Line Break RPV Reactor Pressure Vessel RSEL Resultant Static Equivalent Load SBA Small Break Accident S/RV Safety/Relief Valve S/RVDL Safety/Relief Valve Discharge Line STP Short Term Program SRSS Square Root Sum of the Squares xviii Revision 2

NEDO21888 SECTION 1 INTRODUCTION Revision 2

1.0 INTRODUCTION

In June of 1976 at the request of the Mark I Owners Group, General Electric Company initiated a program to provide the domestic Mark I utilities with detailed containment load definitions for specific hydrodynamic events appli-cable to their plants. More precisely, Mark I containment pressure suppression component load definitions were to be provided for the postulated Lossof Coolant Accident (LOCA) and Safety/Relief Valve (S/RV) actuation events. This report represents the culmination of these activities and provides the hydro dynamic load definition procedures to be utilized by each utility in a plant unique structural evaluation. Complete load definitions for the components discussed herein are obtained through consideration of these loads in conjunc-tion with the seismic, deadweight and buoyancy effects.

The loads and methodologies given in Sections 4, 5 and 6 of this report are defined in sufficient detail that a plant unique analysis can be performed for each Mark I containment. The loads are presented on .a generic basis except where geometric or system differences require that the loads be defined on a plant unique basis. In cases where it has been more efficient to do so, the method to calculate a plant unique load, rather than the load itself, is presented. For each loading condition, the analytical and/or experimental bases for the stated definition are also presented. A complete list of acronyms utilized in this report is given in the preface.

1.1 DESCRIPTION

OF THE MARK I CONTAINMENT The Mark I containment is a pressure suppression system which houses the Boiling Water Reactor (BWR) pressure vessel, the reactor coolant recirculating loops and other branch connections of the Nuclear Steam Supply System (NSSS).

It consists of a drywell, a pressure suppression chamber (or wetwell) which is approximately half filled with water, and a vent system which connects the drywell to the wetwell suppression pool. The suppression chamber is toroidal in shape and is located below and encircling the drywell. The drywellto wetwell vents are connected to a vent header contained within the airspace of the wetwell. Downcomer pipes project downward from the vent header and termi-nate below the water surface of the suppression pool. The pressure suppression 11 Revision 2

chamber is shown in relation to the steel drywell in Figure 1.11. Figure 1.12 shows a typical cross section of the suppression chamber.

In the highly unlikely event of a high energy NSSS piping failure within the drywell,.reactor water and/or steam would be released into the drywell atmos-phere. This postulated event is referred to as a Loss of Coolant Accident (LOCA). As a result of increasing drywell pressure, a mixture of drywell atmosphere, steam, and water would be forced through the vent system into the pool of water maintained in the suppression chamber. The steam vapor would condense in the suppression pool, thereby limiting internal containment pressure. The noncondensible drywell atmosphere would be transferred to the suppression chamber and contained therein. Section 2 of this report presents a complete description of these events.

BWRs utilize safety/relief valves (S/RV) attached to the main steam line as a means of primary system overpressure protection. The outlets of these valves are connected to discharge pipes which are routed to the suppression poo1 as shown in Figures 1.12 and 3. The discharge lines end in either Ramshead or TQuencher type discharge devices. Figure 1.14 presents the configuration of each type of discharge device.

Following the actuation of an S/RV, steam enters the safety/relief valve discharge line (S/RvDL), pressurizing the line and forcing the air and water initially in the line to be expelled to the wetwell and contained therein.

Subsequent to water and air clearing of the S/RVDL, the steam released by the S/RV actuation flows through the S/RVDL and discharge device into the suppression pool and is condensed therein. Section 2 of this report presents a complete description of these events.

1.2 MARK I CONTAINMENT PROGRAM For the Mark I containment design, the pressure and temperature loads associated with a LOCA, were based on experimental technology obtained from testing of the 12 Revision 2

NEDO-21888 Bodega Bay and Humboldt Bay Power Plant pressure suppression concepts.

The tests were performed to demonstrate the viability of the pressure sup-pression concept for reactor containment design by simulating the LOCA with various equivalent piping break sizes up to twice the crosssectional break area of the largest reactor system pipe. The test data provided quantitative information for establishing containment design pressures.

In performing large scale testing of an advanced design pressuresuppression containment (Mark III), and during inplant testing of Mark I containments, suppression pool hydrodynamic loads not explicitly included in the original Mark I containment design basis were identified. These additional loads could result from dynamic effects of drywell air and steam being rapidly forced into the suppression pool during a postulated LOCA, and from suppres-sion poo1 response to S/RV operation generally associated with plant transient operating conditions. Since these hydrodynamic loads were not explicitly considered in the original design of the Mark I containment, the NRC staff in early 1975 requested a detailed reevaluation of the containment system from each domestic utility with a Mark I containment.

Recognizing the joint need to respond to the NRC requests for additional information and the essential similarity of all the Mark I plants, the domestic Mark I utilities formed an Owners Group on April 23, 1975. The Owners Group provided a strong, unified, and consistent approach to resolution of the open issues. through the pooling of individual resources. The memher utilities of the Mark I Owners Group and their respective plants are listed in Table 1.21.

The Mark I Owners Group retained the General Electric Company to develop and manage a program which would address and resolve the stated concerns.

A two-phase program was established; it was described to the NRC in letters submitted during the week of May 5, 1975. The Phase I effort, called the Short Term Program (STP), provided a rapid confirmation of the adequacy of 13 Revision 2

NEDO21888 the containment to maintain its integrity under the most probable course of the postulated LOCA considering the latest available information on the important suppression pool dynamic loads. The first phase demonstrated the acceptability of continued operation during the performance of Phase II, called the Long Term Program (LTP). The LTP included detailed testing and analytical work to define precisely the specific hydrodynamic loads for which each containment would be assessed to establish conformance to the original intended design safety margins.

The STP was completed in late 1976 following the docketed submittal hy each utility of the documentation listed in References 1.21 through 1.211 and the associated plant unique analysis of the external torus support system and the externally attached piping. Review of this documentation by the NRC led to issuance of the Mark I Containment Short Term Program Safety Evaluation Report in December 1977 (Reference 1.212). This report concluded that licensed domestic BWR Mark I facilities could continue to operate safely, without undue risk to health and safety of the public, during an interim period while the Long Term Program was conducted.

The Long Term Program activities were initiated in June 1976. A detailed description of this program and plans for its implementation are give in References 1.013 and 1.014. These references define the objectives of the Program, provide Program task descriptions, and describe the integrated activities leading to a definition of hydrodynam~ic loads for reevaluation of the containment structure by the individual utilities. In all, seven Program work areas were established:

1.0 Program Action Plan 2.0 Preliminary Load Evaluation Activities 3.0 Structural Acceptance Criteria 4.0 Generic Structural Evaluation 5.0 Load Evaluation 6.0 Load Mitigation Development Testing 7.0 Load Definition Report The extensive testing and analytical efforts performed under the eighteen separate tasks of Area 5.0 and the pooi swell and S/RVDL discharge mitigation tasks of Area 6.0 are the primary bases for the load definitions of Area 7.0.

These load definitions are presented in this report.

Revision 2 14

NEDO-2 1888 1.3 LOAD DEFINITION REPORT OBJECTIVE The objective of this Load Definition Report (LDR) is to provide load definition procedures for the postulated LOCA and S/RV actuation events for use in structural reevaluation of the pressure suppression chamber, vent system, S/RV discharge lines and other Mark I containment components.

15/16 Revision 2

NEDO21888 Table 1.21 MARK I UTILITIES AND PLANTS IN THE MARX I OWNERS GROUP Utility Name Plant Name Boston Edison Company Pilgrim Boston, Massachusetts Carolina Power & Light Company Brunswick 1,2 Raleigh, North Carolina Commonwealth Edison Company Dresden 2,3 Chicago, Illinois Quad Cities, 1,2 Detroit Edison Company Fermi 2 Detroit, Michigan Georgia Power Company Batch 1,2 Atlanta, Georgia Iowa Electric Light & Power Company Duane Arnold Cedar Rapids, Iowa Jersey Central Power & Light Company Oyster Creek Morristown, New Jersey Nebraska Public Power District Cooper Columbus, Nebraska Niagara Mohawk Power Corporation Nine Mile Point 1 Syracuse, New York Northeast Utilities Service Company Millstone Berlin, Connecticut Northern States Power Company Monticello Minneapolis, Minnesota Philadelphia Electric Company Peach Bottom 2,3 Philadelphia, Pennsylvania Power Authority of the State of New York Fitzpatrick New York, New York Public Service Electric and Gas Hope Creek 1,2 Newark, New Jersey Tennessee Valley Authority Browns Ferry 1,2,3 Knoxville, Tennessee Yankee Atomic Electric Company Vermont Yankee Westboro, Massachusetts 17 Revision 2

DOWNCOMERS TORUS SUPPRESSION CI4AMBEI~

PRESSURE SUPPRESSION POOL Figure 1.11. Typical Mark I Containment 18 Revision 2

NEDO21888 VENT HBADER MAIN VENT DOWNCOMER 5AFETY~RELIEF VALVE DISCHARGE PIPE SIRV DISCHARGE PIPE DEVICE Figure 1.12. Typical Composite Section Through Suppression Chamber Revision 2 19

NEDO21888 REACTOR VESSEL.

DRYWELL SAFETYIRELIEF VALVE MAIN STEAM LINE SiRVDL VACUUM BREAKER*

~iRV DISCHARGE LINE POOL WATER LEVEL

~ TORUS Figure 1.13. Typical Mark I Plant S/RV Discharge Line Configuration 110 Revision 2

NEDO21888 RAMSHEAD DISCHARGE DEVICE T.OUENCHER END CAP HOLES DISCHARGE (OPTIONAL) DEVICE HOLES SYMMETRICAL ABOUT HORIZONTALE~

(BOTH SIDE)

Figure 1.14. Discharge Devices Employed in Mark I Plants 111/112 Revision 2

NEDO21888 REFERENCES FOR SECTION 1 1.21 J. M. Humphrey, et al., Mark I Containment Evaluation Short Term Program Final Report, Volume I, Program Description and Summary of Conclusions, General Electric Company, Report No. NEDO20989, September 1975.

1.22 R. H. Buchholz, et al., Mark I Containment Evaluation Short Term Program Final Report, Volume II, LOCA Related Hydrodynamic Loads, General Electric Compauy, Report No. NEDC209892, September 1975.

1.23 J. M. Humphrey, et al., Mark I Containment Evaluation Short Term Program Final Report, Volume III, Load Application and Screening of Structural Elements, Report No. NEDC209893, September 1975.

1.24 Mark I Containment Evaluation Short Term Program Final Report, Volume IV, Structural Evaluation, Prepared by the Bechtel Power Corporation for the General Electric Company, Report No.

NEDC209894A, September 1975.

1.25 Mark I Containment Evaluation Short Term Program Final Report, Volume V, Independent Assessment of the Mark I Short Term ProgramI Prepared by Teledyne Materials Research for the General Electric Company, Report No. NEDC209895, September 1975.

1.26 Mark I Containment Evaluation Short Term Program Final Report, Addendum 1 to Volume IV, Structural Evaluation, Prepared by the Bechtel Power Corporation for the General Electric Company, Report No. NEDC209894A, September 1975.

1.27 Mark I Containment Evaluation Short Term Program Final Report, Addendum 2, Loads and Their Application for Torus Support System Evaluation, General Electric Company, Report No. NEDC20989, Addendum 2, June 1976.

1.28 5. Kayhan, et al., Mark I Containment Evaluation Short Term Program Final Report, Addendum 3, Vent Header and Vent Pipe Impact Loads General Electric Company, Report No. NEDC20989, Addendum 3, August 1976.

1.29 Mark I Containment Evaluation Short Term Program Final Report, Addendum 4, Revisions to Volume IV, Structural Evaluation, Prepared by the Bechtel Power Corporation for the General Electric Company, Report No. NEDC209894A, November 1975.

1.210 Description of Short Term Program Plant Unique Torus Support Systems and Attached Piping Analysis, NUTECH, Report No.

~l020l2, Revision 2, June 1976.

113 Revision 2

NEDO21888 1.211 Letter from L. J. Sobon (GE) to V. Stello (NRC), SubjectMark I Short Term Program Report Questions, October 29, 1976.

1.212 Mark I Containment Short Term Program Safety Evaluation Report, NUREG0408, Nuclear Regulatory Comission, December 1977.

1.213 Letter from L. J. Sobon (GE) to V. Stello (NRC), SubjectMark I Containment Program Program Action Plan (Revision 3), February 1978.

1.214 Mark I Containment Program Program Action Plan, Revision 3, dated February 15, 1978, General Electric Company, Nuclear Energy Division, February 1978.

114 Revision 2

SECTION 2 REVIEW OF PHENOMENA Revision 2

2.0 REVIEW OF PHE~1OMENA This section describes the sequence of events of the Mark I containment related phenomena for the postulated LOCA and S/RV actuation conditions. A basis is provided for understanding the loading conditions which result from these events.

For a postulated pipe break inside the drywell, three LOCA categories are con-sidered. These three categories, selected on the basis of postulated break size, are referred to as the Design Basis Accident (DBA), Intermediate Break Accident (IBA) and Small Break Accident (SBA). S/RV actuation can occur as a result of a number of system conditions. Although the load magnitudes depend on the initial system conditions prior to S/RV discharge, the sequence of the loading phenomena is the same for all conditions.

2.1 DESIGN BASIS ACCIDENT The DBA for Mark I plants employing jet pumps within the BWR is the instantaneous doubleended guillotine break of the BWR recirculation pump suction line at the reactor vessel nozzle safe-end to pipe weld. For Mark I plants that do not employ jet pumps within the BWR, the same DBA break definition applies except that the break location is the BWR recirculation pump discharge line at the reactor vessel nozzle safe-end to pipe weld. As discussed in Section 4.1.1, these postulated break conditions result in the maximum flow rate of primary system fluid and energy into the drywell, through the vent system and into the wetwell. The DBA results in the maximum pressurization rate and peak pressure in the drywell; it therefore produces the most limiting pool swell and vent system thrust loads. The DBA event is evaluated up to the time that the low pressure Emergency Core Cooling System (ECCS) starts to flood the reactor vessel, which occurs at approximately 30 seconds after the pipe break as pre-sented in Section 4.1.1.3. The use of this break to determine the DBA loading for all Mark I plants is verified by a direct comparison of the drywell and wetwell pressure responses resulting from a recirculation line break with that from a main steam line break.

21 Revision 2

The sequence of events within the wetwell which follow the postulated break, is divided into two phases:

a. Pool Swell This phase covers the dynamic effects of drywell and vent system air being forced through the vent system into the suppression pool to the wetwell airspace.

b. Steom Condensation This phase covers the dynamic events during the period following initial air clearing when the flow into the suppression poo1 is a steam air mixture. The steam is con-densed at the downcomer exit while the air rises through the pool to the wetwell airspace.

The reactor will automatically scram due to high drywell pressure. Main steam line isolation will occur due to low reactor water level. No mechanical S/RV actuation will occur because of the rapid reactor vessel depressurization and large rate of reactor fluid and energy inventory loss through the break. It is assumed that spurious actuation of a single S/RV can occur at any time during the DL With the postulated instantaneous rupture of a recirculation line, a pressure wave traveling at sonic velocity would expand from the break location into the drywell atmosphere and through the vent system. The wave amplitude attenuates rapidly as it expands into the larger drywell volume. The wave front enters the vent system with nearly uniform amplitude, but is greatly attenuated from its initial value at the break location.

The rapid bulk pressurization of the drywell immediately following a postulated DBA and prior to vent clearing theoretically generates a weak compressive wave in the downcomer water legs. This wave could propagate through the suppression pool and induce a muchattenuated loading on the torus shell.

Immediately following the postulated DBA pipe rupture, the pressure and temperature of the drywell atmosphere and vent system increase rapidly. These combined pressure and temperature transients induce mechanical and thermal loadings on the vents, vent header and downcomers. With the drywell pressure increase, the water initially standing in the downcomers accelerates into the 22 Revision 2

NEDO21888 pool until the downcomers clear of water. During this water clearing process, the suppression pool fluid is accelerated causing drag loads on structures within the suppression pool located immediately below the dowocomers. Following downcomer water clearing, the downcomer air, which is at essentially drywell pressure, is exposed to the relatively low pressure of the wetwell, producing a downward reaction force on the torus. The consequent bubble expansion causes the pool water to swell in the torus, compressing the airspace above the pool.

During the early stages of this process, the pool swells in the bulk mode (i.e., the water is accelerated upward by the rising air bubble motion) and structures close to the pool surface experience impact and drag loads as the pool impacts and flows past the structure. Following the initial air bubble expansion and pool surf ace rise, the bubble pressure decreases as the bubble overexpands and the pool liquid mass decelerates. The net effect of the pool deceleration is an upward lifting force on the torus. Eventually, the bubbles break through to the torus airspace and an air/water froth mixture continues upward due to the momentum previously imparted to the water slug. The upward motion of this froth mixture causes impingement loads on the torus and other structures encountered, but the loads are lower in magnitude than the impact loads associated with bulk pool swell.

As the air/water mixture in the suppression pool experiences gravityinduced phase separation, pool upward movement stops and the liquid falls hack. Struc-tures within the path of the fluid motion experience a downward loading; the submerged portion of the torus experiences a small pressure increase. Follow-ing pool fallback, there are waves on the suppression pool surface which induce low magnitude loads on the downcomers, torus and ~anyother structures close to the water surface.

The transient associated with drywell air venting to the pool typically lasts for 3 to 5 seconds. Since the air originally contained within the drywell and vent system is transferred to the wetwell airspace, the wetwell will experience a rise in static pressure. Following air carryover, there will be a period of high steam flow through the vent system. The discharge of steam into the pool and its subsequent condensation causes pool pressure oscillations which will be transmitted to submerged structures and the torus shell. This pheno menon is referred to as condensation oscillation. As the reactor vessel 23 Revision 2

.depressurizes, the steam flow rate to the vent system decreases. The reduced steam flow rate leads to a reduction in the drywell/wetwell pressure differen-tial. Steam condensation during this period of reduced steam flow is charac-terized by movement of the water/steam interface up and down within the downcnmer as the steam volumes are condensed and replaced by surrounding pool water. This phenomenon is referred to as chugging. During the steam condensation period the downcomers experience a lateral loading due to the asymmetric collapse of steam bubbles at the downcomer exit. Also, the submerged structures and containment walls experience pressure oscillations due to steam bubble forma-tion and collapse.

Shortly after the postulated pipe rupture, the Emergency Core Cooling System (ECCS) automatically begins to pump water from the plant condensate storage tank and/or the suppression pool into the reactor pressure vessel (RPV) to flood the reactor core. Eventually water cascades into the drywell from the break, causing steam condensation and drywell depressurization. As the drywell pressure falls below the pressure in the wetwell .airspace, the drywell vacuum relief system is actuated and air from the wetwell enters the drywell, equaliz-ing the drywell and wetwell pressures slightly above their initial values.

Following vessel flooding and drywell/wetwell airspace pressure equalization, suppression pool water is continually recirculated from the pool to the reactor vessel by the ECCS pumps. The core decay power results in a slow heatup of the suppression pool. The suppression pool cooling mode of the Residual Heat Removal (RHR) system is manually actuated to remove energy from the suppression pool to return the containment to normal temperature conditions.

2.2 INTER~DIATE BREAK ACCIDENT The IBA for a BWR Mark I plant is a postulated pipe rupture small enough that rapid reactor depressurization will not occur, but large enough that the High Pressure Coolant Injection (HPCI) system cannot maintain reactor vessel water level. The IBA is defined as a liquid line break of 0.1 ft2.

24 Revision 2

• The high drywell pressure resulting from the postulated accident conditions will scram the reactor. The sequence of events following this scram could eventually lead to closure of the main steamline isolation valves (MSIV) due to low reactor water level. The closure of the MSTVs would result in an increase in REV pressure that is relieved by opening the~S/RVs.

Following the postulated break, steam fills the drywell causing the drywell pressure slowly to increase and displace the water initially in the submerged portion of the vent system into the suppression pool. The drywell pressure transient is sufficiently slow that the dynamic effect of the water clearing in the vents is negligible. The subsequent clearing of air from the vent system occurs more slowly than for the DBA and thus imparts lower loadings to the wet well components. As the flow of air, steam and water continues from the drywell to the wetwell, the wetwell airspace pressure increases. Following the initial purge of air from the drywell, steam begins to flow through the vent system and condenses within the suppression pool. As with the DBA event, the conden-sation oscillation and chugging phenomena occur during the steam condensation process.

~The Automatic Depressurization System (ADS) will actuate due to high drywell pressure and low reactor water level at approximately 300 seconds for a plant with turbine driven feedwater pumps. For plants with motor driven feedwater pumps it is assumed that the feedwater system continues to maintain the vessel level until the pumps are manually tripped at 600 seconds. The ADS will initiate approximately 300 seconds after the motor driven pumps are tripped, or 900 seconds after the break occurs. The reactor will be depressurized approximately 200 seconds after the ADS is initiated. Thus, the lEA is evaluated to 500 seconds for plants with turbine driven feedwater pumps and 1100 seconds for plants with motor driven feedwater pumps. During operation of the ADS, steam from the RPV is vented directly to the suppression pool via the S/RVDL. As the reactor depressurizes, the Core Spray systems and the Low Pressure Coolant Injection (LPCI) mode of the RHR system will be activated to flood the REV and cool the core. Eventually, water will cascade into the drywell causing steam condensation and drywell depressurization. As the drywell depressurizes below 25 Revision 2

NEDO21888 the pressure within the wetwell airspace, the wetwell to drywell vacuum breakers open, equalizing the containment pressures and terminating the event. Since the reactor depressurization transient is less rapid for the IBA than for the DBA more decay heat is discharged into the suppression poo1 which results in the suppression poo1 temperature being higher for the IBA than for the DBA at the time of complete reactor depressurization.

2.3 SMALL BREAK ACCIDENT The SBA for a BWR Mark I plant is a postulated pipe rupture in which the fluid loss rate from the reactor system is insufficient to depressurize the reactor and small enough that HPCI operation is sufficient to maintain reactor water 2

level. The SBA is defined as a 0.01 ft steam break. Following the break, the drywell pressure will slowly increase until the high drywell pressure scram set point is reached. Following the reactor scram, MSIV closure may occur due to the water level transient in the reactor. If the main steam lines isolate the reactor system, pressure will increase and safety/relief valves will open intermittently to control system pressure.

The drywell pressure increase gradually depresses the water level in the vents until the water is expelled and air and steam enter the suppression pool. The rate of airflow will be such that the air will bubble through the pool without causing pool swell. The steam will be condensed and drywell air will pass to the wetwell airspace. A gradual pressurization of the wetwell results at a rate dependent upon the break size and degree of drywell steam/air mixing.

Eventually, the steam/air flow through the vents results in essentially all the drywell air being transferred into the wetwell. Following the air transfer, wetwell pressurization occurs at a rate dependent on the suppression pool heatup rate. Condensation oscillation will not be present due to the low mass flux, but chugging may occur.

Operator actions will determine the subsequent course of events. It is assumed that there will be no operator action for 10 minutes following the pipe break.

After 10 minutes, it is assumed that the operator will rapidly depressurize the reactor via ADS. An individual plant may choose to conduct a plant unique analysis to demonstrate that a less rapid cooldown rate is acceptable.

26 Revision 2

NEDO-21888 2.4 SAFETY/RELIEF VALVE ACTUATION Prior to the initial actuation of a safety/relief valve caused by a normal operational transient, the S/RV discharge lines contain air at atmospheric pressure and suppression pool water in the submerged portion of the piping.

Following S/RV actuation, steam enters the S/RVDL compressing the air within the line and expelling the water slug into the suppression pool. As the water is cleared, the S/RVDL undergoes a transient pressure loading. The submerged portion of the line also experiences water clearing thrust loads caused by directional changes of momentum within the line and discharge device. The water jets entering the pool from the discharge device result in drag loads on submerged structures within the influence of the jets.

Once the water has cleared from the discharge device, the compressed air enters the pool in the form of high pressure bubbles. These bubbles expand resulting in an outward acceleration of the surrounding pool water. The momentum of the accelerated water results in an overexpansion of the bubbles, causing the bubble pressure to become negative relative to the ambient pressure of the surrounding pool. This negative pressure slows and reverses the motion of the water, leading to a compression of the bubbles and a positive pressure relative to that of the pool. The bubbles continue to oscillate in this manner for approximately one second as they rise to the pool surface. As the bubbles oscillate, the associated local pool motion causes drag loads on nearby sub-merged structures. The positive and negative pressures developed within the bubbles attenuate with distance and result in an oscillatory pressure loading on the submerged portion of the torus shell.

Following air and water clearing from the line, steam is discharged through the line to the suppression pool and condensed therein. The steam condensation phenomenon produces a highfrequency, lowmagnitude oscillatory loading on the torus shell.

FoJ.lowing closure of an S/RV, the steam pressure in the S/RV decreases rapidly as the steam flows out into the pressure suppression pool. At a sufficiently low steam pressure, pool water reenters the S/RVDL. A rapid depressuri zation of the line then occurs as the steam remaining in the line is condensed 27 Revision 2

NEDO21888 by the inflowing water. This depressurization causes the water to reflood into the S/RVDL and the vacuum breaker valve on the S/RVDL to open, allowing drywell gas to enter the line. The reflooding water may rise in the line to a level somewhat above its initial preS/RV actuation level before equilibrium is reestablished. The actual reflood water level depends primarily on the size of the S/RVDL vacuum breaker.

2.5 OTHER CONSIDERATIONS In addition to the LOCA and S/RV discharge events, the dynamic behavior of the pool water during and after seismic motion is evaluated. Horizontal and vertical seismic motion is transmitted to the pool through the torus support system. The seismic motion generates a sloshing wave behavior on the pool surface which produces minimal loads on the downcomers, submerged structures and the torus shell.

28 Revision 2

SECTION 3 LOAD COMBINATIONS Revision 2

3.0 LOAD COMBINATIONS This section identifies the timing sequence of the loading conditions for the structural components in the wetwell due to the hydrodynamic phenomena described in Section 2. In addition, those sections of the report which define the hydro dynamic structural loading are identified.

The combinations of loading conditions affecting the various major structural components are presented in Figures 3.01 through 3.05. These figures identify the hydrodynamic loading conditions resulting from LOCA and from S/RV discharges.

Seismic loadings, and structural and water deadweight loads are not presented in this report. The figures in this section provide the timing sequences of the loading conditions for all three postulated LOCAs, the DBA, IBA, and SBA.

The lengths of the bars in the figures indicate the time periods during which a loading condition may exist. A loading condition such as condensation oscillation is assumed to exist continually during the indicated time period. For S/RV dis-charge lcads, the duration of the loading is short, but the loads may occur at any time during the indicatedtime period. Loads are considered to act simultaneously on a structure at a specific time if the loading condition bars overlap at that time.

The bar charts for the DBA show the loading condition combinations for postulated breaks in the recirculation line, the main steam line or any other breaks large enough to produce significant pool swell loads. The bar chart for the IBA shows conditions for a break size sufficiently large such that the HPCI system cannot prevent ADS actuation on lowwater level, but smaller than that break size which would produce significant pool swell loads. A break size of 0.1 ft2 is assumed for the lEA. The bar charts for the SBA show conditions for a break size equal to 2

0.01 ft in crosssectional flow area. For an SBA, the HPCI system would be able to maintain the water level and the reactor would be depressurized by means of operator initiation of the ADS.

Table 3.01 identifies the S/RV discharge loading conditions.

31 Revision 2

NEDO21888 Table 3.02 identifies the timing nomenclature used in the bar charts.

Table 3.03 identifies the S/RV and LOCA loads which potentially affect struc-tural components and identifies the appropriate section of the LDR defining the loads. For S/RV piping and other structures within the wetwell, the location of the structural component must be considered to determine if any of the identi-fied conditions affect the structure.

32 Revision 2

Table 3.01 S/RV DISCHARGE LOAD CASES FOR MARX I STRUCTURAL ANALYSIS Any 1 ADS Mul t iple*

Valve Valves Valves Initial Conditions 1 2 3 A. First Actuation Al A2 AS B. First Actuation Leaking S/RV** B3 C. Subsequent Actuation C3

• Number (one or more) and location of S/RVs assumed to actuatewill be determined by plant unique analyses.

• • Only one S/RV is assumed to leak for plants equipped with Ramshead dis-charge devices for normal operating transients. The loads for TQuencher discharge devices are not affected by leaking S/RVs. No S/RVts are con-sidered to leak prior to a LOCA.

33 Revision 2

Table 3.02 EVENT TIMING NOMENCLATURE Time Description The onset of condensation oscillation (see Section 4.4)

The beginning of chugging (see Section 4.5)

The end of chugging (see Section 4.5)

Time of complete reactor depressurization (see Section 4.1)

ADS actuation on high drywell pressure and low reactor water level, break size and plant depend-ent. The ADS may also be actuated by the operator for a small break accident.

34 Revision 2

Table 3.03 STRUCTURAL LOADING Other Wetvell Interior Structures s~.

5, ..

s 25 5 O.J 0

• U U a -J U, ~.

5 . 3

~ 0E 0 C

S. 3 20 I 0 S. W O~ 0

0. 5~ 5 C U ,

• 0. .J 0 I~ S.. 3 .

.0 U, U~

0 0 5 ~43U 2 5

0 0.

43. ~

00 Ii S e U S sm 3 0 0 0 C > 30 > 0 S.. U. C 3 ~ 00 Loads < <5 ~

4.1 Containment Pr.ssure and Thmperatur.. X X X X X X X X x 4.2 Vent System Thrust Loads X X X 4.3 Pool Swell 4.3.1 Torus Net Vertical Loads X X 4.3.2 Torus Shell Pressure Histories X X 4.3.3 Vent System Impact sod Drag X X I 4.3.4 Impact and Drag on Other Structures I X I 4.3.5 Froth Impingement I I X I X 4.3.6 Pool Fallback X I X 4.3.7 LOCA Jet X I 4.3.8 LOCA Bubble Drag X X I 4.4 Condensation Oscillation 4.4.1 Torus Shell Loads I K 4.4.2 Load on Submerged Structures X I X 4.4.3 Lateral Loads on Doqmcomera I I 4.4.4 Vent System Loads I X 4.5 Chugging 4.5.1 Torus Shell Loads I X 4.5.2 Loads on Submerged Structure. I X I 4.5.3 Lateral Loads oo Downoomere X I 4.5.4 Vent System Loads C I 5.2 TQuencher Loads 5.2.1 Discharge Line Clearing x 5.2.2 Torus Shell Pressures X I 5.2.6 Jet Loads on Submerged Structures I X I I 5.2.5 Air Bubble Drag x x x K 5.2.6 Thrust Loads on TQuencher Arms I 5.2.7 S/RVDI. invironneotal Temperature x 5.3 Reushead Loads 5.3.1 Discharge Line Clearing x 5.3.2 Torus Shell Praeeuree I I 5.3.4 Jet Loads on Submerged Structure. I I X x 5.3.5 Air Bubble Drag xl I I I 5.3.6 S/RVDL Environmental Temperature x 35 Revision 2

VENT SYSTEM AIR, STEAM AND LIQUID FLOW AND PRESSURE TRANSIENTS SECTION 4.2 a S 5 a SINGLE S/RV ACTUATION 4S/RV EVENT CASE Al)

SECTIONS 5.2 OR 5.3 0 a 2

0 C.)

0 POOL UWILL

-J I SECTION 4.3 I CONDENSATION OSCILLATION SECTION 4.4 I

NOTE: CONSIDERATION SHOULD BE GIVEN TO REACTION LOADS FROM ATTACHED STRUCTURES.

I I I CHUGGING

-.0.1 82 83 TIME AFTER LOCA (s.d Figure 3.01. Loading Condition Combinations Structures Affected: Vent Header, Main Vents, Downcomers and Torus Shell Accident Condition: DBA Revision 2 36

NEDO2 1888 LOCA PRESSURE AND TEMPERATURE TRANSIENTS SECTION 4.1 4

3 SINGLE S/RV ACTUATION*

a (S~RV EVENT SECTIONS 5.2CASE ORAl) 6.3 z

0 I-.

0 z

0 U

S/RV ACTUATION ON z SETPOINT (S/RV EVENT CASE A3. C31 ADS ACTUATION 0

-J I (SIRV EVENT CASE A2) I I CONDENSATION OSCILLATION SECTION 4.4 I

CHUGGING SECTION 4.5 NOTE: CONSIDERATIONS SHOULD BE GIVEN TO REACTION LOADS FROM ATTACHED STRUCTURES.

LOADING DOES NOT COMBINE WITH OTHER S/RV CASES I I I I ti IADS>BO 82 83 4 TIME AFTER LOCA (med Figure 3.02. Loading Condition Combinations Structures Affected: Vent Header, Main Vents, Downcomers, Torus Shell and Submerged Structures Accident Condition: IBA 37 Revision 2

NEDO21888 N

LOCA PRESSURE AND TEMPERATURE TRANSIENTS SECTION 4.1 5

3.

SINGLE S/RV ACTUATION I.- ISJRV SECTIONS EVENT 6.2CASE OR 6.3 Al)

OPERATOR INITIATION OF ADS C,

2 EVENT CASE A2) 0 0

-J IRV ACTUATION ON SET-POINT II/~V EVENT CASE A3, C31 CHUGG ING I SECTION 4.5I NOTE: CONSIDERATION SHOULD BE GIVEN TO REACTION LOADS PROM ATTACHED STRUCTURES.

LOADING DOES NOT COMBINE WITH OTHER S/RV CASES.

I I t2 600. 83 14 TIME AFTER LOCA (eec)

Figure 3.03. Loading Condition Combinations Structures Affected: Vent Header, Main Vents, Downcomers, Torus Shell and Submerged Structures.

Accident Condition: SBA 38 Revision 2

NEDO21888 LOCA PRESSURE AND TEMPERATURE TRANSIENTS SECTION 4.1 0

z 0

C) 0 FROTH IMPINGEMENTI SECTION 4.3.5 z

0

-J

~~~~~~~1 I POOL SWELL FALLBACK SECTION 4.3.6 FWEL~MPA~

AND DRAG SECTION 4.3.4 77 I NOTE: 1. STRUCTURES BELOW MAXIMUM POOL SWELL HEIGHT I I

-0.1 0.7 ~1.5 TIME AFTER LOCA (sec)

Figure 3.04. Loading Condition Combinations Structures Affected: Small Structures Above P001*

Accident Condition: DBA For S/RV lines consider possibility of loads produced by S/RV actuation Case Al.

Revision 2 39

NEDO21888 LOCA PRESSURE AND TEMPERATURE TRANSIENTS SECTION 4.1 a

5 2

S a e SINGLE S/RV ACTUATION ES/RV EVENT CASE Al)

SECTIONS 5.2 OR 5.3 a a a en I I

S z CONDENSATION OSCILLATION SECTION 4.4 S

hPOOLUWELLI PALLDACK I CHUGGING SECTION 4.5 L~.2~2ZeJ D LOCA MR SUIELE SECTION 4.3.8 D LOCA WATER JET PORMATION SECTION 4.3.7 NOTE: CONSIDERATION SHOULD BE GIVEN TO REACTION LOADS FROM ATTACHED STRUCTURES.

LOADING DOES NOT COMBINE WITH OTHER S/RV CASES.

I I

-0.1 -.0.7 -.1.5 ~1 83 TIME AFTER LOCA Esec)

Figure 3.05. Loading Condition Combinations Structures Affected: Submerged Structures

• Accident Condition: DBA 310 Revision 2

SECTION 4 LOCA RELATED LOADS Revision 2

4.0 LOCA RELATED LOADS This section specifies the methodology used to determine the wetwell and vent system primary structural loads encountered during a postulated LOCA.

The drywell pressure and temperature responses which are calculated for a postulated DRA are also presented for a typical Mark I plant.

The primary structural loads generated during a LOCA are a result of the following:

a. The pressures and temperatures within the drywell, vent system and wetwell torus

b. The fluid flow through the vent system

c. The initial LOCA bubble formation in the pool and the resulting displace-ment of water due to pool swell

d. Steam flow into the suppression pool (condensation oscillation and/or chugging)

e. S/RV actuation loads (discussed in Section 5).

The spectrum of break sizes addressed consists of the DBA, IBA and SBA as pre-viously defined in Sections 2.1, 2.2 and 2.3 of this report, respectively.

4

• 01 / 4.02 Revision 2

NEDO21888 4.1 CONTAINMENT SYSTEM TEMPERATURE AND PRESSURE RESPONSE The following sections present the drywell and wetwell pressure and temperature responses due to the DBA, IBA and SBA.

4.1.1 Design Basis Accident A typical set of containment temperature and pressure responses to a DBA are pre-sented in Figures 4.1.11 and 4.1.12, respectively. Following the pipe break, the drywell pressure increases and the water initially in the submerged portion of the vent system accelerates into the suppression pool. Until the water is com-pletely cleared from the vent system, the air pressure in the wetwell does not begin to change significantly. Starting near point A (Figure 4.1.12) both drywell and wetwell pressures increase as the vents clear. The REV blow down continues into the drywell and flow continues from the drywell through the vent system and into the wetwell. Flow to the wetwell is initially air, but sub-sequently becomes a mixture of air, water and steam between points B and D. Point B is characterized by the depletion of the fluid inventory initially occupying the recirculation loop. Point C occurs at the depletion of the sub cooled fluid inventory of the recirculation loop and the REV, and at the transi-tion to saturated fluid blowdown from the RPV.

The drywell pressure increases more rapidly than the wetwell pressure until the mass flow rate into the drywell drops sharply. At approximately point D, the liquid level inside the reactor has dropped below the elevation of the pipe break and the blowdown, which was previously all liquid, becomes a steamliquid mixture. The flow through the vent system now begins to decrease and results in a corresponding decrease in the vent system pressure. This permits the drywell and wetwell pressures to converge until they differ only by the hydrostatic pres-sure determined by the vent system submergence at point E. This description of the timing of events applies to both temperature and pressure transients in the drywell and wetwell.

4.1.11 Revision 2

NEDO21888 4.1.1.1 Analytical Procedure The details of the analytical models used to simulate the shortterm transient response of the drywell and wetwell to a DBA are presented in References 4.1.11 and 4.1.12. These models briefly are as follows:

a. A vessel blowdown model which simulates the reactor vessel response to a LOCA and determines the break mass and energy flow rate. (Described in Reference 4.1.11)

b. A drywell model which determines drywell thermodynamic conditions as a result of the mass and energy flows into and out of the drywell. (Des-cribed in Reference 4.1.11)

c. A model which determines vent clearing time and flow behavior during the LOCA. (Described in Reference 4.1.12)

d. A model which simulates suppression poo1 temperature response by a simple mass and energy balance. (Described in Reference 4.1.11)

e. A werwell free airspace model which is used to calculate the airspace pressure and temperature response. (Described in Reference 4.1.11)

The models described have been verified by comparison with model tests performed at the Bodega Bay and Humboldt Bay type containment test facilities as described in References 4.1.12 and 4.1.13.

4.1.1.2 Assumptions

a. MayWitt fuel decay heat and sensible energy as defined in Table 4.1.11 for all plants. This assumption is consistent with the original Mark I plants licensing bases documented in plant FSARs or PSARs.

b. The blowdown model is the Homogeneous Equilibrium Model (HEM) which is fully described in Reference 4.1.14. This blowdown model replaces the slip flow model described in Reference 4.1.11.

4.1.12 Revision 2

NEDO21888

c. Steam condensation on drywell structures and internal components is conservatively neglected.

d. Plant unique calculated vent system loss coefficients, which include the downcomer exit loss, as identified in Table 4.1.12, are used. The method of calculating these loss coefficients has been verified against the 1/12 scale 3D tests described in Reference 4.1.15.

e. The flow of liquid, steam and air in the vent system is assumed to be a homogeneous mixture based on the instantaneous mass fractions in the drywell. This assumption yields increased vent system flow density resulting in higher flow losses and therefore higher drywell pressure.

These conservative drywell pressures were used as the forcing function for the plant unique pool swell tests (see Section 4.3.1).

f. The DBA for plants employing jet pumps within the BWR is the instan-taneous doubleended guillotine break of the recirculation pump, suction line at the reactor vessel nozzle safeend to pipe weld. For plants that do not employ jet pumps within the BWR, the DBA break location is the BWR recirculation pump discharge line at the reactor vessel nozzle safeend to pipe weld. The fluid inventory initially occupying the recirculation loop includes the subcooled fluid in the recircula tion system and the reactor vessel region external to the shroud. This maximizes the initial mass and energy release rates to the drywell.

g. The bulk wetwell pool and airspace temperatures are assumed equal throughout the transient. This assumption maximizes the wetwell air-space temperatures and pressures.

h. The initial drywell temperature is 1350F at the time of the break. This is the normal operating temperature of the drywell as supplied by the utilities.

i. The initial drywell relative humidity is 20% and the initial wetwell relative humidity 100%.

• The 20% drywell humidity represents the mini-mum value of the normal range and maximizes the mass of noncondensibles in the drywell. The 100% wetwell humidity maximizes the vapor pressure and thus results in the maximum total wetwell pressure.

4. 1.13 Revision 2

NEDO21888

j. The reactor vessel control volume is assumed to include the fluid and structural masses and energy, of the following primary system components, thereby conservatively predicting the blowdown mass and energy and the resulting drywell and wetwell pressures and temperatures:

(1) Reactor vessel (2) Recirculation loops (3) Main steam lines to the inboard isolation valves (4) RCIC steam line to the first normally closed valve (5) RHR shutdown line to the first normally closed valve (6) LPCI line to the first normally closed valve (7) Core spray line to the first normally closed valve (8) HPCI line to the first normally closed valve (9) Feedwater line to the point in the feedwater system where the temperature during normal operation is equal to the saturation temperature at the final calculated reactor vessel pressure.

Feedwater mass below this temperature will not flash during reactor vessel depressurization and therefore will not discharge to the containment.

(10) Isolation condensors for applicable plants.

k. The reactor is operating at 102% of licensed thermal power. This value for a typical plant is identified in Table 4.1.13. This assumption maximizes the core decay heat.

1. The initial temperature of the suppression pool is assumed to be equal to the arithmetic mean of the operating range supplied by the Mark I owners. A typical plant suppression pool temperature is shown in Table 4.1.13.

4.1.14 Revision 2

NEDO21888

m. The highest torus water level within the normal operating range is assumed, thereby providing the maximum downcomer submergence and mini-mum free airspace. See Table 4.1.13 for typical plant conditions.

n. The area of the instantaneous guillotine break is as tabulated in Table 4.1.1.4.

o. No credit is taken for normal auxiliary power. This maximizes the time it takes the ECCS systems to become operable.

p. The single failure of one RHR pool cooling loop is assumed. This has no impact on the short term containment pressure and temperature response.

q. For plants with a drywelltowetwell pressure differential sys-tem, the initial drywell and wetwell pressures are selected based on the AP system operating at the nominal value of the normal range specified. The nominal value was set equal to the arithmetic mean of the maximum and minimum values supplied by the Mark I Owners.

For plants without a drywelltowetwell pressure differential sys-tem, the initial drywell and wetwell pressures are conservatively assumed to be 0.75 psig. Typical plant conditions are shown on Table 4.1.13.

4.1.1.3 Analysis Results Plant unique containment pressure and temperature responses have been generated for each plant. These plant unique results are presented in Plant Unique Load Definition documents and submitted individually to the Mark I owners. Typical responses are shown in Figures 4.1.11 and 4.1.12, for a plant with condi tions.as tabulated in Table 4.1.13.

4.1.15 Revision 2

NEDO21888 4.1.1.4 Determination of Limiting Break Size and Location In order to verify that the recirculation line break will result in the most severe drywell pressure rate and peak pressure loading for all Mark Is, the system response during a postulated main steam line was compared with the response during a recirculation line break. The analytical bases and the conclusions are presented in this section.

The containment system response to a recirculation line break is described in Section 4.1.1. The main steam line break is similar except that the initial blowdown is steam instead of liquid. Drywell pressurization and flow through the vent system into the wetwell is similar to the recirculation line break.

The criteria used to determine the limiting event for containment structure load-ing were the drywell rate of pressurization and the peak drywell pressure. For each plant, the maximum break areas were determined for both main steam and recirculation lines and are listed in Table 4.1.14.

During the early portion of the blowdown where the peak drywell pressurization rate and peak drywell pressure are occurring, the plant with the largest ratio of main steam line break area to recircularion line break area was selected as the test case. For other plants, the difference in loading due to a recircula tion line break from that due to a main steam line break will be larger than the difference observed in the test case plant.

Analyses were performed for both a postulated main steam line break and a recircu lation line break using the analytical procedures described in Section 4.1.1. The break area and characteristics of the blowdown were adjusted to represent the appropriate break location. The resulting drywell and wetwell pressure time his-tories from the two analyses are presented, superimposed, in Figure 4.1.13. The recirculation line break analysis shows both a higher rate of pressurization and a higher peak pressure for the drywell and wetwell and therefore would impose a higher overall containment loading. Because the resulting pressure loading deter mined for the test case plant indicates that the recirculation line break is the limitin~g case, it is concluded that for all Mark Is, the recirculation line break generates the most severe DBA containment loading.

4.1.16 Revision 2

NEDO2 1888 Table 4.1.11 FUEL DECAY HEAT AND SENSIBLE ENERGY FOR ALL MARK I PLANTS Normalized Normalized Decay Heat* Decay Heat*

Time and Sensible T-isi~ and Sensible (sac) Energy (sec) Energy l 1.000 4 2 1 x lOi 1.005 l 1 x 104 1.201 x 102 2 x 101 9.657 x 101 2 x 104 1.008 x 103 4 x 101 8.251 x 101 4 x 10~ 8.125 x 103 6 x 101 7.093 x 10 6 x 10 7.394 x 103 8 x 10 6.503 x 8 x l0~ 6.646 x 10 l 3 1 5.310 x lO~ 1 x l0~ 6.245 x 103 2 4.848 x 101 2 x 1055 5.126 x 103 4 5.459 x 101 4 x 105 4.096 x 103 6 5.663 x 101 6 x 105 3.596 x 103 8 5.378 x 10 8 x 10 3.196 x 10 1

1 x 101 4.807 x 101 1 x 106 2.985 x 3

2 x 101 2.051 x 102 2 x 106 2.367 x 103 4 x 10~ 5.501 x 102 x 1066 1.826 x 103 6 x 101 4.220 x 102 6 x 106 1.573 x 103 8 x 101 3.952 x 10 8 x 10 1.430 x 10 7

1 x 102 3.811 x 102 1 x 107 1.287 x l0~

2 x 102 3.365 x 102 2 2 x 10 9. 790 ~ 4 4 x 102 2.827 x 102 4 x l0~ 7.480 x 104 6 x 102 2.549 x 102 6 x 10~ 6.820 x 10 8 x 102 2.365 x 10 8 x l0~ 6.270 x l0~

2 8 4 1 x l0~ 2.229 x 102 1 x 108 6.050 x 104 2 x 1.841 x 102 2 x 10 5.335 x 10 4 x 1.512 x 102 6 x l0~ 1.353 x 102 8 x l0~ 1.257 x 10

• May Witt decay heat normalized to initial power 4.1.17 Revision 2

Table 4.1.12 PLANT VENT SYSTEM LOSS COEFFICIENTS Plant Vent System Loss Coefficients Browns Ferry 1, 2, 3 5.32 Brunswick 1, 2 5.17 Cooper Station 5.51 Dresden 2, 3 5.17 Duane Arnold 4.65 Enrico Fermi 2 5.51 Fitzpatrick 5.17 Hatch 1, 2 5.51 Hope Creek 1, 2 5.51 Millstone 5.17 Monticello 5.17 Nine Mile 1 5.6 Oyster Creek 1 5.6 Peach Bottom 2, 3 5.17 Pilgrim 5.17 Quad Cities 1, 2 5.17 Vermont Yankee 5.17 4.1.18 Revision 2

NEDO2 1888 Table 4.1.13 TYPICAL PLANT CONDITIONS AT INSTANT OF DBA PIPE BREAK 102% Licensed Power (14WT) 1703 Initial Suppression Pool Temperature (F) 77.5 Downcomer Submergence (ft) 4.25 3 Drywell 134200 Airspace Volume (ft )

Wetwell 97580 Drywell 1.25 Airspace Pressure (psig)

Wetwell 0.10 4.1.19 Revision 2

NEDO21888 Table 4.1.14

SUMMARY

OF MARK I PLANT BREAK AREAS Break Area (ft 2 ) Area Ratio Main Steam!

Plant Main Steam Line Recirculation Line Recirculation Browns Ferry 1, 2, 3 3.934 4.216 0.933 Brunswick 1, 2 3.185 4.258 0.748 Cooper Station 3.. 055 4.158 0.735 Dresden 2, 3 2.402 4.261 0.564 Duane Arnold 2.197 2.515 0.874 Enrico Fermi 2 4.091 4.119 0. 993*

Fitzpatrick 3.083 4.172 0.739 Hatch 1/2 3.076 4.141/3.988 0.743/0.771 Hope Creek 1, 2 3.960 4.177 0.948 Millstone 2.233 4.363 0.512 Monticello 1.803 4.015 0.449 Nine Mile 1 3.887 7.058 0.551 Oyster Creek 1 3.308 6.056 0.546 Peach Bottom 2, 3 3.934 4.147 0.949 Pilgrim 2.102 4.359 0.482 Quad Cities 1, 2 2.402 4.261 0.564 Vermont Yankee 1.801 4.145 0.434

• Test case plant.

4.1.110 Revision 2

NEDO21888 300 F

VENT CLEARING (A) r INVENTORY DEPLETION FSUBCOOLEP FLUID WI DEPLETION IC) r BREAK UNcOVERY ID) 0 (El PRESSURE EOUALIZATION w 1 I- 150 w

0.

uJ WETWELL TEMPERATURE 0

0 10 20 30 TIME AFTER BREAK INITIATION usc)

Figure 4.1.11. Typical Containment Temperature Response to Design Basis Accident (Recirculation Line Break) 4.1. 111 Revision 2

NEDO21888 60 40 I

aw 20 0

0 10 20 30 TIME AFTER BREAK INITIATION (usc)

Figure 4.1.12. Typical Containment Pressure Response to Design Basis Accident (Recirculation Line Break) 4.1. 112 Revision 2

NEDO2 1888 60 w

w A-10 TIME AFTER BREAK INITIATION (ascI Figure 4.1.13. Drywell Pressure Response to Main Steam Line Break and Recirculation Line Break

4. 1. 113/4. 1. 114 Revision 2

NEDO21888 REFERENCES FOR SECTION 4.1.1 4.1.11 W. J. Bilanin, The GE Mark III Pressure Suppression Containment System Analytical Model, General Electric Company, Report No. NEDO20533, June 1974.

4.1.12 General Electric Pressure Suppression Containment Analytical Model, General Electric Company, Report No. NEDO10320, April 1971; Supple-ment 1, May 1971; Supplement 2, June 1973.

4.1.13 Bodega Bay Atomic Park Unit Number 1, Exhibit C Preliminary Hazards Summary Report, Appendix I Pressure Suppression Test Program, Pacific Gas and Electric Company, Docket No. 50205 December 28, 1962.

4.1.14 F. J. Moody, Maximum Discharge Rate of LiquidVapor Mixtures from Vessels, General Electric Company, Report No. NEDO21052, September 1975.

4.1.15 Electric Power Research Institute, ThreeDimensional Pool Swell Modeling of a Mark I Suppression System, EPRI NP906, October 1978.

4.1.115 Revision 2

NEDO21888 4.1.2 Intermediate Break Accident Typical containment temperature and pressure responses to an lEA are presented in Figures 4.1.21 and 4.1.22, respectively. Following the break, the dry well air pressure increases and slowly forces the water initially in the sub-merged portion of the vent system into the suppression pool. The drywell pressure transient is sufficiently slow that the dynamic effect of the water in the vents is negligible and the vents clear when the drywelltowetwell differential pressure is equal to the submergence pressure at point A in Figure 4.1.22. As the flow of air, steam and water continues from the drywell to wetwell, the drywell pressure increases at approximately the same rate as that for the wetwell airspace. After all the drywell air has been purged into the wetwell at point B, the pressure rise for the entire containment slows. The subsequent initiation of the automatic depressurization system at point C, vents primary system fluid into the suppression pool causing its temperature to rise rapidly. The wetwell airspace temperature and pressure then increase. As the ECCS fluid is injected into the reactor, the temperature of the fluid discharged into the drywell will drop thereby condensing the steam in the drywell which reduces the drywell pressure. The wetwell to drywell vacuum breakers open and the wetwell pressure is reduced. The containment response is determined for the lEA up to the time when the reactor vessel pressure equalizes with the drywell pressure, which occurs at approximately 50 psia at point D.

During the depressurization of the reactor vessel, the suppression pool temp-erature increases rapidly due to the transfer of stored reactor vessel energy to the pool. The suppression pool temperature transient will level out after the reactor vessel is depressurized. The pool temperature will rise much less rapidly since the heat input to the pool is limited to the core decay heat, and the RHR system is removing heat from the pool. No S/RV, condensation oscillation, or chugging loads occur after complete reactor depressurization.

4.1.21 Revision 2

NEDO21888 4.1.2.1 Analytical Procedure The pressure and temperature response of the containment to the postulated lEA was determined by performing a mass/energy balance at the following times:

a. When the vents are half cleared

b. When the vents are fully cleared

c. When half the drywell air has been purged to the wetwell

d. When all the drywell air has been purged to the wetwell

e. When the feedwater system is manually shut off at 10 minutes after the accident (for plants with motor driven feedwater pumps only). For plants with turbine driven feedwater pumps, the pumps are assumed to coast down in 7 seconds following reactor isolation; the pressures and temperatures for this point in time are not calculated.

f. Just prior to ADS initiation due to low reactor water level and high drywell pressure

g. When the reactor vessel is depressurized to equalize with the drywell pressure (approximately 50 psia) 4.1.2.2 Assumptions

a. The fuel decay heat and sensible energy used for the DBA analysis shown in Table 4.1.11 are used for the ThA analysis.

A liquid break is assumed with an area equal to 0.1 ft2. The blow b.

down model is the same as the Homogeneous Equilibrium model used in the DBA analyses (Reference 4.1.21).

4.1.22 Revision 2

NEDO21888

c. Feedwater flow coastdown is modeled for plants with turbine driven feedwater pumps. The feedwater flow is manually stopped at 10 minutes after the break, for plants with motor driven feed water pumps.

d. Fluid subcooling is conservatively neglected.

e. The steam is assumed to stratify above the air in the drywell, which gives a higher rate of pressurization of the wetwell.

f. The bulk wetwell pool and airspace temperatures are equal throughout the transient. This maximizes the wetwell airspace temperature and pressure.

g. The hydrodynamic effects of vent clearing are neglected.

h. A constant liquid flashing rate for the break flow is used.

i. The condensation of steam on drywell walls and structures is con-servatively neglected.

j. A single failure of the HPCI (or FWCI where applicable) system is assumed.

For conservatism, the containment response has been determined assuming initial conditions which yield the most severe pressure and temperature in the containment. The conditions are as follows:

a. The reactor is operating a 102% of licensed thermal power as identified in Table 4.1.13 for a typical plant. This maximizes the core decay heat.

4.1.23 Revision 2

NEDO21888

b. The temperature of the suppression pool is equal to the maximum technical specification limit for normal operation. This maximizes the suppression pool temperature and the wetwell airspace temperature and pressure.

c. No credit is taken for normal auxiliary power, which maximizes the time it takes the ECCS system to be operable.

Other plant conditions and assumptions are as described for the DBA in Section 4.1.1.

4.1.2.3 Analysis Results Plant unique containment temperature and pressure responses have been cal-culated and are presented in the Plant Unique Load Definition documents and submitted individually to the Mark I Owners. Typical responses are shown in Figures 4.1.21 and 4.1.22.

4.1.24 Revision 2

5O~

400-DRYWELL AIR PURGED TO WETWELL (SI VESSEL DEPRESSURIZEI)

U. 300- IA) VENT CLEARING 0

w INITIATION OF ADS (CI I I I z

w I II tTl I-.

I I 0 I

U 200-I II g~.a t-.

cn 100 I II 0

10 100 1000 10.000 TIME AFTER UREAK INITIATION (i.c) m I-.

0, 0

Figure 4.1.21. Typical Containment Temperature Response to an Intermediate Break Accident

I w

tTl 0

H 0 0 0~

1 10 100 1000 10.000 TIME AFTER BREAK INITIATIQ~4 (,ecj (U

4 Figure 4.1.22. Typical Containment Pressure Response to an Intermediate Break Accident

NEDO21888 REFERENCES FOR SECTION 4.1.2 4.1.21 F. J. Moody, Maximum Discharge Rate of LiquidVapor Mixtures from Vessels, General Electric Company, Report No. NEDO21052, September 1975.

4.1.27/4.1.28 Revision 2

NEDO2 1888 4.1.3 Small Break Accident Typical containment temperature and pressure responses to a SEA are shown in Figures 4.1.31 and 4.1.32, respectively. Following the break, the drywell pressure will increase slowly and the water level in the vents will be slowly depressed until at point A in Figure 4.1.32, drywell air and steam pass through the suppression pool where the steam is condensed and the air will rise to the free airspace above the pool. At point B, the air is purged from the drywell to the wetwell, and at point C (10 minutes) the operator manually initiates ADS. At point D, just prior to low pressure ECCS initia-tion, the wetwell airspace volume has been reduced to its minimum value and the maximum wetwell pressure occurs. At approximately 20 minutes after the occurrence of the break, the reactor will be depressurized at point E. Fol-lowing reactor depressurization and ECCS flooding of the vessel, the sequence of events is the same as for the IBA described in Section 4.1.2.

During the depressurization of the reactor vessel, the suppression poo1 temp-erature increases rapidly due to the transfer of stored reactor vessel energy to the pool. The suppression poo1 temperature transient will level out after the reactor vessel is depressurized. The pool temperature will rise much less rapidly since the heat input to the pool is limited to the core decay heat and the RHR system is removing heat from the pool. No S/RV, condensation oscillation, or chugging loads occur after complete reactor depressurization.

4.1.3.1 Analytical Procedure The temperature and pressure response of the containment to the SEA has been determined by performing a mass/energy balance on the reactor vessel to obtain the reactor depressurization and temperature histories, and also by performing a mass/energy balance to obtain containment pressures and temperatures at the following times:

a. When the vents are half cleared

b. When the vents are fully cleared 4.1.31 Revision 2

NEDO21888

c. When half the drywell air has been purged to the wetwell

d. When all the drywell air has been purged to the wetwell

e. When the reactor vessel is completely depressurized.

4.1.3.2 Assumptions

a. The fuel decay heat and sensible energy used for the DBA and lEA analyses shown in Table 4.1.11 are used for the SEA analysis.

b. A steam break is assumed with an area equal to 0.01 ft2. The blowdown model is the same as the Homogeneous Equilibrium Model used in the DBA and lEA analyses (Reference 4.1.31).

c. Feedwater flow coastdown is modeled for plants with turbine driven feedwater pumps. The feedwater flow is manually stopped at 10 minutes after the break, for plants with motor driven feed-water pumps.

d. Stratification of the steam above the air in the drywell is assumed. This maximizes the pressurization rate of the wetwell.

e. The bulk wetwell pool and airspace temperatures are equal through-out the transient. This maximizes the wetwell airspace temperatures and pressures.

f. The hydrodynamic effects of vent clearing are neglected.

g. The condensation of steam on drywell walls and structures is neglected.

h. A single failure of one RHR pool cooling loop is assumed.

4

• 1.32 Revision 2

NEDO21888 For conservatism, the containment response has been determined assuming condi-tions at the time of the break which yield the most severe pressure and temp-erature in the containment. The conditions are as follows:

a. The reactor is operating at 102% of licensed thermal power as identified in Table 4.1.13 for a typical plant. This maximizes the core decay heat.

b. No credit is taken for normal auxiliary power. This maximizes the time it takes the ECCS system to be operable. Other assumptions and plant conditions at the instant of the break are as described for the DBA except the initial temperature of the suppression pool, which is as described for the lEA.

4.1.3.3 Analysis Results Plant unique containment airspace temperature and pressure responses are presented in the Plant Unique Load Definition documents and submitted individually to the Mark I Owners. Typical responses are shown in Figures 4.1.31 and 4.1.32.

4.1.33/4.1.34 Revision 2

400 300-VESSEL DEPRESSURIZED (E)

U. MINIMUM WETWELL AIRSPACE VOLUME (0) 0 w

a: INITIATION OF ADS (600 seconds) (CI I I

~

a:

w 200 DRYWELL AIR PURGED TO WETWELL (U) I I PI e

a:-

I-II I ~ C VENT CLEARING (A) H

-I I I 100 WETWELL I I I i 0

I I 10 100 1000 10,000 TIME AFTER BREAK INITIATiON (gee)

(U 4

0, 0 Figure 4.1.31. Typical Containment Temperature Response to a Small Break Accident

50 40 30 I

mal 3W 0

• K L

• 20 H (jJ 10 0

I 10 100 1000 10.000 TIME AFTER BREAK INITIATION ~sec)

(U

.4 p.1.

Figure 4.1.32. Typical Containment PresBure Response to a Small Break Accident

NEDO21888 REFERENCES FOR SECTION 4.1.3 4.1.31 F. J. Moody, Maximum Discharge Rate of LiquidVapor Mixtures from Vessels, General Electric Company, Report No. NEDO21052, September 1975.

4.1.37/4.1.38 Revision 2

NEDO21888 4.2 VENT SYSTEM THRUST LOADS The postulated DEA causes the most rapid pressurization of the containment sys-tem, the largest vent system mass flow rate, and therefore, the most severe vent system thrust loads. The pressurization of the containment for the tEA and SEA is less rapid than for the DEA; thus, the resulting vent system thrust loads for the SEA and lEA are less severe. Consequently, vent system thrust loads are presented for the DBA only.

Following the pipe break, the drywell pressure increases and the water initially occupying the submerged portion of the vent system is accelerated into the pool, clearing the vents. The structural loading during the vent clearing process is due to pressure imbalances between the increasing pressure in the vent sys-tem and the surrounding wetwell airspace. Momentum need not be considered since flow velocities during the vent clearing p~rocess are low. Following vent clearing, there is a period during which bubble formation and breakthrough occurs. This period is characterized by increasing flow velocities and associated pressure losses in the vent system. Following breakthrough, mass flow rates and velocities in the vent system become significant, and momentum must also be considered in determining thrust loading. The fluid flowing through the vent system following breakthrough is assumed to be a homogeneous mixture of air, steam, and water.

Forces on the vent system are derived from mass and momentum equations.

The sign convention used to define positive forces on a typical vent system is shown in Figure 4.21. Vent system thrust loads for this typical plant are shown in Figures 4.22 through 4.25 for 0 to 5 seconds, and Figures 4.26 through 4.29 for 0 to 30 seconds. Nomenclature for this section is given in Table 4.21.

Due to variations in vent system configurations, the nomenclature presented in Table 4.21 defines additional vent system forces and angles which are not applicable to the typical plant vent system geometry presented in Figure 4.21.

The vent system thrust loads presented in the Plant Unique Load Definition documents and submitted to the individual Mark I Owners, contain figures and 4.21 Revision 2

NEDO21888 legends which define the forces applicable to the plant unique vent system configuration.

The force transients presented in Figures 4.22 through. 4.29 show distinct changes in slope or inflection points. These are clearly illustrated in Figure 4.23 and are due to the large changes in flow conditions during the transient, as well as the pressure changes for the drywell and wetwell as determined for the DBA in Section 4.1.1. Up to the time of vent clearing, the absolute magnitude of the forces increases due to rising drywell pressure (assumed constant throughout the vent system) acting against constant wetwell airspace pressure. Following vent clearing, the absolute magnitude of the force on the main vents continues to increase since drywell pressure has not yet peaked and pressure losses from the drywell to the main vents are fairly small. Forces on the vent header and downcomers, however, are characterized by a decrease in magnitude due to large entrance and exit pressure losses between the components when the vent system flow rate has significantly increased. Figures 4.25 and 4.29 present the forces resulting from all main vents and downcomers as well as a net~~ vertical force on the entire vent system resulting from the sum of all component contributions. These figures do not show any net horizontal force on the entire vent system since the sum of all horizontal forces is zero due to the symmetrical configuration of the vent system.

The vent system is subjected to internal pressures as shown in Figures 4.210 and 4.211 for a typical plant. These pressure histories correspond to the force transients shown in Figures 4.22 through 4.29. As stated earlier, vent system pressures are equal to the drywell pressure during the vent clearing process.

4.22 Revision 2

NEDO 21888 4.2.1 Analytical Procedure The analytical procedure employs thrust equations which consider the forces due to both pressure distributions and momentum to define horizontal and vertical thrust forces on the following components of the vent system as identified in Figure 4.21:

a. Main vents

b. Vent header

c. Downcomers.

Because both the main vents and the downcomers are located symmetrically about the center of the vent system, the horizontal vent system thrust loads cancel each other resulting in no net effective horizontal vent system thrust load.

The thrust calculations are based upon plant unique containment pressure response and vent system mass flow rates from analyses similar to those described in Section 4.1.1 for the DBA, but are performed at both minimum and zero initial pressure differential between the drywell and wetwell. The thrust equations define loading on the vent system components during four distinct time periods:

a. Before Vent Clearing The entire vent system (excluding the portion of the dowucomers containing water) is assumed to be at drywell pressure. Thrust forces result from pressure in the vent system and the surrounding wetwell airspace pressure, as determined for the DBA in Section 4.1.1.

The following equations were used to calculate the horizontal and vertical forces on a single main vent, vent header (per mitre bend) 4:2.11 Revision 2

NEDO21888 and a single dowucomer. The nomenclature for terms used below are listed in Table 4.21. Depending on the vent system configuration, not all equations listed below are necessarily used to define vent system thrust forces:

FlVl = (PWW PDW) (A~/n1) (sin 81)

FlHl (PWW PDW) (A.~/n1) (cos 81)

F1V2 = (PWW PDW) (A.~/n1) (sin a sin 81)

FlH2 = (PWW PDW) (A.~/n1) (cos a cos F2V = (PDW PWW) (~C/n3) (sin 82)

F2H = (PDW PWW) (A,.~) (2 2 sin 5)1/2 F3V = (PDW PWW) (~C/n2) (sin 83 sin 82)

F3H = (PDW PWW) (ADC/n2) (cos cos 82)

F4V = (PDW PWW) (A.~C/n2) (1 sin 83)

F4H (PDW PWW) (ADC/n2) ( cos

b. At Vent Clearing At vent clearing all water initially in the downcoiner has been cleared. Since the bubbles at the end of the downcomers are just starting to grow, the flow velocities are sufficiently low such that momentum effects can be ignored. Thrust forces are due to imbalances between drywell and wetwell airspace pressures as obtained from the DBA analyses described in Section 4.1.1. The equations used to calculate the forces are the same as those pre-sented in item (a) above.

4.2.12 Revision 2

NEDO21888

c. After Vent Clearing and Before Bubble Breakthrough After the vents have cleared, bubble formation occurs at each down comer exit. Since the analytical methods described in Section 4.1.1 do not account for bubble formation, mass flow rates predicted by this method during bubble formation are not sufficiently descriptive for thrust force calculations. Therefore, plant unique QSTF experi-mental data (Reference 4.21) were examined to determine the shape of the force curve. The results of this analysis showed that a straight line approximation of the force curve from the time of vent clearing to the time of bubble breakthrough, when flow rate information obtained from the Section 4.1.1 analyses is applicable, would be conservative since an exponential decay of forces was observed in the QSTF. The time period from vent clearing to bubble breakthrough was conservatively chosen as 200 milliseconds. The straight line approximation is shown between points A and B in Figures 4.22 through 4.25 and Figure 4.210. In order to assure the proper transition between vent clearing and bubble breakthrough for those plants that~

propose operation with a differential pressure control, the vent clearing time shall be derived from a containment analysis assuming..-

no drywell/wetwell differential pressure and this time shall be applied to the vent system transients calculated from a containment response with the proposed drywell/wetwell differential pressure.

The equations used to calculate the horizontal and vertical forces on a single main vent, vent header (per mitre bend) and a single downcomer at the time of bubble breakthrough are presented below.

The mass flow rates plus drywell and wetwell pressures, as determined for the DBA in Section 4.1.1, were used in the calcula-tions at the time of bubble breakthrough. The nomenclature is listed in Table 4.21.

4.2.13 Revision 2

NEDO21888 FlVl = ((PwwPl) (A~p/n1) sin FlHl = ((PijwPl) (A~p/n1) - cos 01 FlV2 = ((PWwPl) (A~p/n1) (m~V1)/(n1g)) (sin a sin 61)

FlH2 = ((PWWPl) (A.~/n1) (m~V1)/(n1g)) (cos a cos 01)

F2V = ((P2Pww) (A~/n3) + (m~V3)/(n3g)) sin 02 Revision 2

NEDO21888 F2R = ((P 2PWw) AVE + (~~V9/(n3g~)) (2 2 sin F3V = ((P3PwW) (A~~/n2) + (~V3)/(n2g)) (sin 03 sin 82)

F3H = ((P 3Pww)

(~C/n2) + (~V3)/(n2g)) (cos 03 cos 02)

F4V ((P3Pww) (A.DC/n2) + (~V3)/(n2g)) (1 sin 03)

F4H = ((P3Pww~)

(~C/n2) + (~V3)/(n2g)) ( cos 03)

d. After Breakthrough Following bubble formation and breakthrough, full flow through the vent system is established. During this period, fluid momentum becomes a significant term in the thrust force calculation. Mass flow rates and drywell and wetwell pressures, as determined for the DBA in Section 4.1.1, were used in the calculations for the time period following bubble breakthrough. The equations used to calculate the forces are the same as those presented in item (c) above.

4.2.14 Revision 2

NEDO2 1888 4.2.2 Assumptions

a. Flow losses in the vent system following breakthrough have been selected to yield conservatively high pressures in the main vent, vent header and dowucomers, thus resulting in conservatively high thrust forces. The following flow loss distribution is typical for all Mark I plants and is representative of calculated local loss coefficients:

(1) Pressure drop from the drywell to the main vent is 25% of~

the total calculated pressure drop in the vent system.

(2) Pressd~e drop from the main vent to the yent header is 37.5%:.

of the total calculated pressure drop in the vent system.

(3) Pressure drop from the vent header to the downcomer is 27.5% _

of the total pressure drop in the vent system.

In other words, at any point in time the main vent is at a pressure equal to the drywell pressure minus 25% of the total pressure drop in the system, the vent header is at drywell pres-sure minus (25% + 37.5%) of the pressure drop, and so on. This assumes that losses are distributed at entrances and exits of the components. This is reasonable since form and friction losses in the vent system components themselves are small.

It should be noted that the total pressure drop from the drywell to the downcomer is assumed to be 90% [sum of items (1), (2), and (3) above] of tho total pressure drop from the drywell to the wetwell.

The remaining 10% is assigned as exit loss from the downcomer to the suppression pool.

b. Flow in the vent system is completely turbulent and loss coefficients are independent of Reynolds number.

4.2.21 Revision 2

NEDO2 1888

c. The flow of liquid, steam, and air following vent clearing is conservatively assumed to be homogeneous.

d. Momentum effects are negligible prior to vent clearing because fluid velocities are very low during this period.

e. Unsteady fluid flow considerations are negligible following vent clearing because they have a second order effect on the calculated thrust loads.

f. Fluid flow from the drywell to the vent system is assumed uniform.

The effect of nonuniform flow in the downcomers observed in the EPRI 1/12 scale pool swell tests (References 4.22 and 4.23) is discussed in Section 4.2.4.

4.2.22 Revision 2

NEDO21888 4.2.3 Analysis Results Plant unique force transients and containment pressure responses have been generated for the following two sets of initial containment conditions:

a. Maximum submergence and minimum normal operating pressure differential between the drywell and the wetwell

b. Maximum submergence and zero pressure differential between the drywell and the vetwell.

The results are presented in the Plant Unique Load Definition documents and submitted to the individual Mark I Owners. Results for a typical plant are presented in Figures 4.22 through 4.210.

The vent system thrust loads are calculated and presented for the initial 30 second period following the DEA. Thrust loads after 30 seconds stay virtually constant at the value calculated at 30 seconds.

Some Mark I plants normally operate with zero pressure differential between the drywell and the wetwell and therefore, only one set of results are presented for these plants.

4.2.31/4.2.32 Revision 2

NEDO21888 4.2.4 Application The horizontal and vertical main vent thrust transients shown in Figures 4.22 and 4.26 represent the resolution of the thrust load which is acting on the end cap of the main vent. This loading is actually distributed over the end cap area, and is shown in Figure 4.212. This figure also shows main vent mitre bend forces for the two plants (Browns Ferry and Oyster Creek) which have this configuration. It should be noted that the Browns Ferry main vent mitre bend lies between the drywell and the bellows on the main vent. The Oyster Creek mainvent mitre bend lies on a continuous run of piping (i.e., downstream of the bellows).

The vertical and horizontal vent header thrust transients shown in Figures 4.23 and 4.27 represent the vent header loading per mitre bend. Vertical loading is due to the contributions of the individual dowucomer pairs, which were assumed equal for each pair in the analysis. As discussed in Section 4.2.2 the flow in the downcomers is not uniform. Figure 4.213 defines the effect of this nonuniform flow.

The horizontal and vertical downcomer thrust transients shown in Figures 4.24 and 4.28 are loads for a single downcomer, and are the resultant forces due to turning the flow through the mitre bend. The resolution of this load into vertical and horizontal point loads is shown in Figures 4.214 and 4.215.

Figure 4.214 also shows the mitre bend loading for those plants which have two downcomer mitre bends. Duane Arnold, which is the only plant with straight vertical downcomers, has no downcomer forces. The discussion pertaining to nonuniform dowecomer flow on vent header forces also pertains to the forces generated for downcomers. Figure 4.214 defines the effect of this distributed flow.

In addition, total vertical thrust loads and net vertical thrust loads, shown in Figures 4.25 and 4.29, are defined as follows:

FlVlT = Main vent end cap vertical force multiplied by the number of main vents 4.2.41 Revision 2

NEDO-.21888 FlV2T = Main vent mitre bend vertical force multiplied by the number of main vents (Browns Ferry and Oyster Creek only)

F2VT = Vent header vertical force per mitre bend multiplied by the number of vent header mitre bends F3VT = Downcomer mitre bend vertical force multiplied by the number of dowucomers F4VT = Second dowucomer mitre bend vertical force multiplied by the number of downcomers FNETV = FlVlT + FlV2T + F2VT + F3VT + F4VT For plants which do not have main vent mitre bends, or less than two downcomer mitre bends, forces which apply to these bends will be zero.

4.2.42 Revision 2

NEDO21888 Table 4.21 N0~NGLATURE FOR SECTION 4.2 PDW Drywell pressure P14W Wetwell airspace pressure P1 Main vent pressure P2 Vent header pressure P3 Downcomer pressure FlVl Vertical force on a single main vent end cap FlHl Horizontal force on a single main vent end cap FlV2 Vertical force on a single main vent mitre bend (applicable to Browns Ferry and Oyster Creek only)

FIH2 Horizontal force on a single main vent mitre bend (applicable to Browns Ferry and Oyster Creek only)

F2V Vertical force on vent header (per mitre bend)

F2H Horizontal force on vent header (per mitre bend)

F3V Vertical force on a single downcomer mitre bend F3H Horizontal force on a single downcomer mitre bend F4V Vertical force on second mitre bend of dowecomer (if applicable)

F4H Horizontal force on second mitre bend of downcomer (if applicable)

FlVlT Total main vent end cap vertical force FlVl x number of main vents FlV2T Total main vent mitre bend vertical force = FlV2 x number of main vents F2VT Total vent header vertical force F2V x number of vent header mitre bends F3VT Total vertical force (first downcomer mitre bend) =

F3V x number of downcomers 4.2.43 ReviBion 2

NEDO21888 Table 4.21 (Continued)

NOMENCLATURE FOR SECTION 4.2 (Continued)

F4VT Total vertical force (second downcomer mitre bend) =

F4V x number of dowucomers FNETV FNETV = FlVlT + FlV2T + F2VT + F3VT + F4VT AVE Vent header flow area Avp Total main vent flow area ADC Total downcomer flow area Number of main vents Number of downcomers n3 Number of vent header mitre bends Total mass flow rate T

Fluid velocity in main vent V1 Fluid velocity in vent header V2 V3 Fluid velocity in downcomer Angle of main vent with horizontal 81 Angle of first downcomer mitre bend with horizontal 83 Angle of second dowacomer mitre bend with horizontal a Angle of main vent mitre bend with horizontal 900 (vent header mitre bend angle)

B 4.2.44 Revision 2

NEDO21888 PLAN FiVi FiVi SECTION AA FiVi - VERTiCAL FORCE ON MAIN VENT END CAP Fl Hi - HORIZONTAL FORCE ON MAIN VENT END CAP F1V2 = VERTICAL FORCE ON MAIN VENT MITRE BEND (ONLY IF a

• 0)

FlH2 - HORIZONTAL FORCE ON MAIN VENT MITRE BEND (ONLY IF a

• 0)

F2V - VERTICAL FORCE ON VENT HEADER (PER MITRE BEND)

F21-I HORIZONTAL FORCE ON VENT HEADER (PER MITRE BEND)

(SEE FIGURE 4.2.15 FOR RESOLUTION OF HORIZONTAL FORCESF~

F3V - VERTICAL FORCE ON DOWNCOMER MITRE BEND F3H - HORIZONTAL FORCE ON DOWNCOMER MITRE BEND FORCES ARE SHOWN IN THEIR ASSUMED POSITIVE DIRECTION Figure 4.21. Definition of Positive Thrust Loads 4.2.45 Revision 2

0

-100 4

a:

0 U.

a:- H C

-200 400 0 I 2 3 4 5 TIME AFTER BREAK INITIATION (s.d (U

4 t~) Figure 4.22. Single Main Vent Forces (05 eec)

60 I

w U

z N

• 0 N 0 a

• IL N

H

-4 cc cc 0 1 TIME2 AFTER BREAK INITIATION 3 lied 4 5 24 Figure 4.23. Vent Header Forces Per Mitre Bend (05 sec)

5 4

3 2

w a:- 0 U.

.1 H

-2

.3

.4

-5 0 1 2 3 5 4

TIME AFTER UREAK INITIATION (see)

Figure 4.24. Single Downcomer Forces (05 sec)

W0 800 700 600 400 300 I

~ 200 w

a: U

~l00 0

U.

I 0 N)

I.

.0

.100

.200 400

.400 00 (U

I. 0 1 2 3 6 0,

TIME AFTER BREAK INITIATION (see) r3 Figure 4.25. Total Vertical Forces, Net Vertical Force (05 eec)

0

-100 w N N

U 0

U. N)

I H C

o C C

-200

-300 4 0 5 10 15 20 26 30 0,

I, TIME AFTER IREAK INITIATION (usc) 0 I~.J Figure 4.26. Single Main Vent Forces (030 see)

50 40 30 4

.1:- w z LI N C 0

IL I 20 H C

C 10 0

0 5 10 15 20 25 I-,.

TIME AFTER BREAK INITIATION (see) 0 N

Figure 4.27. Vent Header Forces Per Mitre Bend (030 sec)

S 4

3 2

ji U

U IL t.

20 to Js ta I- I-A ha 1 C 03 03 2

4

-4

-6 (U 0 5 10 15 20 25 30 4

H 0, TIME AFTER UREAK INITIATION (see)

I-.

0 0

Figure 4.28. Single Downcomer Forces (030 eec)

m 000 700 600 00 400 300 200 Js w

U 131

~ 100

-is 0 U. 0 I-.

1.) 0 H

-100

-200

-300

-400 400

-600

-700 0 5 10 15 20 25 30 TIME AFTER BREAK INITIATION (aec)

Figure 4.29. Total Vertical Forces, Net Vertical Force (030 sec)

100 90 80 70 I

w IL 60 w

IL - MAXIMUM DRYWELL TO WETWE LI PR ESSURE DIF F ER ENTIAL ha a:-

I ha C- 40 H 03 03 03 30 m P2 ________ _______ _________

20 m.a~ a a ~ a a a a _______

10 U....

0 I I I 0 2 3 4 5 (U

4 TIME AFTER UREAK INITIATION (eec) 0, 0

ha Figure 4.210. Pressure Time Histories (05 sec)

100 90 so 70 160 w

IL a:- 2:

ha a. t?1 is I I-. ha U H C

03 03 30 20 10 0

(U 4

H 0 TIME AFTER BREAK INITIATION ieee) ha Figure 4.211. Pressure Time Histories (030 sec)

NEDO21888 FiVi DISTRIBUTED PRESSURE LOAD ON MAIN VENT END CAP IS RESOLVED INTO VERTICAL AND HORIZONTAL POINT LOADS.

FOR PLANTS WITH MAIN VENT MITRE BENDS (BROWNS FERRY AND OYSTER CREEKI FiVi FIV2 AND F1H2 ARE ACTUALLY DISTRIBUTED AROUND THE BEND AND ARE THE RESULTANTS OF INTERNAL PRESSURE AND SHEAR FORCES ON THE PIPING Figure 4.212. Application of Thrust Force on Main Vent End Cap and Main Vent Mitre Bend 4 .2 .416 Revision 2

NEDO21888 PLAN 3ECTEON F2V Is THE VENT HEADER VERTICAL FORCE PER MITRE BEND AND REPRESENTS THE CONTRiBUTIONS OF DOWNCOMERS 1 THROUGH 6 LET (F2v) 1 BE THE CONTRIBUTION TO F2V FROM DOWNCOMER 1 (F2V)2 BE THE CONTRIBUTION TO F2V FROM DOWNCOMER 2 (F2V)6 BE THE CONTRIBUTION TO F2V FROM DOWNCOMER 6 FROM THE ANALYSIS, (F2V)1 = (F2V)2 = (F2v)3 (F2V)4 = .(F2V)5 = (F2V)6 F2V/6 HOWEVER, FROM EPRI TESTING (REFERENCE 4.2-3),

(F2V)1 (F2V/6) (0.79); (F2V)4 = (F2V16) (0.95)

IF2V)2 = (F2V/6) (0.9); (F2V)6 = (F2V/6) (1.19)

(F2V)3 = (F2V/6) (1.1); (F2V)6 (F2V/6) (1.17)

THESE FACTORS REPRESENT THE MAXIMUM DEVIATION FROM THE UNIFORM FLOW ASSUMPTION, APPLY ONLY AFTER VENT CLEARING AND ARE ASSUMED TO REMAIN CONSTANT DURING THE REMAINING VENT HEADER THRUST LOAD TRANSIENT Figure 4.213. Application of Vent Header Forces 4.2.417 Revision 2

NEDO21888 F3V F3V P32.4 DOWNCOMER F4H SINGLE MITRE SENO DOUSLE MITRE UEFQ VENT HEADER TOP VIEW DOWNCOMER IN THE ANALYSIS F3V. F3H, F4V, AND F4H ARE ASSUMED EQUAL FOR DOWNCOMERS 1 THROUGH 6.

HOWEVER, FROM EPRI TESTING (REFERENCE 4.2-3). THE FOLLOWING FACTORS APPLY AFTER VENT CLEARING:

(F3V) 1 F3VXO.79; (F3V)4 = F3VXO.95 (F3V)2 F3V X 0.90; IF3V)5 = F3V X 1.18 (F3V)3 F3V X 1.01; (F3V)6 = F3V X 1.17 THE SAME FACTORS WOULD APPLY FOR F3H. F4V, AND F4H [I.E. (F4H)i - F4H X 0.794]

Figure 4.214. Application of Downcomer Forces

4. 2. 418 Revision 2

NED 0 21888 VENT HEADER

-~

F2H2 F214 Figure 4.215. Resolution of Horizontal Forces 4.2.419 Revision 2

NEDO2 1888 REFERENCES FOR SECTION 4.2 4.21 J. M. Humphrey, Mark I Containment Program 114 Scale TwoDimensional Plant Unique Fool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.22 J. E. Torbeck, et al., Mark I Containment Program 1/12 Scale Pressure Suppression Pool Swell Tests, General Electric Company, Report No. NEDO13456, May 1976.

4.23 The Electric Power Research Institute, Three Dimensional Pool Swell Modeling of a Mark I Suppression System, EPRI, NP906, October 1978.

4.2.420 Revision 2

NEDO2 1888 4.3 POOL SWELL LOADS In the event of a postulated design basis LOCA due to a break in a large pressurized line, the drywell and vent system are pressurized and the water initially in the downcomers is accelerated downward into the suppress ion pool. After the downcomers are cleared of water, air is discharged into the wetwell below the pool surface and the pooi swell transient starts.

The vent system air is initially at drywell pressure. When the design basis LOCA occurs, the pressure rapidly increases in the drywell and vent system, which results in the water leg in the downcomers being pushed out into the suppression pool. When this clearing process is completed, the air behind the downcomer water slug produces a bubble at the end of the downcomer. The water above the bubble is accelerated upward as the bubble expands. As the bubble expansion continues, the pool water rises in the torus and compresses the airspace above the pool surface. Eventually, the bubble breaks thrDugh to the torus airspace and the displaced pool liquid settles back to its original level.

Associated with pool swell are the following eight primary loads:

a. Torus net vertical loads

b. Torus shell pressures

c. Vent system impact and drag loads

d. I.mpact and drag loads on other structures above the pool

e. Froth impingement loads

f. Pool faliback load

g. Downcomer water clearing jet load on submerged structures

h. LOG-A bubble induced drag loads on submerged structures 4.31/4.32 Revision 2

NEDO21888 4.3.1 Torus Net Vertical Load Histories In the postulated LOCA, the downcomer air, which is at essentially drywell pressure, is injected into the suppression pool, producing a downward reaction force on the torus. The consequent bubble expansion causes the pool water to swell in the torus, compressing the airspace above the pool and producing an upward reaction force on the torus. The bubble pressure decreases as the bubble overexpands and the pool liquid mass decelerates. Eventually, the bubbles break through to the torus airspace (the bubble and torus airspace com-municate) and the displaced poo1 liquid settles back toward its original level.

These vertical loads create a dynamic imbalance of forces on the torus, which acts in addition to the weight of the water applied to the torus.

This dynamic force history lasts for only a few seconds and typically includes a downforce shortly after vent clearing, followed by an upforce as the airspace is compressed.

A typical force history is shown in Figure 4.3.11.

4.3.1.1 Bases and Assumptions The torus net vertical load transients are based on 1/4 scale 2D plant unique test results performed in the QSTF (Ref. 4.3.14). The geometric characteristics of the torus and vent system and the LOG-A driving conditions were represented in this facility for each plant tested. The test facility design and 0peration is based on a scaling law analysis which is presented and verified in Reference 4.3.11.

For each plant unique series of tests, the plant short term DBA drywell pres-surization history was calculated using an analytical model of the drywell response as described in Section 4.1.1. The resulting pressure history (scaled for the test) was used as a lower bound for the test drywell pressuri-zation history, assuring a conservative load definition. Figure 4.3.12 4.3.11 Revision 2

NEDO21888 presents a comparison of an analytically determined drywell response to the measured response for a representative QSTF test. The QSTF tests were per-formed at the minimum plant operating AP and maximum downcomer submergence for the plant. These conditions will result in a conservative prediction of the net torus vertical loads. The plant unique test conditions are included in Reference 4.3.14.

As indicated in Raferences 4.3.11 and 4.3.14, orifices were placed in the 1/4 scale vent system in order to accurately model the vent system resistance (Section 4.1.1 specifies the plant unique flow resistances used). Reference 4.3.11 describes the scaling laws used to obtain flow resistances (fL/D) which produced correctly scaled enthalpy flow out of the downcomers. Scaling was based on the following considerations:

Ia. As identified in Reference 4.3.13 the flow resistance of the 1/4 scale vent system was obtained by scaling the plant unique flow resistances (fL/D) by the quantity l/(length ratio of the model) to satisfy the enthalpy flow scaling law.

b. Additionally, the vent flow resistance was decreased to account for the fact that the initial air temperature in the drywell for the 1-14 scale tests was 700F, while the plant drywell operating temperature is higher. In order to obtain the correctly scaled enthalpy flow, the QSTF flow resistance had to be decreased (thereby increasing the QSTF enthalpy flow) by the inverse ratio of the actual plant and QSTF drywell temperatures. For example, for a drywell operating temperature of 135F, the adjustment factor to the vent flow resistance is 460 + 70 460 + 135 .891, resulting in the correct enthalpy flow to the bubbles.

c. As discussed in Reference 4.3.14 the vent system flow resistance was evenly split by placing orifices at the top of the vent pipe and in the downcomers. ~his configuration is a reasonable approxi-mation of the continuous flow loss distribution in the actual Mark I vent system.

4.3.12 Revision 2

NEDO-21888 Figure 4.3.13 compares the enthalpy flow measured in a typical 1/4 scale test and a calculated enthalpy flow once quasi steady flow has been obtained. The calculated enthalpy flow was found using the drywell pressure history calculated as described in Section 4.1.1 and assuming steady state Fanno flow of 100% air through the vent system. Figure 4.3.13 demonstrates that the quasi steady state enthalpy flow achieved in the test exceeds the calculated enthalpy flow and is therefore conservative. The enthalpy flow produced in the test facility is also conservative prior to reaching its quasi steady state flow condition due to the conservatism of the drywell pressure history (see Figure 4.3.12) and the selection of the vent system loss coefficient for the QSTF tests.

For all QSTF tests the test conditions were set to correspond to the plant operat-ing conditions which produce the most severe loading. For example, increased dry well to wetwell Al decreases the torus vertical loads and vent header impact loads. The QSTF tests were, performed with the minimum plant operating eiP as the maximum test Al. Other parameters treated in a conservative manner include downcomer submergence, torus water level, drywell pressurization, and vent system flow resistance. In addition, the geometric tolerances of the facility were established such that any effect on the measured loads would be con-servative. The load definitions affected by these conservatisms include those discussed in this section and Sections 4.3.2 and 4.3.3.

The net torus load histories are based on a spatial integration of the pres-sure transducers located on the QSTF torus shell. Reference 4.3.12 presents net tor~is load data from the first series of QSTF tests as measured by both the pressure transducer integration and the torus load cell. This data con firms that the pressure transducer integration accurately measures the net torus vertical load.

4.3.13 Revision 2

NEDO21888 The net torus loads are based on results from the QSTF, which is a 2D facility with average vent spacing. Reference 4.3.12 addresses the possible difference between loads in a threedimensional test facility versus those in a two dimensional facility. That report concludes that loads derived from two dimensional test results do not need to be adjusted to account for three dimensional effects. Thus, the plant unique net torus load transients shown in the Plant Unique Load Definitions are appropriate for a three dimensional load definition. However, for added conservatism, each loading phase of the plant unique net torus load transients will have additional margins. For the 5

downward loading function, DOWN (DOWN) +2 x 10 (PEAKDOWN ) 2 is mean mean - -

derived in terms of a fraction of the load at the time of the peak downward load, and that fraction is applied to the entire downward loading phase. For the upward loading function, the UP (UP) + 0.215 (UP ) is applied mean mean to the entire upward loading phase. In the above discussions, mean refers to the average of the QSTF plant specific test results (lbf). These margins shall be applied to the QSTF mean load function prior to scaling the load function to full scale equivalent conditions. The additional margin for theT zero ~P condition, for those plants that have an operating Al, will be based upon~ a single test. i 4.3.1.2 Load Definition The torus net vertical dynamic loads are given on a plant unique basis as load histories. Static loads (i.e., water and structural weights) are not included in these load histories. The net dynamic load is given as an equivalent pressure acting on the projected plan area of the torus. In the Plant Unique Load Definitions, load transients are given for each Mark I plant. For those plants whose normal operating condition includes a drywell!

wetwell Al, load transients are given for both the zero and actual AP conditions. The peak net torus vertical loads are listed in Tables 4.3.11 and 4.3.12.

The vertical load transients given in the Plant Unique Load Definitions are mean (average) load magnitudes for those cases where the testing basis included multiple tests (i.e., normal operating conditions). The load transient for the zero Al condition, for those plants that have an operating Al, was based upon a single test.

4.3.14 Revision 2

NEDO2 1888 The use of mean values is appropriate since there are a number of conservatisms inherent in the load definition. The calculation of the plant drywell pressurization rate includes the conservatisms outlined in Section 4.1.1.2.

Furthermore, the test drywell pressurization transient always bounded the cal-culated transient, as illustrated in Figure 4.3.13. Finally, the QSTF tests were performed to match each plants unique geometry and operating conditions and facility geometric tolerances were set to produce the most severe loading condition.

The weight of the pool water can be included in the net torus vertical load definition by subtracting the effective water pressure from the load transient.

The effective water pressure is equal to the weight of the pool water divided by the torus horizontal projected area, i.e.,

W F (t) = P (t) water net,tot. net A torus where lnet (t) = mean (average) torus load definition Pnet, total (t) = total pool swell load on torus, including dynamic load and static water pressure, psi W weight of pool water, lbf water At = torus horizontal projected area, in.2.

For determination of the water mass available for resisting torus uplift, refer to Reference 4.3.14.

4.3.15/4.3.16 Revision 2

NEDO21888 Table 4.3.11 MEAN PEAK NET TORUS VERTICAL LOADS*

OPERATING DRYWELL/WETWELL PRESSURE DIFFERENTIAL (Al)

Up Down AP Submergence Plant (psi) (psid) (ft) Deflector Browns Ferry 1,2,3 2.99 6.54 1.1 3.50 No Brunswick 1,2 6.14 9.75 0.0 4.33 Yes Cooper Station 5.50 9.12 1.0 4.38 No Dresden 2,3 4.93 7.66 1.0 4.00 Yes Duane Arnold 1.99 3.61 1.1 3.38 No Fermi 2 2.66 4.89 1.0 3.33 Yes Fitzpatrick 3.96 5.96 1.7 4.42 Yes Hatch 1 3.23 5.62 1.5 4.00 Yes Hatch 2 5.42 10.74 0.0 4.33 No Hope Creek 1,2 3.46 7.60 0.0 3.33 No Millstone 3.54 6.12 1.2 3.33 Yes Monticello 6.44 9.91 1.0 4.25 Yes Nine Mile Point 5.06 8.17 1.0 4.25 Yes Oyster Creek 4.71 8.35 1.0 4.06 Yes Peach Bottom 2,3 4.84 8.82 1.1 4.40 Yes Pilgrim 5.57 7.61 1.5 4.80 Yes Quad Cities 1,2 4.93 7.66 1.0 4.00 Yes Vermont Yankee 4.33 7.20 1.7 4.54 Yes

• The values given are based upon initial QSTF tests and may change due to additional tests with different operating conditions and plant modifications.

Additional margin as defined in Section 4.3.1.1 is to be added to UP and DOWN loads.

4.3.17 Revision 2

NEDO21 888 TABLE 4.3.12 SINGLE TEST PEAK NET TORUS VERTICAL LOADS*

ZERO DRYWELL /WETWELL PRESSURE DIFFERENTIAL Up Down Submergence Plant (psi (psi (ft) Deflector Browns Ferry 1,2,3 3.13 10.50 3.50 No Brunswick 1,2 ** ** 4.33 Yes Cooper Station 6.70 13.18 4.38 No Dresden 2,3 5.89 11.91 4.00 Yes Duane Arnold 2.82 5.86 3.33 No Fermi 2 3.28 8.01 3.33 Yes Fitzpatrick 6.08 13.00 4.42 Yes Hatch 1 5.33 10.68 4.00 Yes Hatch 2 ** ** 4.33 No Hope Creek 1,2 ** ** 3.33 No Millstone 5.80 10.07 3.33 Yes Monticello 7.91 12.58 4.17 Yes Nine Mile Point 5.86 11.75 4.25 Yes Oyster Creek 5.57 13.15 4.06 Yes Peach Bottom 2,3 6.16 10.27 4.40 Yes Pilgrim 7.44 11.94 4.80 Yes Quad Cities 5.89 11.91 4.00 Yes Vermont Yankee 7.20 15.93 4.54 Yes

• The values given are based upon initial QSTF tests and may change due to additional tests with different operating conditions and plant modifica-tions. Additional margin as defined in Section 4.3.1.1 is to be added to UP and DOWN loads.

• • These plants have a normal operating condition of zero drywell/wetwell pressure differential, so no single test was performed at this condition.

Their mean peak loads from a series of four tests at this condition are reported in Table 4.3.11.

4.3.18 RevisiOn 2

NEDE2 1888 12 8

4 w

~ 0 LU C

0~

-4

-8 12 600 800 1000 TIME AFTER BREAK INITIATION (misc)

Figure 4.3.11. Typical Net Torus Vertical Loading History 4.3.19 Revision 2

NEDO21888 38 32 TEST CALCULATED U)

ID LU C-.)

z LU

~ 16 U,

Cd, LU PEAK DOWNLOAD a-12 8

4 0

0 0.2 0.4 0.6 0.8 1.0 1.2 TIME AFTER BREAK INITIATION hoc)

Figure 4.3.12. Comparison of Typical Analytical and Test Drywell Pressure Histories 4 .3. 110 Revision 2

NEDO21 888 3.6 3.2 2.8 2.4 U

a,

~0 2.0 x

0

-J U-a-

-J 1.6 C

I 4..-

z LU 1.2 0.8 0.4 0

0 100 200 300 400 500 600 700 TIME AFTER VENT CLEARING (msec)

Figure 4.3.13. Enthalpy Flow Comparison 4.3.111/4 .3.112 Revision 2

NEDO21888 REFERENCES FOR SECTION 4.3.1 4.3.11 D. L. Galyardt, et al., Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Scaling Evaluation, General Electric Company, Report No. NEDO21627, January 1978.

4.3.12 Mark I Containment Program, Comparison of GE and EPRI Torus Load Test Results, Prepared for the General Electric Company by Nuclear Services Corporation, Report No. NEDO21973, September 1979.

4.3.13 J. E. Torbeck, et al., Mark I Containment Program 1/12 Scale Pressure Suppression Pool Swell Tests, General Electric Company, Report No.

NEDO13456, May 1976.

4.3.14 J. M. Humphrey, Mark I Containment Program, 1/4 Scale TwoDimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program, 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Test, General Electric Company, Report No. NEDO24615, June 1980.

4 .3. 113/4 .3. 114 Revision 2

NEDO21888 4.3.2 Torus Shell Pressure Histories The net vertical load transient on the torus during pool swell was described in Section 4.3.1. In this section the same torus load is described in terms of the local torus shell pressure history. When integrated over the torus shell inside surface, these local pressure histories result in the net torus load transient due to pool swell.

4.3.2.1 Bases and Assumptions The local torus shell pressure histories are based on test data from the QSTF and the 1/12 Scale 3DTest Facility (References 4.3.21 and 4.3.22). The loads are defined by providing an average submerged pressure history which is applicable to the submerged portion of the torus shell; a table of multipliers which specifies the variation of the average submerged transient pressure at different positions on the shell; and a pressure transient applicable to the entire torus airspace, which applies to all points in that region.

The average submerged pressure transients, which, in magnitude, are equal to the net force applied to the submerged portion of the torus during pool swell divided by the torus horizontal cross section area, are given on a plant unique basis. Torus shell pressure transients measured during the QSTF plant unique tests formed the bases for plant unique torus shell pressure transients.

The equivalent differential pressure corresponding to respective margins in Section 4.3.1.1 will be added to both downward and upward loading phases.

The spatial variation of the average submerged pressure transient magnitude is based on both the 1/4 scale and 1/12 scale tests. The variation in the torus circumferential direction is based on the 1/4 scale 2D test results; the variation in the torus longitudinal direction is based on the 1/12 scale 3D test results. The spatial variation of the average submerged pressure magni-tude varies with time during the poo1 swell transient. For this reason, the spatial variation depends not only on position on the submerged portion of the torus shell, but also on time. Only the magnitude of the average sub-merged pressure transient varies at different positions on the torus shell; the timing of the pressure transient is the same at all submerged locations.

Finally, the airspace pressure transients are based on the 1/4 scale tests.

4.3.21 Revision 2

NEDO21888 Representative pressure histories for the average submerged pressure transient and airspace pressure transient are given in Figures 4.3.21 and 4.3.22.

The assumptions made when determining the torus shell pressure histories are:

a. The normalized longitudinal variation of the submerged pressure history as measured in the 1/12 Scale 3D tests is applicable for all Mark I plants.

b. Effects due to the torus mitre joint are assumed to be negligible.

c. The loads derived from testing in the relatively rigid QSTF are assumed to provide a conservative load definition for the entire range of structural stiffnesses of the actual Mark I tori.

The longitudinal variation of the torus shell pressure magnitudes is primarily the result of the spacing of the downcomer pairs along the length of the torus. The downcomer spacing is not exactly the same in all of the plants, but is quite similar. Therefore, the results from the 1/12 scale tests can be applied to all the Mark I plants.

The effects of the torus mitre joints on the pool swell phenomenon are dis-cussed in Reference 4.3.22, where it is concluded that the effects are negligible.

4.3.2.2 Load Definition The Plant Unique Load Definitions contain the plant unique average submerged torus pressure histories and airspace pressure histories. In all these transients, the initial static pressure has been subtracted out so that only the dynamic pressure history is shown. Those figures which correspond to plant operating conditions at~which multipletests were run were con struct~d such that the loads represent mean (average) load~. The figures for zero L~P (for those plants whose normaVoperatinAconditiOns include ~

drywell/wetwell ~P) are based upon a sing1~ t~st.~

4.3.22 Revision 2

NEDO 21888 The longitudinal and azimuthal spatial variations are implemented by means of multiplicative factors M and M~. These factors were determined by 7

dividing the pressure measured at a specific location by the integral of the pressure over the submerged portion of the torus shell. Figure 4.3.23 shows the coordinate system used in the calculation of the multipliers. The normalized length, z/Z, is the torus centerline distance from the centerline of the main vent divided by the torus centerline length from the main vent to the center of the nonvent bay.

Table 4.3.21 contains the factors which adjust the magnitude of the average submerged pressure transient to correspond to any submerged posi-tion on the torus shell. Those factors are given for nondimensional posi-tion on the torus shell and time during poo1 swell and are applicable to all Mark I plants. At points on the shell between those where the factor is defined, linear interpolation may be used to determine the appropriate factor... The submerged pressure transient changes only in magnitude at different points on the torus shell; the timing remains the same at all points. The airspace pressure transient is the same at all points on the unwetted portion of the torus shell.

The tozus submerged pressure longitudinal and azimuthal multipliers are applied as follows. If, P (t) = The average submerged pressure magnitude at avg. sub time t, lloc sub (t) = the local submerged pressure magnitude at time time t, M, M = the azimuthal and longitudinal submerged pressure z

multipliers corresponding to the submerged location and time of interest,

Then, P (t)=P (t)xM xM bc. sub avg. sub z 0 4.3.23 Revision 2

NEDO21888 For the load definitions corresponding to normal plant operating conditions, multiple tests were run in the QSTF. The submerged and airspace pressure histories defined in the Plant Unique Load Definitions correspond to the mean values of the loads measured in these multiple tests.

The submerged pressure transients defined above do not include the static pressure on the torus shell resulting from the water in the torus. The water pressure at any submerged location on the torus shell can be calcu lated by:

P = .433 h w

where P = static water pressure at a submerged location on the w

torus shell, psi h = vertical head of water above submerged location on the torus shell, ft.

The static water pressure calculated by the above equation should be included.~

in the submerged pressure transients -~after the~-transients have been adjusted to the desired spatial position on the torus shell.4 Adding the static water pressure to the dynamic submerged pressure trace produces the total submerged pressure transient on the inside of the torus shell in psi relative to the initial wetwell airspace pressure. To put the total submerged pressure transient and the airspace pressure transient in terms of absolute pressure, The initial wetweli airspace pressure should be added to the total submerged and airspace pressure transients.

4.3.24 Revision 2

NEDO21888 Table 4.3.21 TORUS SHELL PRESSURE HISTORY MULTIPLIERS M AND M z e LONGITUDINAL VARIATION, M z

z/k Time 0.0 0.361 0.552 0.895 1.000 Start of Accident 1.0 1.0 1.0 1.0 1.0 Peak Download 0.852 0.925 1.040 1.134 1.134 Zero Download 1.059 0.971 0.990 0.999 0.999 Peak Upload 1.109 0.992 0.996 0.945 0.945 Reduced Upload 1.077 0.979 1.005 0.966 0.966 Zero Upload 1.0 1.0 1.0 1.0 1.0 AZIMUTHAL VARIATION, Me e

165 1500 1350 1200 900 Time 1800 1950 2100 2250 2400 2700 Start of Accident 1.0 1.0 1.0 1.0 1.0 1.0 Peak Download 1.205 1.197 1.083 0.889 0.638 0.638 Zero Download 0.947 0.952 0.940 1.015 1.145 1.145 Peak Upload 0.908 0.935 0.940 1.023 1.186 1.186 Reduced Upload 0.963 0.983 0.9 72 1.010 1.067 1.067 Zero Upload 1.0 1.0 1.0 1.0 1.0 1.0 4.3.25 Revision 2

NEDO21888 Note: Differential pressure corresponding to the downward load margin will be added to the pressure transient for the downward loading phase time period.

40 30 LU Lr

~2Q U) a.

10 0

0 TIME AFTER BREAK INITIATION (misc)

Figure 4.3.21. Typical Average Submerged Torus Pressure History 4.3.26 Revision 2

NEDO21888 Note: Differential pressure corresponding to the upward load margin will be added to the pressure transient for the upward loading phase time period.

40 30 U)

~20 LU a-10 0

0 200 400 600 800 1000 1200 1400 TIME AFTER BREAK INITIATION (misc)

Figure 4.3.22. Typical Torus Airspace Pressure History 4.3.27 Revision 2

NEDO21888

~~- Z/2 2700 900 1500 Figure 4.3.23. Submerged Location on Typical Torus Shell 4.3.28 Revision 2

NEDO2 1888 REFERENCES FOR SECTION 4.3.2 4.3.21 J. M. Humphrey, Mark I ContaThment Program, 1/4 Scale TwoDimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.22 The Electric Power Research Institute, ThreeDimensional Pool Swell Modeling of a Mark I Suppression System, EPRI NP906, October 1978.

4.3.29/4.3.210 Revision 2

NEDO21888 4.3.3 Vent System Impact and Drag As the poo1 surface rises, it impacts structures in its path. The resulting loading condition of primary interest is the impact on the vent system.

The impact phenomenon consists of two events, the impact of the pool on the structure, and the drag on the structure as the poo1 flows past it follow-ing impact. The entire process is referred to simply as impact in this document; however, the load definition includes both the impact and drag portions of the loading transient.

The vent system components which are potentially impacted during poo1 swell include the downcomers, the vent header, and the main vents. Figure 4.3.31 shows the coordinate system which will be used to discuss the vent system. As for the torus vertical loads (Sections 4.3.1 and 4.3.2), only a sector of the vent system will be considered, since the same geometry repeats around the entire torus. A sector includes the vent system from the centerline of a main vent to the position midway to the adjacent main vent.

The sequence of impact of the poo1 on the vent system follows the same pattern in all the plants. The poo1 rises fastest where the downcomers are closely spaced (z/~.= 1.0), and slowest where the downcomers are widely spaced (z/2. = 0.0). Impact first occurs on the angled portion of the downcomers and then on the vent header near z/Z = 1.0. From there the impact point moves along the vent header toward the main vent. At any longitudinal (z/9~) posi-tion on the header, impact first occurs at a particular 0 and spreads out in the 0 direction from that point. For plants without vent header deflectors the initial impact occurs at 0 = 00. For plants with deflectors, the initial impact is plant unique but is generally near 0 = 300.

The load definition for the vent system impact is specified in a different form for each of the three major components of the vent system.

For the downcomers a generic impact pressure transient is specified which is assumed to apply to the entire impacted surface of the downcomer. This load definition bounds the loads for all plants.

4.3.31 Revision 2

NEDO2 1888 The load definition for the vent header is more complex than that for the downcomers. A number of local impact pressure transients, which correspond to the average downcomer spacing impact velocity, are defined for specific locations on the vent header. A plant unique vent header impact velocity distribution curve forms the basis for adjusting the impact pressure transients at points on the vent header where the impact velocity is differ-ent from that of the average downcomer spacing. Finally, definitions of plant unique longitudinal and circumferential time delays specify when the local impact pressure transients occur in time relative to each other.

The impact load definition for the main vent is found by using the procedure outlined in the section on Impact and Drag on Other Structures Above the Pool, Section 4.3.4. The impact load definition for vent header deflectors is specified in Section 4.3.9.

4.3.3.1 Bases and Assumptions The plant unique 1/4 scale 2D pool swell tests (Reference 4.3.31) provide the primary basis for the vent system impact and drag load definition. In the 1/4 scale tests the plant unique vent header, vent header deflector (if applicable), and downcomer design and operating conditions, were accurately simu-lated. In these tests impact pressure transducers were mounted on the lower sur-face of the vent header and downcomers. These pressure transducers measured the impact pressure transient applied to the vent header and downcomers as the p001 first impacted and then flowed past these structures. The 1/4 scale impact pressure transients were scaled to -full scale (using Froudes Law of Com-parison which requires scaling of pressure by the geometric length ratio and scaling of -time by the -square root of the geometric length ratio) and then assigned to positions on the full scale plant header corresponding to the locations of the pressure measurements in the 1/4 scale tests.

The 1/4 scale vent header load -cell measurements were used as the basis for the downcomer impact/drag load definition. The small load measured just before vent header impact was converted to an equivalent pressure on the bottom 50 4ncluded angle -on -the slanted portion of -the downcomers adjacent to the vent header: -

4.3.32 Revision 2

NEDO ~2l888 The 1/4 scale tests were also used as a basis for determining the plant unique vent header impact velocity and circumferential time delay. The impact velocity is necessary for determining the normalized impact velocity distribution and the longitudinal time delays.

The 1/12 scale 3D tests (Reference 4.3.32) were the basis for specifying the longitudinal vent header impact velocity distribution and the longi-tudinal impact time delays. The 1/12 scale tests were performed with a torus and vent system geometry of a typical Mark I plant, which resulted in a longitudinal distribution of the pool swell displacement and velocity corresponding to the typical plant vent system configuration. This dis-tribution is applicable to all the Mark I plants. The 1/12 scale data were interpreted such that the longitudinal impact velocity distribution and longitudinal time delays correspond to the plant unique height of the header above the initial poo1 surface for each plant.

The 1/4 scale 2D tests were performed at both the nominal plant operating conditions, and at the zero drywell/wetwell AP condition if it differed from the nominal operating conditions. In general, multiple tests were run at the operating conditions, while a single test was run at the zero L~P condition. The loads defined for the nominal plant operating conditions correspond to the mean (average) of the loads measured during the.multiple tests. The zero t1P condition loads correspond to the loads measured in the single zero /~P test.

The following assumptions are made when defining the vent system impact and drag loads:

a. The impact pressure is proportional to impact velocity squared.

b. The duration of the impact pressure transient is inversely pro-portional to the impact velocity.

C. The plant unique timing of impact on the vent header and the varia-tion of impact velocity along the vent header z axis is that predicted by the EPRI 1/12 scale 3D tests, with main vent orifices only.

4.3.33 Revision 2

NEDO? 1888

d. The impact loads derived from tests with the relatively rigid QSTF vent header are assumed to provide a conservative load definition for the actual Mark I vent headers.

Assumptions a. and b. imply that the impact impulse (integration of impact pressure over the time of impact) is directly proportional to the impact velocity. In the 1/4 scale 2D generic sensitivity test report (Reference 4.3.33) the peak force, impulse, and duration of the net vent header force 2

are plotted versus V , F, and 1/V respectively. The trends indicated there provide justification for Assumptions a. and b.

The justification for Assumption c. is that the spacing of the downcomer pairs along the header, while not exactly the same in any two plants, is quite similar in all the plants. Therefore, the results from the 1/12 scale tests modeling a typical Mark I plant can be applied generically.

4.3.3.2 Load Definition

a. Downcomers In the 1/4 scale tests the measured impact/drag loading on the downcomers was found to be of very low magnitude. Figure 4.3.32 provides the generic downcomer load definition. The downcomer pressure transient defined in Figure 4.3.32 (which specifies an impact pressure of 8.0 psid) should be applied uniformly over the bottom 500 of the angled portion of the downcomer only, perpendicular to the local downcomer surface, as shown in Figure 4.3.33. The impact pressure transient begins as soon as the rising poo1 reaches the lower end of the angled portion of the downcomer as determined by the pool swell displacement/velocity procedure in Section 4.3.4.

The transient ends at the time of maximum pool swell. The structural analysis for downcomer impact shall either be dynamic, accounting for the virtual mass of water near the submerged portions, or a dynamic load factor of two shall be applied. -~

4.3.34 Revision 2

NEDO21888

b. Vent Header The Plant Unique Load Definitions provide the local impact pressure transients obtained from QSTF measured impact pressures, for each of the Mark I plants, as well as figures which show the locations of the local impact pressure transients. Figure 4.3.34 shows the general form of the vent header local impact pressure transients. The local

4. 3. 34A/B Revision 2

impact pressure transients are given in units of psid relative to the torus airspace pressure at the time of vent header impact. The pressure difference between the inside of the vent header and the torus airspace at the time of vent header impact is specified on a plant unique basis in Table 4.3.31, which is determined from plant unique QSTF test data.

The impact pressure transients in the Plant Unique Load Definitions (PULD) correspond to the average downcomer spacing impact velocity.

Since the impact velocity varies along the length of the header, the local impact pressure transients need to be adjusted for impact velocity. This is accomplished via the figures in the PULD which specify the vent header impact velocity dis~ribution normalized to the average downcomer spacing impact velocity.

For a normalized impact velocity equal to 1.0, no adjustment of the local impact pressure transient is necessary. For other impact velocities both the impact pressure magnitude and the time duration of the transients need to be adjusted according to the following:

P V2 At 1/V where P = local impact pressure magnitude, psi V = normalized impact velocity given in the PULD, ft/sec At time duration of the local impact pressure transient.

A.ll points of the transient should be adjusted relative to the start of the transient, sec.

The Plant Unique Load Definitions also contain the plant unique longitudinal time delays, At, and the plant unique circumferential time delays, At 4.3.35 Revision 2

NEDO21888 The impact pressure transient at any point on the vent header begins at a certain time after the pool first impacts the header. This time is the sum of two quantities: the longitudinal time delay, which accounts for the longitudinal (z) distance from the position of first impact to the point of concern; and the circumferential (e) time delay, which accounts for the time it takes for the point of impact to sweep around the header in the cir-cumferential direction. Therefore, the time of impact at any point on the vent header is specified by:

t.=t +At +At io z 0 where

t. = time of impact at any particular point on the header, sec 2.

= time of first impact on the header as determined by the pool swell displacement/velocity procedure in Section 4.3.4, sec At2 = plant unique longitudinal time delay based on EPRI 1/12 scale 3D main vent orifice tests, for point of interest on the header, sec.

At0 = plant unique circumferential time delay for point of interest on the header, sec.

C. Main Vent Since the main vent is a large structure and is oriented at an angle to the pool surface, it should be subdivided into smaller sections and the impact and drag loads calculated separately on each subdivision.

4.3.36 Revision 2

NEDO21888 The impact and drag loads on the main vent are evaluated in the following manner:

1. Subdivide the submerged portion of the main vent pipe into six equally wide segments (see Figure 4.3.35). If this subdivision results in AL < 0.3D, fewer segments may be used such that AL - O.3D.

2. Determine the velocity and acceleration histories at Points 1 through 7 in Figure 4.3.35 from the QSTF data and appropriate corrections for longitudinal variations along the torus (at Point 7, only the initial impact velocity is required).

3. Using the velocity components normal to the vent pipe, calculate the impact and steady drag pressure using the method in Section 4.3.4 (Cylindrical Structures). At Point 7, only impact force is to be considered.

Using the acceleration components normal to the vent pipe, calculate the acceleration drag pressure using the equation P/7rD\4+f~static buoyancy\ 1 a 2g \144/\ DL /

Where P is the acceleration pressure averaged over the projected a

3 area (psi), p is the density of water (lbm/ft ), D is the diameter

• 2 in feet, V the cross flow acceleration (ft/sec ), g is the conversion 2

constant (lb ft/lb sec ), and (1/144) is the conversion factor for m f

-the pressure units.

4. Sum the pressures due to impact, viscous drag and acceleration drag and multiply by D to obtain force per unit length-at Stations 1 through 7.

5. To obtain a smooth loading history for the main vent as a whole, the linear interpolation method suggested for the vent header deflector in Section 3.5 of Reference 4.3.34 may be used.

4. 3. 36A/B Revision 2

Table 4.3.31 INTERNAL VENT HEADER/TORUS AIRSPACE PRESSURE DIFFERENTIAL AT TIME OF VENT HEADER IHPACT Pressure Differential (psi Plant Operating Al Zero Al

-Brunswick 1,2 11.4

• Browns Ferry 1, 2, 3 5.2 5.8 Cooper 8.9 11.3 Dresden 2,3 7.4 8.9 Duane Arnold 6.1 8.6 Fermi 2 6.5 7.1 Fitzpatrick 6.9 9.9 Hatch 1 6.9 10.1 Hatch 2 9.4

• Hope Creek 1,2 6.3

• Millstone 8.2 12.2 Monticello 11.2 13.6 1 10.5 12.2 Nine Mile Point Oyster Creek 8.3 11.2 Peach Bottom 2,3 10.6 11.0 Pilgrim 9.7 11.4 Quad Cities 1,2 7.4 8.9 Vermont Yankee 7.1 12.7

• Normal operating condition is zero drywell/wetwell pressure differential.

4.3.37 Revision 2

MIDWAY BETWEEN TWO MAIN VENTS MAIN VENT G~

MAIN VENT HEADER B MEASURED IN C-L~i ha uuu H

I C 03 03 03 (TYPI H

0, H

0 ha Figure 4.3.31. Vent System Coordinates

-LU

~ 8.0 w

a-TIME WHEN POOL TIME OF TIME (mined REACHES LOWER MAXIMUM END OF ANGLED POOL SWELL PORTION OF DOWNCOMER Figure 4.3.32. Downcomer Impact and Drag Pressure Transient 4.3.39 Revision 2

NEDO21888 VENT HEADER A

DO WNCOMER lANG LED SECTION) woy DOWNCOMER SECTION AA (VERTICAL SECTION)

Figure 4.3.33. Application of Impact/Drag Pressure Transient to Downcomer 4.3.310 Revision 2

NEDO21888 P

3 - TEST DATA APROXIMATION LU LU a.

1-U 4.

a-

~1 .~5 11 t2 t3 4 15 TIME Figure 4.3.34. Vent Raader Local Impact Pressure Transient 4.3. 311 Revision 2

NEDO21888

£~...H1gheLt pool

• 1eb-atior~

ccl a pODI *~tvitI~n at V a C~ ~ t1*1COr1.~ with a V

• pool 4In~act velocity at sLati:?

a

• pool *ccelerat~on at itaLic Figure 4.3.35. Schematic Diagram Illustrating the Methodology for Main Vent Impact and Drag 4.3.312 Revision 2

NEDO21888 REFERENCES FOR SECTION 4.3.3 4.3.31 J. M. Humphrey, Mark I Containment Program 1/4 Scale TwoDimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.32 The Electric Power Research Institute, ThreeDimensional Pool Swell Modeling of a Mark I Suppression System EPRI NP906, October 1978.

4.3.33 J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: LDR Load Tests Generic Sensitivity, General Electric Company, Report No. NEDO23545, June 1979.

4.3.34 Mark I Containment Vent Header Deflector Load Definition, Task Number 7.3.3, General Electric Company, Report No. NEDO24612, April 1979.

4.3.313/4.3.314 Revision 2

NEDO-2] 888 4.3.4 Impact and Drag on Other Structures Above the Pool During the pool swell transient, the rising pool will impact structures above the initial pool surface and below the maximum pool swell height. The loading on structures above the pool, other than the vent system (which includes the vent header, vent header deflector, and downcomers), is speci-fied in this section.

As the poo1 surface rises during poo1 swell, it impacts structures located within its range of travel. Loads are generated due to both the impact forces and drag forces. The timing and amplitude of the loading on a particular structure depends on the velocity of the poo1 surface as it impacts and flows past the structure. To facilitate the calculation of these impact and drag loads, the first part of this section provides a method for calculating the poo1 swell impact velocity at any point in the torus. The procedures for calculating the impact and drag loads are then given.

Two sets of procedures are given. For calculations made previously using the original set of procedures (4.3.4.3 and 4.3.4.4) the NRC requires an additional 35/. margin in the impulse and drag load immediately following impact. Application of these factors is discussed in Section 4.3.4.6.e.

An alternative set of procedures defined by the NRC acceptance criteria is provided in Section 4.3.4.6.

4.3.4.1 Bases and Assumptions The primary input to the calculation of the pool swell impact and drag loads is the pool swell impact velocity. The pool swell impact velocity calculation is based on analysis of high speed motion picture data obtained during tests performed at the 1/4 Scale 2D Test Facility (QSTF) and at the 1/12 Scale 3D Test Facility described in References 4.3.41 and 4.3.42, respectively.

Plant unique QSTF pool swell tests were performed to obtain 2D pool swell transients corresponding to the average downcomer spacing in the torus.

For each plant the torus and vent system geometry and the plant operating 4.3.41 Revision 2

HEDO21888 conditions were modeled. Downcomer spacing in the test model was scaled to be equivalent to the average downcomer spacing in the plant being tested.

Motion pictures of the 1/12 scale 3D pool swell tests were used to obtain the displacement and velocity profiles along the longitudinal axis (z) of the torus. A single representative geometry was used in conjunction with selected pressurization rates to represent generic plant behavior.

4.3.4la Revision 2

NEDO21888 The assumptions made when defining the pooi swell velocity and displacement behavior are as follows:

a. The pool swell phenomenon is symmetric from sector to sector (a sector is defined in Figure 4.3.41).

b. The normalized pool swell 2D displacement and velocity profiles obtained from the QSTF tests are representative of the displacement and velocity profiles at every longitudinal position in the torus.

c. The longitudinal distribution of displacement and velocity obtained from the 1/12 scale 3D tests is applicable to all Mark I plants.

Reference 4.3.42 discusses the 1/12 scale 3D pool swell tests. These tests indicated that pool swell is symmetric from sector to sector, justifying Assumption a.

The pool swell shape obtained in the QSTF tests (Reference 4.3.41) applies to the average cell spacing. Assumption b. is justified because even in regions with different downeomer spacings, the shape and velocity profiles will not be substantially different.

Assumption c. is based on the fact that the downcomer spacing geometry for the typical plant modeled in the 1/12 scale 3D tests is similar to that in each of the Mark I plants.

The assumptions employed in the load definition for pool swell impact loads are as follows:

a. The impact load is calculated assuming the impacted structure is rigid.

b. The impact pressure is parabolic in time. The duration of impact is calculated based on assuming that impact occurs over the bottom 500 included angle of the impacted structure.

4.3.42 Revision 2

NEDO21888

c. The hydrodynamic mass factor, Kb, is assumed to be 0.2 for all structures of concern.

d. Based on the results reported in Reference 4.3.43, the impact load on gratings is defined as negligible (however, a drag load is specified).

The value of the hydrodynamic mass factor, Kb, was determined by comparing the measured impact impulses on the QSTF vent header during the Part 1 test series (Reference 4.3.44) to the impact calculated using the method explained in this section. K,0 was determined by the comparison. In all cases K.~ was calculated to be less than 0.20; K.0 was conservatively assumed to be 0.20.

The assumptions employed in the load definition for pool swell drag are as follows:

a. Standard drag coefficients ar~ applicable.

b. The maximum velocity .at the (x,z) position of the structure is used for the- standar4 drag calculation, instead of the actual pool swell velocity at structure location (x,yz); it is assumed that such a drag load bounds the sum of the actual standard drag and acceleration.

drag at the same location at all times. ~

4.3.4.2 Determination of Pool Swell Displacement and Velocity.

To calculate the pool swell impact and drag loads on other structures above the pool, the pool swell displacement and velocity transients must be known at the (x,z) location of the structure of interest (Figure 4.3.41 shows the coordi-nate system used for this analysis). This subsection describes a procedure for calculating the pool swell displacement and velocity transient at any point in the torus.

4..3.43 Revision 2

NEDO21888 Figure 4.3.42 is a typical plot of the pool surface displacement longitudinal distribution. This describes the pool swell displacement at any z position normalized to the displacement corresponding to the average downcomer spacing.

In other words, Figure 4.3.42 is a plot of L~y(z,t)

~y(t) where Ay(z,t) = the pool displacement at time t above the initial pool surface

~y5(t) = the same parameter corresponding to the average downcomer spacing.

Figure 4.3.43 is a plot of a typical plant pool surface displacement in the xy plane, ~~y(x,t), for the average downcomer spacing. The pool swell dis-placement at any (x,z) location in the torus and at any time t is given by the product

~y5(z,t A procedure similar to that indicated above is used to find the pool swell velocity transient at any point in the torus. Figure 4.3.44 shows the typical longitudinal velocity distribution and Figure 4.3.45 shows a typical plant velocity transient in the (x,y) plant for the average downcomer spacing.

The pool swell velocity at any (x,z) location in the torus and at any time t is given by the product v(z,t) ~ v(x,t) 4.3.44 Revision 2

NEDO21888 where v(z,t) is from Figure 4.3.44 v(t) v(x,t) is from Figure 4.3.45 for a typical plant.

I By using Figures 4.3.42 and 4.3.44 along with the appropriate plant unique...

figures corresponding to Figures 4.3.43 and 4.3.45, the pool swell displace-ment and velocity transients can be constructed for any (x,z) location in a Mark I plant.

4.3.4.3 Load Definition for Impact The procedure for calculating the pool swell impact loads on structures (other than the vent system) above the pool also uses the pool swell impact velocity as a primary input.

During impact each structure must stop or redirect the momentum of the water which impacts it. An upper bound value of this momentum is given by the product of the hydrodynamic mass of the structure and the impact velocity.

The momentum arrested during impact is also equal to the impact impulse.

In fact, the impulse is less than the product of hydrodynamic mass and impact velocity because much of the impacting momentum is redirected rather than totally arrested. In equation form the above becomes I K.0 Mh V where I impact impulse, lbfsec N,0 = hydrodynainic mass of impacted structure, lbm (Table 4.3.41)

V = impact velocity normal to the structure surface being n

impacted, ft/sec.

4.3.45 Revision 2

NEDO21888 lbmft g = gravitational constant = 32.2 *2 lbfsec Kb = proportionality factor = 0.2.

The impact impulse can be expressed in terms of force per unit projected area of the structure by dividing by the projected area of the structure. In addi-tion, the impact impulse is given as a ratio of impulse per impact velocity by dividing by the impact velocity. The result is I Kb~

VAg 144 n c where I = impact impulse per unit area, psisec A = projected area of the structure, ft 2 The duration of impact is calculated assuming impact occurs over the bottom 500 included angle of the structure. If the structures cross section is not circular, a circumscribing circle is assumed for this calculation. The impact duration is assumed to be equal to the time necessary for the impact velocity to travel through the 50 included angle, i.e.,

At = d/v n

where d = distance corresponding to 500 included, angle, ft (see Figure 4.3.46)

At = impact duration, sec.

The impact load transient is assumed to be parabolic in time. Together with the definition of the impulse and duration, this completes the definition of~..

the impact load on the structure.

4.3.46 Revision 2

NEDO2 1888 4.3.4.4 Load Definition For Drag The pool swell drag load on other structures, which follows the impact load and continues until the time of maximum pool swell, is calculated using the standard drag equation and the maximum pool swell velocity at the (x,z) location on the structure of interest. The maximum pool swell velocity is found using the procedure outlined for determining the pool swell displace-ment/velocity characteristics (Section 4.3.42). The drag load is then given by CD 2 F =pAV D 2g c n where F = drag load on structure, lbf D

p = density of pool water 62.4 lbm/ft3 A = projected area of structure, ft2 V = impact velocity normal to the impacted surface, ft/sec n

C = standard drag coefficient for structure of interest D

lbmft Sc = gravitational constant = 32.2 lbfsec2 4.3.4.5 Application of Impact and Drag Loads on Structure The impact and drag pressure transient should be distributed uniformly over the surface. The load should be applied in the upward direction most critical to the structural evaluation of the structure within the specified load range.

The range of directions which should be considered include a variation of +/-100 from the upward vertical.

4.3.47 Revision 2

N EDO21888 When the impact loading is primarily impulsive and calculations have already been performed using the methodology presented in Sections 4.3.4.3 and 4.3.4.4, (using the impulse equation on page 4.3.45 with Kb 0.2 for cylinders and K..~ = 0.62 for flat structures) simple adjustments may be made to provide required margins. Under these conditions, a parabolic pulse shape may be used provided corrections are made to account for the 35% margin in the impulse and with additional corrections for the drag force immediately following impact.

For structures with a natural frequency less than 30 Hz, loading can be treated impulsively (i.e., independent of pulse shape) when the conditions fall into the region above the straight line shown in Figure 4.3.47.

The following corrections must be applied to the previously calculated stresses:

a. The calculated stresses will first be multiplied by a factor of 1.35 to account for the data scatter in the impulse data.

b. The calculated stresses will then be multiplied by an additional factor to account for the presence of drag following the impact.

This factor is determined as follows:

(1) Calculate the drag pressure, (P ) as described in drag Sections 4.3.4.6.a and 4.3.4.6.b.

(2) Form the ratio:

P /P drag max where P is the amplitude of the parabolic pulse used in the max original stress analysis multiplied by 1.35.

(3) Determine the dynamic load factor (DLF) from Figure 4.3.48 corresponding to the two cases: (1) parabolic pulse without drag and (2) parabolic pulse followed by drag.

4.3.48 Revision 2

NEDO2 1888 (4) Multiply the calculated stress by the factor DLFwith drag /DLF w/o drag 4.3.4.6 Alternate Procedure for Pool Swell Impact and Drag on Other Internal Structures The impact and drag loads for internal structures above the suppression pool (except the vent header, downcomers, and vent header deflectors) shall be such that the structures are classified as either cylindrical (e.g.,

pipes), exposed flat surfaces (e.g., I beams), or gratings. The following load specifications for each of the three structural classifica-tions shall be used. Noncylindrical structures can be conservatively defined as equivalent flatsurfaced structures. However, if such an approach is too conservative, a similar technique may be used with an impact data base which is appropriate for the structural configuration of interest. The longitudinal velocity distribution shall be based on the main vent EPRI pool swell tests, using Figures 4.3.42 and 4.3.44.

a. Cylindrical Structures For cylindrical structures, the pressure transient which occurs upon water impact and subsequent drag is depicted in Figure 4.3.49.

The parameters in Figure 4.3.49 shall be defined as follows:

(1) The maximum pressure of impact P will be determined by max P

max = 144 g c where P = the maximum pressure averaged over the projected max area (psi),

p = the density of water (lbm/ft 3 ),

4. 3. 48A Revision 2

NEDO21888 V = the impact velocity (ft/sec), and 2

gc = gravitational constant (ft lbm/lbf sec (2) The hydrodynamic mass per unit area for impact loading shall be obtained from the correlation (cylindrical target) depicted by Figure 68 in Reference 4.3.43. A margin of 35% will be added to this value to account for data scatter.

(3) The impulse of impact per unit area shall be determined by:

MR ___

I p A (l4~8~)

where I = the impulse per unit area (ps~sec),

2 MR/A = the hydrodynamic mass per unit area (lbm/ft ),

and V = the impact velocity (ft/sec).

(4) The pulse duration will be determined from the following equation:

t = 21 p /~ max (5) The pressure due to drag following impact shall be determined by:

C pV max D = 144g 2 cI where P = the average drag pressure acting on the projected D

area of the target (psi),

C = the drag coefficient as defined by Figure 4.3.410, D

4.3.48B RevisiOn 2

NEDO-21888 3

p = the density of water (lbm/ft ), and V = the maximum vertical velocity attained by the max pool.

(Note that different velocities are used for the determination of impact and drag loads.)

b. FlatSurface Structures For flatsurface structures, the pressure transient which occurs upon water impact and subsequent drag is depicted in Figure 4.3.411.

The parameters in Figure 4.3.411 shall be defined as follows:

(1) The pulse duration (T) is specified as a function of the impact velocity:

T = 0.0016W for V < 7 ft/sec 0.011 W

= V for V > 7 ft/sec where W = the width of the flat surface (feet) and V = the impact velocity (ft/sec).

(2) The pressure due to drag following impact shall be determined by:

C /pV2 D 2 \l44g J where P = the average drag pressure acting on the frontal D

area of the structure (psi),

4.3.48C Revision 2

NEDO 21888 C = the drag coefficient (C = 2 flat strips D D normal to flow, independent of Reynolds number),

3 p = the density of water (lbm/ft ), and V = the maximum vertical velocity attained by the pool.

max (3) The hydrodynamic mass per unit area for impact loading shall be obtained from the correlation (flat targets) in Figure 68 in Reference 4.3.43. A margin of 35% shall be added to this value to account for data scatter.

(4) The impulse of impact per unit area shall be determined by:

- MR(v \

A \144 g/

where I = the impulse per unit area (psisec),

2 MR/A = the hydrodynamic mass per unit area (lbm/ft ),

and V = the impact velocity (ft/sec).

(5) The maximum pressure (P max

) shall be calculated from the impulse per unit area and the drag pressure as follows:

21 P

max i D

c. Gratings The static drag load on gratings in the poo1 swell zone of the wetwell shall be .calculated for gratings with open areas greater than or equal to 60% by forming the product of the pressure dif-ferential (Figure 4.3.412) and the total grating area (not only the 4.3.48D Revision 2

NEDO21888 area of the metal bars). The pressure differential curve in Figure 4.3.412 is based on a velocity of 40 ft/sec. If the maxi-mum pool velocity in the area where gratings are located differs from 40 ft/sec, the force on the grating will be calculated as follows:

F = APxA grating max 5 To account for the dynamic nature of the initial loading, the load shall be increased by a multiplier given by:

F /D = 1 + [1 + (0.0064 Wf) 2 I 1/2 ; for Wf < 2000 in/sec.

SE where F = static equivalent load,. lb SE f W = width of grating bars, inches f = natural frequency of lowest mode, Rz and D = static drag load, lbf If Wf > 2000 in./sec (not expected for gratings) the force on the bars of the gratings will be calculated by the method outlined above for flatsurfaced structures.

d. Load Application These load specifications correspond to impact on rigid struc-tures. When performing the structural dynamic analysis, the rigid body impact loads shall be applied; however, the mass of the impacted structure shall be adjusted by adding the hydrodynamic mass of impact, except for the gratings. The value of the hydrodynamic mass shall be obtained from the appropriate correlation in Figure 68 in Reference 4.3.43.

4. 3. 48E Revision 2

NEDO21888 In performing the structural dynamic analysis, the drag following impact (as shown in Figures 4.3.49 and 4.3.411) shall be included in the forcing function. The transient calculation shall be continued until the maximum stress in the structure has been identified.

4. 3. 48F Revision 2

NEDO21888 Table 4.3.41 HYDRODYNAMIC MASS AND ACCELERATION DRAG VOLUMES FOR TWODIMENSIONAL STRUCTURAL CONPONENTS (LENGTH L FOR ALL STRUCTURES)

SECTION THROUGH ACCELERATION DRAG BODY AND UNIFORM HYDRODYNAMIC MASS BODY VOLUME VA FLOW DIRECTION (PATTON, 19651 2L 2,rR2L CIRCLE piR ELLIPSE pza2L ,rala+bI L Qb

-I ELLIPSE p,rb2L wbla+b) L fl-i PLATE wa2L I

a/b LIII pira2L aLI4b+al RECTANGLE ~4--~m~

10 1.14 pua2L aLI4b+1 14,ra) 5 1.21 aL(4b+1 .2lira) 2 1.36 pra2L aL(4b+1 .36wa) 1 1.51 p,ra2L aL(4b+1 .51 Wa) 1/2 1.70 pga2L aL(4b+?.70,ra) 1/5 1.96 pire2L aL(4b+1 .9Bwal 1/10 2.23 pwa2L aL(4b+2.23wa1 a/b 2 0.85 pna2L aL(2b+0.SSwa)

DIAMOND ~ I 0.76 p,ra2L aL(2b+0.76ira) 1/2 0.67 pira2L aL(2b+0.67wa1 1 ~5 0.61 pza2L aL(2b+0.Blwo) 4.3.49 Revision 2

NEDO21888 Table 4.3.41 (Continued) ACCELERATION DRAG HYDRODYNAMIC MASS AND ACCELERATION DRAG VOLtTh~S FOR TWODIMENS TONAL STRUCTURAL CONPONENTS (LENGTh L FOR ALL STRUCTURES)

SECTION THROUGH BODY AND UNIFORM HYDRODYNAMIC MASS BODY VOLUME VA FLOW DIRECTION (PATTON. 1965) a/c2.6, b/c3.6 2L [2.1lwa2+2c(2abc)J L I-BEAM 2.11 psra Icr~

a/c2.6.b/c3.6 2.11 pwa2L [2.11za2+2c(2a~tbc)] L I-BEAM C

~~2b-b.J HYDRODYNAMIC MASS AND ACCELERATION DRAG VOLUMES FOR THREE-DIMENSIONAL STRUCTURES BODY AND FLOW HYDRODYNAMIC MASS ACCELERATION DRAG DESCR IPTION DIRECTION (PATTON, 1965) VOLUME VA CIRCULAR S/3R3 B/3R3 DISK b/a pw/6be2 till.... b ELLIPTICAL DISK 3 0.9 puj6b2 O.9w/6ba2 2 0.826 pw/6ba2 0.826w/6ba2 1.5 0.746 pw/Bba2 0.748w/6ba2 1.0 0.637 pw6/b2 0.637w~6ba2 Revision 2 4 .3. 410

NEDO21888 Table 4.3.41 (Continued)

HYDRODYNAMIC MASS AND ACCELERATION DRAG VOLUMES FOR TWODIMENSIONAL STRUCTURAL COMPONENTS (LENGTh L FOR ALL STRUCTURES)

BODY AND FLOW HYDRODYNAMIC MASS ACCELERATION DRAG DESCRIPTION DIRECTION (PATTON, 1965) VOLUME VA b/a RECTANGULAR PLATE 4- b 1.5 0.478 0.680 pwl4a 2b pw/4a2b 0.479,r/4a2b 0.680w/4a2b

+

TRIANGULAR 0.840 pw/4a2b 0.840w/4a2b 2

0.953 pw/4a2b 0.953 ,r/4a2b 2.5 pir/4a2b ,r/4a2b 3

-4a~

pw/4~2b w/4a2b PLATE pa3 (tan a3 dane 3/2 3,r SPHERE QR p 2/3 R3 2wR3

4. 3. 411 Revision 2

V

+

C-L~3 z

H ha ha H 03 03 03 (U

4 I-.

0, H

0 I-a Figure 4.3.41. Typical Mark I Torus Sector

NEDO 21888 1 .3 1.2 11

. l.a 2

w w

U

-J B.

0.9 U

IL-0 w

N

-j

~ 0.8 0

2 0.7 0.6 0.5 0 0.2 04 05 0.8 1.0 1.2 NORMALIZED POSITION ON VENT HEADER, WQ)

Figure 4.3.42. Typical Pool Surface Displacement Longitudinal Distribution 4.3.413 Revision 2

NEDO-21888 8.0 7.0 6.0 5.0 I-z w

uJ U

-J Cd 4.0 0

-J U

w 3.0 2.0 1.0 0

0 0.2 0.4 0.5 0.8 1.0 HORIZONTAL DISTANCE (x/R)

Figure 4.3.43. Typical Plant Pool Surface Displacement in XY Plane 4.3.414 Revision 2

NEDO- 21888 1.3 1.2 1.1 1.0 is I-C) 0.9 0

-J a

a U

C U.

Co 0 0.8 a

N

-A C

0 2

0.7 0.6 0.5 0 0.2 0.4 0.6 0.8 1.0 1.2 NORMALIZED POSITION ON VENT HEADER tzIQ)

Figure 4.3.44. Typical Pool Surface Velocity Longitudinal Distribution 4 .3.415 Revision 2

NEDO21888 60 50 40 30 U

C LU

= 0.455

= 0.418 20 t = 0.493 t = 0.531 t- 0.380 t = 0.342 t = 0.569 10

- 0.304

= 0.256 t- 0.229 t 0.191 0

0 0.2 0.4 0.6 0.8 1.0 1.2 HORIZONTAL DISTANCE (x/R)

Figure 4.3.45. Typical Plant Velocity Transient in XY Plane 4.3.416 Revision 2

NF.DO- 21888 STRUCTURE OF Figure 4.3.46. Calculation of Distance Over Which Impulse Acts 4.3.417 Revision 2

NEDO21888 TYPICAL IMPACT STRUCTURES K~-~ ~q U

A a

I-U C

a.

W (feet)

Figure 4.3.4-7. Impulsive Impact Loading Region for Structures with a Natural Frequency Less Than 30 Hz 4.3.418 Revision 2

NEDO21888 a

CO CO a

a.

T TIME 2

IL

-A 0

0 I-U C

U.

0 C

0

.J U

iC z

0 0 0.1 0.2 0.3 0.4 0.5 0

IITN Figure 4.3.48. Effect of Drag Following a Parabolic Impact Pulse 4.3.418A Revision 2

NEDO21888

~max a

Co Co a

a.

I-Li C

a.

a C:,

C a

C Ip

7. TIME Figure 4.3.49. Pulse Shape for Water Impact on Cylindrical Targets

4. 3. 41BB Revision 2

2.0 1.5 IS z t.1 1.) 0 0

-IS Co 1.0 N)

H 03 03 0 03 C

0.5 0

10 1.0 100 2IDg)

FROUDE NUMBER (V (U

4 H

0, H

0 ha Figure 4.3.410. Drag Coefficient for Cylinders Following Impact

~max w

Co Co a

a.

I-C.)

C a.

a C

a C

7. TIME Figure 4.3.411. Pulse Shape for Water Impact on Flat Targets 4.3.418D Revision 2

NEDO21888 (N

a Co

-A C

I.-

z w

~ 10 w

IL II.

0.7 0.8 OPEN AREA FRACTION Figure 4.3.412. Pressure Drop Due to Flow Across Gratings 4.3.418E/F Revision 2

NEDO21888 REFERENCES FOR SECTION 4.3.4 4.3.41 J. M. Humphrey, Mark I Containment Program 1/4 Scale TwoDimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.42 The Electric Power Research Institute, ThreeDimensional Pool Swell Modeling of a Mark I Suppression System, EPRI NP906, October 1978.

4.3.43 T. R. McIntyre, et al., Mark III Conf.irmatory Test Program OneThird Scale Pool Swell Impact Tests Test Series 5805, General Electric Company, Report No. NEDO13426, August 1975.

4.3.44 D. L. Galyardt, et al., Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Scaling Evaluation, General Electric Company, Report No. NEDO21627, January 1978.

4.3.419/4.3.420 Revision 2

NEDO21888 4.3.5 Froth Impingement Loads During the LOCA pool swell transient one of the loading conditions in the torus is due to froth impingement. This section describes these loads.

Froth is an airwater mixture which rises above the pool surface and may impinge on the torus walls and structures within the torus airspace. After-ward the froth will fall back, creating froth fallhack loads. There are two mechanisms by which froth may be generated:

a. As the rising pool strikes the bottom of the vent header and/or the vent header protection device, froth is formed which travels upward and to both sides of the vent header.

b. A portion of the water above the expanding air bubble becomes detached from the bulk pool; this water is influenced only by its own inertia and gravity. The water breaks into froth which rises up into the air space beyond the maximum bulk pool swell height.

4.3.5.1 Bases and Assumptions The basis for the froth impingement load definition is motionpicture data obtained from experimental pool swell tests performed on the 1/4 scale two dimensional test facility (QSTF) (Reference 4.3.51). Although the QSTF tests are not scaled after bubble breakthrough, they do give an approximation of the froth formation phenomenon.

The following assumptions were made when defining the froth impingement loads:

a. No froth velocity is assumed in the torus longitudinal direction.

b. The first froth formation mechanism described in Section 4.3.5 results in negligible loads on the torus shell.

4.3.51 Revision 2

NEDG2 1888

c. For the first froth formation mechanism described in Section 4.3.5, the froth density is assumed to be 20% of full water density for structures with maximum cross section dimension of less than one foot and a proportionally lower density for structures greater than one foot.

For the second froth formation mechanism described in Section 4.3.5, the froth density is assumed to be 100% of full water density for structures or sections of structures less than or equal to 1 ft in all linear dimensions, and 25% water density for structures larger than 1 ft. The exception to this are structures located above the vent header, for which the froth density is assumed to be 10% of full water density.

Ten and twentyfive percent of full water density is also used for the froth fall back density for structures directly above and to the side of the vent header, respectively.

Assumption a. is based on analysis of the 1/12 scale 3D pool swell test movies (Reference 4.3.52), which showed little froth velocity in the torus longitudinal direction.

Assumption b. concerns the froth which sprays off the vent header as the pool impacts that structure. This phenomenon was noted in the QSTF movies (Refer ence 4.3.51). The froth spray which forms is of relatively low density; if this froth spray reaches the torus wall, it will have diffused to such an extent so as to produce a negligible load on the torus shell.

Assumption c. addresses the froth density. As discussed above, the froth density of the first froth formation mechanism appears relatively small in the QSTF movies. For small structures Assumption c. conservatively assumes 100% water density for the second froth formation mechanism, and a more realistic 25%

water density for larger structures. Structures above the vent header are shielded from froth (this was indicated in the QSTF movies, Reference 4.3.51) and therefore a 10% water density for the froth is asswned in this region.

4.3.5.2 Load Definition Procedure Observation of the QSTF motionpictures indicated two predominant mechanisms for froth generation and also identified the regions where these loads apply.

4.3.52 Revision 2

NEDO21888 Two regions of the torus are identified such that within each region the loading is the result of one of the two mutually exclusive froth formation mechanisms.

Region I, as indicated in Figure 4.3.51, is the region where loading results from the poo1 surface impacting the vent header and/or deflector. Region II, as indicated in Figure 4.3.52, is the region where loading results from froth formed when the water above the rising bubble separates from the bulk pool and rises into the airspace.

For those structures in Region I, the froth impingement pressure is calculated using the following equation which includes deceleration due to gravity: [

((2.5 V) 2 2gH)

Pf = x144 4.3.52A/B Revision 2

NEDO21888 where H = vertical elevation difference between bottom of the structure and 450 tangent on the bottom of the header.

P = froth impingement pressure, psi.

f 3 p = froth density. Equals 20% water density, lbm/ft for X < 1 ft and Pf = (0.2/X) for X > 1.0 ft; X is structure maximum cross sectional dimension V = maximum pool surface velocity prior to impact on the vent header, ft/sec.

= 32.2 lbmft/lbfsec2 g = 32.2 ft/sec2 Figure 4.3.51 shows a typical structure and the specified direction of application of the froth impingement loading within Region I. The load shall be applied in the direction most critical to the structure within the 90 degree sector bounded by the horizontal opposite the vent header and the vertical.

Figure 4.3.53 shows the time history assumed for the froth load.

Alternately, the froth source velocity, mean jet angle, and froth density in Region I may be derived from a detailed analysis of the QSTF plant specific highspeed films. Individual movie frames shall be analyzed to determine the mean angle value and uncertainty limits for the source velocity, wave jet width (wf), the angle of the wave jet relative to the horizontal (e), and the impingement pulse time (t). The mass of the froth jet is then given by:

mh/2 mf

- (;~) sin where mf = the mass of the wave jet, mh = the hydrodynainic mass associated with the combined vent header deflector and vent header impact, 4.3.53 Revjsion 2

N EDO21888 Xf = the froth source velocity, and V. = the pool velocity just prior to vent header impact.

2.

The froth density (p ) at any point in Region I is then given by:

f m

pf = f wVT ff where L is the length of the vent header and wf varies to account for spreading of the froth jet with increasing distance from the source:

As before, the vertical component of the source velocity shall be decelerated to the elevation of the target structure to obtain the froth impingement velocity. The load shall be applied in the direction most critical to the structure within the 900 sector defined in Figure 4.3.51.

Uncertainty limits for each parameter shall be applied to assure a con-servative load specification. In applying the uncertainty limits,, it must be noted that the source velocity derived from the leading edge of the wave jet will be higher than the spatial and temporal mean froth velocity, and could lead to a nonconservative estimate of the froth density. Consequently, a separate, lowerbound froth source velocity must be used to determine the froth density.

For those structures in Region II the froth impingement velocity, Vf~ is cal-culated assuming the froth travels vertically to the structure of concern under the influence of gravity only. The froth velocity is based on a source velocity equal to the maximum pool surface velocity directly beneath the structure under consideration which is corrected for subsequent deceleration from the elevation of the maximum velocity.

4.3. 53A Revision 2

NEDO-? ~888 This velocity is determined by the procedure described in Section 4.3.4.2.

The froth impingement pressure is then calculated by

= flf Xf f g x144 where Vf = froth impingement velocity, ft/sec Pf = froth density, determined by the criteria described in Section 4.3.51, lbm/ft3 The characteristics of the loading time history are shown in Figure 4.3.54.

At is the time needed for the froth to rise from the maximum poo1 swell f

height to the structure. Figure 4.3.55 specifies the range of possible directions of load application for structures in Region II. The same range of directions of application shown in Figure 4.3.55 for the torus (x,y) plane

(+/-450 from vertical) should be considered for the torus (y,z) plane. The calculated pressure should be applied to the structure of concern as a rectangular pulse 100 milliseconds in length in the direction most critical to the capability of the structure.

The froth fallback pressure is calculated by 2

= Pff Vff ff 144 g where Pff = froth fallback pressure, psi Vff = froth fallback velocity, ft/sec Pff = froth fallback density = 25% of full water density except for the region directly above the vent header, where it equals 10% of full water density, lbm/ft3.

4.3.5-4 Revision 2

NEDO2 888 The froth faliback velocity is calculated by allowing the froth to freefall from the height of the upper torus shell directly above the subject struc-ture. The froth fallback pressure is applied uniformly to the upper pro-jected area of the structure of concern in the direction most critical to the behavior of the structure. The froth fallback is specified to start when the froth impingement load ends, and lasts for 1.0 seconds. Figure 4.3.56 indicates the directions of application which should be considered. The same range of directions of application shown in Figure 4.3.56 for the torus (x,y) plane (+/-450 from vertical) should be *considered for the torus (y,z) plane.

~*3.55/4.3.56 Revision 2

NEDO-21888 Figure 4.3.51. Definition of Froth Impingement Region I 4.3.57 Revision 2

NEDO21888 R

MAXIMUM POOL SWELL PROFILE Figure 4.3.5-2. Definition of Froth Impingement Region II 4.3.58 Revision 2

NEDO-2l888 12 a

CO Pf CO a

a.

80 msac TIME tn5ec(

TIME OF VENT HEADER IMPACT Figure 4.3.53. Froth Loading Ristory Region I 4.3.59 Revision 2

NEDO 21888 a

Pt CO CO a

a.

100 meec b.

TIME OF MAX IMUM POOL SWELL

+ ~tf

-I TIME Irniecl Figure 4.3.5-4. Froth Loading History Region II 4.3.510 Revision 2

NEDO21888 TORUS Figure 4.3.5-5. Froth Impingement Region II Possible Directions of Load Application 4.3.511 Revision 2

NEDO-21888 TORUS STRUCTURE Figure 4.3.56. Possible Directions of Froth Fallback Load Application 4.3.512 Revision 2

NEDO 21888 REFERENCES FOR SECTION 4.3.5 4.3.51 J. M. Humphrey, Mark I Containment Program 1/4 Scale TwoDimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO-21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.52 The Electric Power Research Ins?itute, ThreeDimensional Pool Swell Modeling of a Mark I Suppression System, EPRI NP906, October 1978.

4.3.513/4.3.514 Revision 2

NEDO 21888 4.3.6 Pool Fallback Loads Following the pool swell transient, the pool water falls back to its original level and in the process generates fallback loads. After the pool surface has reached its maximum height due to pool swell, it falls back under the influence of gravity, creating drag loads on structures inside the torus shell which are between the maximum bulk pool swell height and the downcomer exit level. The fallback load starts as soon as the pool reaches its maximum height and ends when the pool surface falls past the structure of concern. The fallback load on the torus shell is included in the definition of the torus veritcal load due to pool swell (Sections 4.3.1 and 4.3.2).

4.3.6.1 Bases and Assumptions The basis for the pool fallback load definition is the pool swell data obtained from analysis of movies from the QSTF tests described in Reference 4.3.61 and from the 1/12 scale 3D tests described in Reference 4.3.62.

The procedure for determining fallback loads utilizes the following major assumptions:

a. The fallback load on structures below the downoomer exit level is negligible.

b. Fallback loads on the vent header, downoomers, and main vents are considered negligible.

c. The drag force is the sum of standard drag (proportional to velocity squared) and acceleration drag (proportional to acceleration).

d. The standard drag forces on circular cylinders are computed by using a conservative drag coefficient of C = 1.2, independent of the Rey-d nolds number.

4.3.61 Revision 2

e. Drag forces on structures with sharp corners (e.g., rec-tangular and I beams) are computed by considering forces on an equivalent cylinder of diameter Deq = -7 V L where L is the maximum transverse dimension and is defined as max the diameter of a circumscribed cylinder about the cross section of the structure.

f. Long slender structures are modeled in segments of length (L),

which do not exceed its diameter CD or D ). The total force eq on the structure is obtained by integrating the individual segment force over the entire length of the long structure.

Longer segments are used as long as the equivalent uniform f low velocity and acceleration are evaluated conservatively for every point on the segment.

g. Interference effects due to the proximity of solid boundaries are considered for each segment that has its center less than 1.5 diameters from the boundary. Multipliers are needed to increase both the acceleration and the standard drag. Similarly, multipliers are needed to include the interference effects be-tween neighboring structures where the centers of the segments are separated by less than 3D (D (D1 + D9/2, the average diameter of the two structures).

For structures near walls, the multiplier (1 + A ) sh~il be w

used to increase the acceleration drag and the multiplier (1 + D ) shall be used to increase the standard drag. Bounding w

expressions for A and D are given below as functions of w w Xw CXw = r/D - 1/2, where r is the distance from the segment center to the boundary and D is the diameter of the structure):

4.3.6-la Revision 2

NEDO21888 0.05<X <1.0 Aw 0*05/X~,

V DV 0.12/X V Xw z 0.05 AV 1.0 DV 2.4 For structures with neighbors that are less than 3D away and within 300 of being pa.ra.Llel, the multiplier Cl+A1) shall be used to increase used to the acceleration drag and the multiplier Cl+D I ) shall be and are increase the standard drag. Bounding expressions for A1 5- 1, where is the given below as functions of X1 CX1 r12/

distance between segment centers):

0.05 ~ 2.0 A 0.2 (D 2

I D2 DI 0.2/X I where D is the diameter of the structure under consideration and 1

D 2 is the diameter of the neighbor. If more than one neighbor must

~e considered, the and values may be s~ed over the neighbor structures. For X I less than 0.05, the two neighbor structures shall be considered as an effective single structure.

The effects of wall proximity and neighbor structures may be super-imposed in order to c~1pute overall multipliers as follows:

Cl + A + Iklk A) w Cl+D +ID k 1k

)

4.3.6lb Revision 2

NEDO218 88 4.3.6.2 Load Definition Procedure Fallback loads are applied to the top surfaces of the structures affected.

If the top surface area of the structure of concern is large, the area should be subdivided and the fallback loads calculated individually for each smaller area. As a guideline, if any linear dimension of a surface is greater than 1 foot, the area should be divided into smaller areas.

The procedure requires determining the maximum pool swell height above the top surface of the structure. The pool swell displacement/velocity procedure described in Section 4.3.4 is used for this. Freefall of the bulk fluid from this height produces both standard drag and acceleration drag, with the total drag given by the sum.

The standard drag on the structure of length L is given by F C.DL 5 JJeq 2g C

where Deg is the equivalent diameter. For cylinders, Deq is simply the diameter, while for structures with sharp corners (.e.g., rectanales and I b~ins) t~ must be eq computed by considering forces on an equivalent cylinder of diameter Deq = J~ ~.max where I.~ is the maximum transverse dimension. Itrnax is defined as the diameter of a circumscribed cylinder about the crosssection of the structure Cfor example, I~uax = Ca2 + b2)1/2for a rectangular crosssection of s2.de a and b).

The acceleration drag is given by FA = £Ya

~c ~

dt where VA is the acceleration volume, as given in Table I of Reference 4.3.63.

For free fallback, du equals the gravitational acceleration, g.

dt To provide sufficient conservatism for fallback densities greater than 25%, the minimum value of CD should be taken as 1.2 Ci.e., CD > 1.2).

4. 3.62 Revision 2

The fallback load is applied uniformly over the upper projected surface of the structure in the direction most critical to the behavior of the structure. Figure 4.3.6-1 indicates the possible directions of applica-tion to be considered. This range (~ 450 from the vertical) applies to the torus Cy,z) plane as well as the torus (x,y)plane shown in Figure 4.3.61.

4 .3.62a Revision 2

NEDO21888 TORUS.

Figure 4.3.61. Possible Directions of Fallback Load Application 4.3.63/4.3.64 Revision 2

NEDO21888 REFERENCES FOR SECTION 4.3.6 4.3.61 J. M. Humphrey, Mark I Containment Program 1/4 Scale 2D Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.62 The Electric Power Research Institute, ThreeDimensional Pool Swell Modeling of a Mark I Suppression System, EPRI NP906, October 1978.

4.3.63 F. J. Moody, Analytical Model for Estimating Drag Forces on Rigid Submerged Structures Caused by LOCA and Safety Relief Valve Ramshead Air Discharges, General Electric Company, Report No. NEDO21471, September 1977.

4. 3. 65/4. 3. 66 Revision 2

NEDO2 1888 4.3.7 LOCA Jet Load As the drywell pressurizes during a postulated LOCA, the water slug (if any) initially standing in the submerged portion of each downcomer is accelerat&

downward into the suppression pool. As the water slug enters the pool, it forms a jet which could potentially load structures which are intercepted by the discharge. Forces due to the pool acceleration and velocity induced by the advancing jet front are also created.

described in this section.

The calculation of these loads is I Jet 1o~ding affects structures which are enclosed by the jet boundaries and lasts from the time that the jet first reaches the structure until the time when the last particle of the water slug passes the structure. Pool notion can create loads on structures which are within the region of motion for the duration of the water jet.

4.3.7.1 Load Definition I

The calculation procedure described to obtain LOCA jet loads is based on both expe:2imental data obtaii~ed from tests performed at the 1/4 Scale 2D Test Facility (QSTF) (Reference 4.3.72) and also on an analytical model described in Reference 4.3.71.

Plant unicue downcomer clearing information was obtained experimentally during the QSTF testing in the form of LOCA jet fluid velocity and acceleration histories.

The r.ajor assumptions included in the methodology are as follows:

1. When a structure is engulfed in the jet path, the force on the structure can be calculated using a standard velocitysquared drag equation. If a structure partially or fully intercepts the jet, a momentum balance is used to define a proportionality factor to be used in place of the standard drag coefficient.

2. The jet is completely dissipated when the last fluid particle leaving the downcomer catches up to the front end of the jet; 4.3.71 Revision 2

NEDO2 1888 the dissipated jet provides no more drag.

3. Pool velocity and acceleration are induced by the advancing water jet, inducing both standard velocity and acceleration drag loads on submerged structures.

4.3.7.2 Evaluation Procedure The procedure used to calculate forces on submerged structures in the LOCA jet path is presented below. The two models for the water jet and for the bulk pool motion induced by the jet are presented separately, followed by the load equations.

The following plant unique LOCA jet information is obtained from the QSTF plantspecific test series:

¶ = an identifier for each fluid particle in the jet, equal to the time at which the fluid particle was discharged from the downcomer.

VD(T) = jet discharge velocity for particle T, ft/sec dVDQr di Water Jet aD(T) = jet discharge acceleration for particle T, ft/sec2 j

The maximum jet penetration, 9. is determined by v 2 ~

= DaD ~max~

max ,ft p

where imax denotes the last time step, i.e., vent clearing time.

If 9. < L, there is not jet load on the structure.

p If ip > L, there is a jet load on the structure.

L = vertical CX) distance (ft) from the bottom of the downcomer to the 5

4.3.72 Revision 2

NEDO21888 structure of concern. Note that the X coordinate defined here is different from that defined previously for the torus.

The plant unique LOCA jet information is used to construct a plot of the jet front position vs. time. This is done by plotting the position (denoted as X.).

j

4. 3. 72A Revision 2

NEDO 21888 vs time (elapsed time following start of DBA) of a number of fluid particles which exit the downcomer during the clearing transient. This timespace plot (t,x) is based on the following equation:

x. = (~r) (t¶)

Whenever a line of constant T overtakes an earlier line, the earlier line ceases to extend and the overtaking line continues. The timespace plot enables one to enter a location x, and time t, read the corresponding value of T, and then obtain the associated jet velocity from the plant unique LOCA jet data.

As the jet travels through the pool, the particles at the rear, which were discharged from the downcomer at higher velocities, catch up with particles at the front, which were discharged at lower velocities. When this over-taking occurs both particles continue on at the higher velocity. When the last fluid particle leaving the downcon.er catches up to the front of the jet, the jet dissipates, which shows that as the rear particles catch up to the particles in front, the jet becomes shorter and wider.

For a struc~ture at position x, values of T and t can be obtained from the tir.iespace plot explained above. Then the above equation for Xj and the plant unique LOCA jet data can be employed to determine jet area at position x for various values of time, t. The jet area is given by:

___________________________ AD vD(~) _____

A(x,t) dV (~) dV (~-)

1 ___ D 1i/l~ D dT (tT~ kvD(T),i dt /Ix.

where A~ is the discharge area.

The area of the jet front is approximated by the area the jet possesses one 1/2 initial discharge diameter D = (4A.~/7r) behind the actual jet front.

4.3. 73 Revision 2

NEDO2 1888 If I < D, the jet front area should be approximated as x = ~ L The s 2 s~

jet radius is calculated in order to determine which structures overlap the jet, and how much overlap actually occurs.

Bulk Pool Motion Induced by the Water Jet Forces due to poo1 motion induced by the advancing jet should be calculated for structures that are within four downcomer diameters below the downcomer exit elevation. The flow field is computed by modeling the moving jet front as a hemispherical cap centered one downcomer diameter behind the water jet front posi-tion. It contains the same amount of water as the jet and moves with the velocity of the water jet front. The formulas for the hemispherical radius (R) and the trajectory of the hemispherical center (X ) are c

R =~[~L (t) +

3(xf(t)3] /~ for Xf (t) ~ D 5

R (t) 9Xf Ct) 1 for Xf (t) < D s 2 L 2D Xc (t) = XfCt) D for Xf (t) > D xc (t) = 0 for Xf (t) ( D where D = downcomer diameter, ft Xf (t) = position of water jet front, ft The equivalent uniform velocity and acceleration at the location of the structure (X,Y) is obtained from the time dependent potential function induced by the jet front 3i~X~Xc dX (x,Y,t)= fr(i~) - +/-(f~) dt 4.3.74 Revision 2

NEDO2 1888 where r = (xX )2 +y2 Y = transverse distance of the structure from the jet axis (XX C ) = distance of structure from effective jet front center along the jet axis The local uniform flow velocity is U~, (X,Y,t) = V The acceleration is a (X, ,t) =

This calculation is performed for r > R and X > X . If either of these s c conditions are not satisfied, then the procedure to calculate forces due to bulk pool motion induced by the water jet is not necessary, and the water jet loads bound the induced loads due to pool motion.

Load Calculation Structures that are within four dowacomer diameters below the downeomer exit elevation will sustain a loading, first from the flow field induced by the jet, then from the jet itself if it is within the range of the jet.

Forces from the flow field are both standard drag and acceleration drag.

The force from the jet is standard drag only, since particles within the jet travel at constant discharge velocity (i.e., there is no acceleration).

The standard drag force on the submerged structure is computed based on the normal component of velocity intercepting the structure, the projected area of the structure intercepted by the normal component of velocity, and the jet or flow field area.

The standard drag force is 2

F5(t) = CD A ~ ~ lbf 2 gc 4.3.75 Revision 2

NEDO2 1888 where CD = applicable standard drag coefficient for a structure fully submerged inside the jet boundary

= 2 for momentum stoppage with no momentum if the structure fully or partially intercepts the jet or flow field

= 4 if the structure turns the jet or flow field back on itself

= projected area of the structure intercepted by the normal component of velocity, ft2

= velocity component normal to the surface

= VD (T) cos8 for a structure intercepted by the jet VD CT) = discharge velocity for T for the structure position L

5, ft/sec u = U~cose = V4 (cose) for a structure within the flow field, ft/sec 6 = angle between the normal to surface A~ and the vertical plane, degrees p = density of pool water = 62.4 ibm ibm ft gc = 32.2 The acceleration drag force is FA (t) = £UWV where p = density of pool water = 62.4 ibm

= acceleration of flow field at the sub

~t~Y 2

merged structure, ft/sec V acceleration drag volume for flow normal to the surface, ft3

= 32.2 ibmft lbf sec2

= time dependent potential function induced by jet front 4.3.76 Revision 2

REFERENCES FOR SECTION 4.3.7 4.3.71 F. J. Moody, Analytical Model for Liquid Jet Properties for Predicting Forces in Rigid Submerged Structures, General Electric Company, Report No. NEDO21472, April 1978.

I 4.3.72 J. M. Humphrey, Mark I Containment Program 1/4 Scale TwoDimen-sional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO-21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.77/4.3.78 Revision 2

NEDD21888 4.3.8 LOCA Bubble - Induced Drag Loads on Submerged Structures During the initial phase of the DBA, pressurized drywell air is purged into the suppression pool through the submerged downcomers. After vent clearing, a single bubble is formed around each downcomer. For the DBA the duration of the LOCA bubble is typically 0.2 seconds from its initial formation until it breaks through the pool surface. It is during the bubble growth period that unsteady fluid motion is created within the suppression pool. During this period all submerged structures below the pool surface will be exposed to transient hydrodynamic loads.

4.3.8.1 Bases and Assumptions The bases of the flow model and load evaluation for the LOCA bubble-induced submerged structure load definition are derived from the model in Reference 4.3.8-1 and are summarized briefly below.

The major assumptions used to develop the analytical model and load evaluation are listed below.

a. Bubble Dynamics The initial LOCA bubble pressure is the same as the drywell pressure at vent clearing. The transient drywell pressure is determined from the QSTF plant unique data of Reference 4.3.84

- Air is an ideal gas.

The bubble is spherical and characterized by the Rayleigh equation.

b. Flow Fields - The flow fields are established by using inviscid, irrotational potential flow theory with the bubble considered as a fluid source.

4.3.81 Revision 2

- The presence of boundaries is considered by using the method of images (sink for the free surface and source for the solid walls). The circular cross section of the torus shell is modeled by a rectangular cross section (Model E contained in the model evaluation report Ref. 4.3.82).

After contact between bubbles of adjacent downcomers, the pool swell flow field above the downcomer exit elevation is derived from QSTF plant unique tests of Reference 4.3.84. The load will act only vertically from this time on. This pool swell drag load is computed using the method described in Section 4.3.4

c. Drag Loads Submerged structures are assumed rigid to on Structures maximize the load.

The standard drag forces on circular cylinders are computed by using a conservative drag co-efficient of Cd = 1.2, independent of the Reynolds number.

Drag forces on structures with sharp corners (e.g., rectangular and I beams) are computed by considering forces on an equivalent cylinder of diameter D eq

=/T L max where L max is the maximum transverse dimension and is defined as the diameter of a circumscribed cylinder about the crosssection of the structure.

4.3.8la Revision 2

NEDO-2 1888 Long slender structures are modeled in segments of length CL), which do not exceed its diameter (D or D ). The total force on the structure is eq obtained by integrating the individual segment force over the entire length of the long structure.

Longer segments are used as long as the equiva-lent uniform flow velocity and acceleration are evaluated conservatively for every point on the segment.

Interference effects due to the proximity of solid boundaries are considered for each segment that has its center less than 1.5 diameters from the boundary. Multipliers are needed to increase both the acceleration and the standard drag. Simi-larly, multipliers are needed to include the inter-ference effects between neighboring structures where the centers of the segments are less than 3D apart, (D = D +

1 D

2) /2, the average diameter of the two structures).

For structures near walls, the multiplier (l+A w shall be used to increase the..acceleration drag and the multiplier (l+D ) shall be used to in-w crease the standard drag. Bounding expressions for A and D are given below as functions of w w Xw (Xw = r/D 1/2, where r is the distance from the segment center to the boundary and D is the diameter of the structure) 0.05<x w <1.0 Aw =0.05/X w D = 0.12/x w w X< 0.05 A = 1.0 w w D = 2.4 w

4.3.8lb Revision 2

NEDO21888 For structures with neighbors that are less than 3D away and within 300 of being parallel, the multiplier (1+A1) shall be used to increase the acceleration drag and the multiplier (1+D I )

shall be used to increase the standard drag.

Bounding expressions for A and I D1 are given below as functions of (X = r /D 1, where I 12 is the distance between segment centers):

D 0.05< X <2.0 A1 0.2 2 I D+D 1 2/

= 0.2/X I where D is the diameter of the structure under 1

consideration and D is the diameter of the 2

neighbor. If more than one neighbor must be considered, the A1 and D1 values may be summed over the neighbor structures. For X I less than 0.05, the two neighbor structures shall be con-sidered as an effective single structure.

The effects of wall proximity and neighbor structures may be superimposed in order to compute overall multipliers as follows:

(1+A w +ZA k 1k )

(l-4-D +ID )

w k 1k Sum of the standard drag and acceleration drag is the total drag force on a structure. The basis for these two components Qf the drag force was established in References 4.3.81 and 4.3.82.

4. 3.8lc Revision 2

NEDO21888 A series of tests *on submerged structures was performed in the Mark I QSTF (Reference 4.3.83). The tests measured LOCA bubbleinduced loads on four cylindrical structures submerged both horizontally and vertically near the downcomer exits. The entire blowdown event for each test was also filmed by a high. speed movie camera. The main objective of the test series was to acquire quantitative total drag loading data on cylindrical structures exposed to an accelerating flow field in an actual scaled prototypical configuration and to compare these results with the predictions of the analytical model given in Reference 4.3.81. Detailed test procedures and measured results are presented in Reference 4.3.83. The model evaluation report (Reference 4.3.82) compares the test data to the results predicted by the analytical model.

4.3.8.2 Load Definition Detailed procedures to calculate drag forces on submerged structures are described in Reference 4.3.81 which provides the methodology of utilizing the unsteady velocity and acceleration flow fields in the suppression pool to obtain drag loads on submerged structures, such as pipes, beams, etc.

4.3.82 Revision 2

NEDO21888 The load definition method first requires establishement of the velocity and acceleration flow fields within the torus by simulating the DBA con-dition at the vent exists. Then, drag loads on submerged structures are calculated based on these transient flow fields.

a. Flow Field Establishment An air charged bubble is considered as the fluid source having a source strength determined by the dynamic characteristics of a bubble. The Mark I torus LOCA bubbles are simulated by multiple sources in a finite pool. By using potential flow theory and the method of images for source and sink considerations to simulate solid walls and a free surface, the velocity and acceleration flow fields within the torus are established.

b. Drag Loads Evaluation The drag force on a submerged structure consists of two components:

the standard drag and the acceleration drag. The standard drag load is computed on a locally uniform flow field evaluated at the local velocity. The acceleration drag load is evaluated based on an inviscid, uniform but unsteady (accelerating) flow field. The sum of these two drag forces gives the total drag load on a sub-merged structure.

4.3.8.3 Selection of Key Parameters For Load Evaluation Procedures This section outlines the key parameters which influence the magnitude and characteristics of the LOCA bubble drag loads on submerged structures. The procedure used to specify the input for the submerged structure drag load model is also discussed.

The plant specific suppression pool geometry should be identified first as follows:

a. Torus shell dimensions

b. Torus water depth 4.3.83 Revision 2

MEDO2 1888

c. Location of submerged structure considered (Figure 4.3.8-1)

d. Locations of downcomers (Figure 4.3.8-1)

Other plant specific data which need to be identified are:

a. DBA transient d.rywell pressure

b. Thermodynamic properties of drywell air and wetwell water

c. Initial drywell/wetwell pressure differential

d. Overall vent system friction factor

e. Downcomer inside diameter

f. Downcomer submergence

g. Initial bubble surface velocity

h. Undisturbed pressure at bubble position before bubble appears.

The submerged structure considered is divided into the appropriate number of sections. The coordinates of each section center, orientation of the struc-ture axes and the structural projected crosssectional area are identified.

The acceleration drag volume for each section of the submerged structure is then obtained from tables given in Reference 4.3.81.

With input parameters identified and calculated, the submerged structure drag load model is used to evaluate the resultant transient drag forces on each structure element being analyzed. In addition, transient bubble dynamic characteristics within the pool can be found. Typical results are shown in Figures 4.3.8-2 and 4.3.8-3. These figures show the LOCA bubble drag forces on a vent header support column in the radial (X) and tangential (Z) directions as a function of time.

4.3.84 Revision 2

NEDO2 1888 INITIAL SUBMERGENCE DISTANCE BETWEEN DOWNCOMER AND TORUS q WATER DEPTH DOWNCOMER ID x

SECTION A-A VENT HEADER SUPPORT COLUMN Figure 4.3.81. Torus and Typical Submerged Structure Geometry 4.3.8s Revision 2

NEDO2 1888

.0 cn w

C-)

0 LL x

-J I-0 I-TIME (sac)

Figure 4.3.82. Sample History of Total XForces on Vent Header Support Column 4.3.86 Revision 2

NEDO21888

.0 -1501 CO w

U 0

U.

-J 0~

TIME (nec)

Figure 4.3.83. Sample Time History of Total ZForces on Vent Header Support Column 4.3.87/4.3.8...8 Revision 2

REFERENCES FOR SECTION 4.3.8 4.3.81 F. J. Moody, Analytical Model for Estimating Drag Forces on Rigid Submerged Structures Caused by LOCA and Safety Relief Valve Ramshead Air Discharges, General Electric Company, Report No. NEDO21471, September 1977.

4.3.82 Submerged Structures Model Evaluation Report, General Electric Company, Report No. NEDO21983, September 1979.

4.3.83 1/4Scale Test Report: Loads on Submerged Structures Due to LOCA Air Bubbles and Water Jets, General Electric Company, Report No. NEDO23817, August 1979.

4.3.84 J. M. Humphrey, Mark I Containment Program 1/4Scale Two Dimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.89/4.3.810 Revision 2

NEDO-21888 4.3.9 Vent Header Deflector Loads Several Mark I Utilities will install vent header deflectors to mitigate pool swell loads on the vent header. The purpose of this section is to define the poo1 swell loads on the vent header deflectors. Figure 4.3,91 shows a typical deflector. The deflector generally consists of a pipe, often with angles or tees welded on the sides, located below the header (see Figure 4.3.92 for typical deflector designs).

4.3.9.1 Bases and Assumptions Two dimensional vent header deflector loads are based on QSTF measurements (Reference 4.3.91) where available or on a combined analytic and experimental method (Reference 4.3.92) where measured loads are not available. Both the measured and calculated two dimensional loads utilize the analytical method and EPRI 1/12 scale tbree~dimensional data (Reference 4.3.93) to generate deflector loads as a function of location along the length of the deflector (~ dimension in Figure 4.3.93).

Deflector loads are presented as the vertical component of the integral over the angle ~ (see Figure 4.3.9-3) of the local pressure and thus are a running load in units of LB/ft as a function of time and location, The load definition includes impact, acceleration drag, steady drag, bouyaney and a static pressure increase with time. The development of the load definition required two specifications:

• The transient fluid flow field past the deflector, and

• The appropriate drag equations (to predict the loads on the deflector from the fluid flow field information).

A detailed explanantion of the procedures used to generate the transient fluid flow field (displacement, velocity, and acceleration) and the resultant de flector loads is given in Reference 4.3.92.

4.3.91 Revision 2

NVEDO21888 4.3.9.2 Summary of Procedure for Analytic Deflector Loads 4.3.9.2.1 - Two dimensional load method a) Flow field prediction QSTF movie data provides the free surface displacement history. Data is taken sufficiently far from the vertical centerline (l8inches) such that the surface displacement history is not disturbed by the presence of the deflector and header and a factor is used to adjust back to centerline velocity. The disp1aceme~it history is curve-fit by assuming the surface acceleration history is proportional to the download history,_ integrating this function twice with respect to time and adjusting the proportionality constant to yield agreement with the displacement data.

Finally, the influence of the deflector drag on the finite water region below the deflector is taken into account by an empirical turn down function deduced from bubble top displacement histories of the water region adjacent to the deflector.

Details of this procedure can be found in Reference 4.3.92.

b) Two-dimensional load Prediction Equation 1 is used to calculate the load per unit length.

F=.pgA~ + [in + J)A.]T) + 1dm.2

-n+

gdy SWP (g+~)

C (1) gc Bouyancy +Acceleration +velocity + Static Pressure Drag Drag Increase with Depth Where A~ = Displaced area as a function of y, Figure 4.3.94, ft2 p Water density, lbm/ft3 in = Hydrodyinanic mass per unit length as a function of y, lbm/ft fl = Location of free surface, Figure 4.3.94, ft n = Free surface velocity, ft/sec fl = Free surface acceleration, ft/sec2 4.3. 92 Revision 2

y = Immersion depth of deflector r:ea.~a1LJfrom bottom of deflector, Figure 4.3.9 4., ft.

S = Immersion of deflector measured from maximum width of deflector, Figure 4.3.95, ft.

W = Maximum width of deflector, Figure 4.3.9-4 , ft.

g =~2onversion constant = 32.2 lb in ft 2 C

lb sec f

The hydrodynamic mass per unit length C including the steady drag component) and its derivative are given by:

din =1 ~WC (y) (2) dy ~ D m =1 PWJYC (E)dE (3) 2 Where C (y) is given for each deflector type in Figures 4.3.96 thru 4.3.910.

D In computing the integral of C Cy) in equation (3) the final dotted line extending D

from the steadystate value of C to C = 0 is used.

D D 4.3.9.2.2. - Three Dimensional Loads Based on EPRI 1/12 scale 3D movie data using main vent orifice test configura-tion, a bounding longitudinal multiplier for displacement, velocity and accel-eration was determined as shown in Figure 4.3.911. This multiplier evaluated at Z/2. = 0, 0.5 and 1.0 was applied to the QSTF displacement, velocity and acceleration histories to generate load histories at Z/2. 0, 0.5 and 1.0.

These load histories are scaled to full scale by:

F 1. F p Quarter Scale Full Scale At corresponding times; e:

e Full scale = 1 e Qtrarter Scaie WhereA= QSTF scale factor (approximately .25) 4.3.93 Revision 2

NEDO21888 4.3.9.3 Summary of Procedures for Use of QSTF Deflector Loads The measured deflector loads in QSTF are for an average cell and must be adjusted to yield values at Z/R. = 0,0.5 and 1.0. In addition, the analytical impact spike is added to the measured force since the measurement system may not have resolved this rapid transient. The assumed spike to be added to the data at the time of impact is shown in Figure 4.3.912. The threedimensional load variation shall be based on the EPRI 1/12 scale 3D main vent orifice test data (Reference 4.3.93).

The procedure to generate measured loads at various Z/9. stations is as follows:

1) Run analysis to generate four load histories, QSTF and 1/k, = 0, 0.5 and 1.0 (see Figure 4.3. 913).

2) Add the impact spike from the QSTF analysis to QSTF measured data at time of impact and adjust time scale to cause time of impact, e, of QSTF analysis and data to agree (see Figure 4.3.914). This impact spike is not added to the type 4 deflector.

3) A relationship between the time scale of each calculated Z/L force history and the calculated QSTF force history is established. From impact to submergence which corresponds to the achievement of steady state drag coefficient, times are related at equal submergences (See Figure 4.3.915). For times greater than the time when C = C the time differences are linearly decreased until no time D D difference exists when F QSTF = 0 This form of time correlation is used because when CD is varying rapidly with submergence (early in time), changes in deflector force are caused principally by changes in suhmergence. When C = C~ and the D D major force contribution is from the static pressure term (later in time), forces are correlated principally by time.

4) Force ratios between calculated forces at Z/2. = 1.0, 0.5 and 0, and the calculated QSTF force are formed at corresponding times established in Step 3.

5) The QSTF measured load with the assumed analytical spike from Step 2 is multiplied by the three ratios established in Step 4 and plotted at the corresponding times to create three load histories.

4.3.94 Revision 2

NEDO2 1888

6) Scale the results of Step 5 to full scali by:

F FS=~1 FStep 5 at times scaled by:

8 e FS = 1 Step 5

-x Where A= QSTF scale factor (less than one)

7) Multiply full scale values by safety factor of 1.15 for experimental uncertainty.

4.3.9.4 Load Definition The deflector loads are calculated on a plant unique basis by either of the above methods and are presented in the Plant Unique Load Definitions. Figure 4.3.9-16 shows a typical deflector load transient at three locations for a typical Mark I plant.

4.3.9.5 Faliback Loads on the Deflector The QSTF plant unique movies (Reference 4.3.9-1) indicate that there are no bulk pool fallback loads on the deflector because the deflector an6 nader tend to push the pool water out toward the torus walls as the pool swells upward. The movies do indicate that there would be some small froth failback load resulting from the froth located between the header and deflector. This load can be bounded by using the froth fallback load procedure explained in Section 4.3.5 For this load definition the froth faliback density should be 10%.

4.3.9.6 Application of Vent Header Deflector Load The vent header deflector load should be applied in the upward direction within a range of directions including a variation of +100 from the upward vertical. In 4.3.95 Revision 2

NEDO21888 applying the load to the deflector in the structural analysis, the inertia due to the added mass of water below the deflector shall be accounted for.

The added mass per unit length of deflector may be estimated by:

where MH = hydrodynamic mass per unit length, lb/ft I = total impulse per unit length associated with the impact transient, lbsec/ft V = impact velocity, ft/sec 2

g = conversion constant, 32.2 ft/lbf sec The froth failback lDad.should be applied .to the downward direction .with a range of directions including a variation of +/-450 from vertical.

Revision 2

NEDO-21 888

- 1.0 x~j~ 0.0 FIGURE 4.3.9-1 TypIcal Vent Header Deflector 4, 39..6 Revision 2

NEDO-21888 TYPE ¶ TYPE 2 PIPE PIPE WITH ANGLES 1YPE 3 *rYPE PIPE WITH TEES WEDGE Figure 4.3.9~2. Typical DeflectOr Design 4.3.99=7 Revision 2

Ii flain vent Vent hrarier I Midway between two main vents z

~0 m co 0 0 C 0,

cm 1~

(U 4

H 0,

H 0

FIGURE 4.3.9-3 Definition of Space Variables ha

NEDO-21 888

/

INITIAL SURFACE

-- ~--

FIGURE 4.3.9-4 VARIABLE DEFINITION V

FIGURE 4.3.9-5 STATIC PRESSURE IFflERSION DEPTH 4.3.9-9 Revision 2

NEDO-?l 888 04 V

5.

6.

F ~I -

____ ____~~

4--,- ~~=Li ~ zi ~ - - - - -- -,

-~

pVd

-- --- --- - - -~ - --4-- .- -- -

_________________ ~ 4 - -,.

~*-- .~1 -1 .

4.

___________ _____ (Ce), from Figure 4.3.9-7

-~

i~ ----4---

a

2. - -~ ---- I- - - E

- El

~ -

=

= ~ ~---.=.~=-- 4 1*--

0.

0.136 0.5 1.0 Vt/d FIGURE 4.3.9-6 Impact and Steady Drag Force Correlation for Type 1 Deflector 4.3

• 9-10 Revision 2

NEDO-2 1888 2.

CD..

1.

.1 0.

1. 10. 100.

FROUDE NUMBER [ V2 / Dg )

Figure 4.3.97 for Type Deflector 4.3.911 Revision 2

NEDO 21888

• V 5 7(diw) 4 F

IT 2w V 3 2

2 I

I I 0.7 U

I 0.207 0 0.136 0.854 1.0 Vt/d Figure 4.3.98. Impact and Steady Drag Force Correlation for Type 2 Deflector 4.3.912 Revision 2

NEDO21888 Kw~1 V

IT 3.

F 2w V

1.

0. 0.136 0.83 1.33 2.

VtJd Figure 4.3.99. Impact and Steady Drag Force Correlation for Type 3 Deflector 4.3.913 Revision 2

NEDO-21 888 V

4.

3.

F 2

2 2.

1.

0.

0.

ZVt/w FIGURE 4.3.9-10 impact and Steady Drag Force Correlation for Type 4 Deflector 4,3,914 Revision 2

NEDO-21 888 Id 13 1.2 1.19 1.-

1.1 4-,

1.0 0.3

• 1~

0~)

0

-J 0.8 0.7 0.5 ft. I I I I I I I 0 0.1 02 0.3 OA Os 03 0.7 03 Ci 1.0 FIGURE 4.3.9-11 Longitudinal Multiplier for Fluid Displacement Velocity and Acceleration 4.3,9-15 Revision 2

NEDO-21 808 F = Vertical upward force on deflector per iz~.it length 8.

d = Diameter of cylinder in deflector types 1 3 V = Impact velocity

= Water density

6. _ = -

t = Tine from begin.ing of impact F ____________ 4 _______________________

p Vd I

4.

2.

~:+/-

+

• ~- i.

I 4 4 ~

I ~ ___

0.

0.

I 0.1 0.2 Vt / d FIGURE 4.3.9-12 Impact Force Transient for Addition to the Empirical Data for Deflector Types 1 3.

4.3.916 Revision 2

NEDO-21 888 F

C

• 1.0 a.,

~1 C * .5 L.

a.,

0.

a.,

U I.-

0

=0 FIGURE 4.3.9-13 CALCULATED LOAD HISTORIES INITIAL IMPACT SPIKE (QSTF ANALYSIS)

F (TIME SHIFTED IF REQUIRED) e FIGURE 4.3.9-14 QSTF MEASURED LOAD WITH ASSUMED ANALYTICAL SPIKE 4 .3. 9-17 Revision 2

NEDO-21 888 0

I L~.

I-U, U,

LJW z-., z L5~.

LJJ ~

z S U,

I ~~1 ~ QSTF 0 0 Li-)

.6-i II O0 z:i V U,L)

U, I-.

4.,

0.

a- >. >-

=

0

= 1.0 0

Is-Isa z

I QSTF TIME FIGURE 4.3.9-15 Time Correspondence Between QSTF Calculated Force And Calculated Forces At Other ~/t Stations 4.3.918 Revision 2

NEDO-21 888 w

0 A.

TIME (nin.cI FIGURE 4.3.9-16 Typical Vent Header Deflector Load 4.3.919 Revision 2

NEDO21888 REFERENCES FOR SECTION 4.3.9 4.3.91 J. M. Humphrey, Mark I Containment Program, 1/4 Scale Two Dimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

J. M. Humphrey, Mark I Containment Program 1/4 Scale Pressure Suppression Pool Swell Test Program: Supplemental Plant Unique Tests, General Electric Company, Report No. NEDO24615, June 1980.

4.3.92 W. Kennedy, et al., Mark I Containment Program, Vent Header Deflector Load Definition, General Electric Company, Report No. NEDO24612, April 1979.

4.3.93 The Electric Power Research Institute, Three Dimensional Pool Swell Modeling of a Mark I Suppression System, EPRI NP906, October 1978.

4.3.920 Revision 2

NEDO21888 4.4 CONDENSATION OSCILLATION LOADS Following the pool swell transient of a postulated LOCA, there is a period during which condensation oscillations occur at the downcomer exit. Condensa-tion oscillation loads are caused by periodic pressure oscillations on the torus shell, submerged structures, and in the vent system. These pressure oscillations are associated with the pulsating movement of the steamwater interface caused by variations in the condensation rate.

The loads specified for condensation oscillation are based on the FSTF tests.

In these tests a prototypical segment (one bay) of a Mark I torus and vent system were subjected to ten steam and liquid blowdowns simulating a range of LOCAs. Duration of the condensation oscillation period varies with break size and is specified in Table 4.4.11. Load combinations are given in the bar charts in Section 3.0. Condensation oscillation loads with the largest ampli-tude occur during the DBA. For the IBA, bounding chugging loads are specified in lieu of condensation oscillation loads.

The purpose of this subsection is to specify the loading conditions which may be produced during this phase of the LOCA transient. Loads on the structures within the Mark I Containment system that may be induced by condensation oscillations are defined in the following subsections:

4.4.1 Torus shell loads 4.4.2 Loads on submerged structures 4.4.3 Lateral loads on dowocomers 4.4.4 Vent systems loads 4.41/4.42 RevisiOn 2

NEDO21888 4.4.1 Torus Shell Loads Oscillating loads on the submerged portion of the torus shell during the con-densation oscillation phenomenon are caused by periodic pressure oscillations superimposed on the prevailing local static pressures. In the following sub-sections the definition of these loads is presented along with a discussion of the bases and assumptions used and the plant unique application considerations.

4.4.1.1 Bases and Assumptions The parameters which were varied in the FSTF tests include dowocomer submerg-ence, initial pool temperature, blowdown fluid phase (liquid and vapor), and initial wetwell pressure. The break size used in the FSTF for the DBA provides steam mass fluxes in excess of those predicted for any Mark I plant. The load-ing conditions established from the FSTF data and the use of conservative application techniques result in bounding condensation oscillation loads for all Mark I plants. The FSTF, test matrix, and test results are described in Reference 4.4.11.

Data from the FSTF tests indicate that the amplitude of the pressure oscilla-tions induced by condensation oscillations on the torus shell is dependent on the break size and the phase of the blowdown fluid (liquid or vapor). The highest pressure amplitude was observed during test M8, the large liquid break test which simulated the DBA conditions (Reference 4.4.11). Data from this particular test run of the FSTF were used as a conservative basis for the load definition for the DBA. For the IBA, chugging loads are specified in lieu of condensation oscillation loads. In the FSTF tests evaluated (Ml, M4, and M9) with a break size equal to 25 percent of the DBA area, strong condensation oscillations did not persist. A break size of 25 percent of the DBA area is larger than the IBA. Some pressure oscillations did occur in these tests during air carryover, but the pressure amplitudes were bounded by the peak values of the chugging pressures which occurred following the air carryover period. The IBA steam tondensation loads therefore are specified as being chugging loads as defined In subsections 4.5.1.2 and 4.5.4.2.

4.4.11 Revision 2

NEDO21888 The onset times and durations for condensation oscillations are specified in Table 4.4.11. The onset times were selected as conservatively small values based on FSTF tests. The durations were based on an analytical study of the DBA and IBA, from which a conservatively large value was selected for the time required to depressurize the system into the chugging regime. The analysis was conducted on a plant configuration which would produce the slowest depressurization rate. The SBA will not result in flow rates sufficiently large to enter the condensation oscillation regime.

Plant unique loads are derived from FSTF data. Flexible wall loads were measured directly in the FSTF which is prototypical of a Mark I plant configuration.

Pressure measurements obtained from various locations on the torus shell show that the longitudinal pressure oscillation amplitude distribution along the torus centerline is essentially uniform. These measurements also show that the varia-tion of torus wall pressure amplitude (from zero at the water surface to maximum at the torus bottom dead center) can be appropriately represented as linearly varying with elevation (see Figure 4.4.12). With the spatial distributions of pressure oscillation amplitude established, the load definition can be specified in terms of pressure at the torus bottom dead center.

Pressure oscillation amplitude variation as a function of oscillation frequency is determined from integrated total pressures at the various frequencies which were derived from torus shell pressure time histories.

The basis for the pressure amplitude specification is several segments of maxi-mum pressure amplitude data measured during FSTF tests M7 and MS. Amplitude vs frequency correlations were compiled for one second intervals of data, and the peak average amplitudes were determined for each 1 Hz frequency band.

These amplitudes as a function of frequency are used as the load specification.

The frequency range in the pressure amplitude specification has been established to cover the range of frequencies observed in the ESTE tests and the range expected in all Mark I plants. Two dominant frequencies (approximately 5 and~

10 Hz) were observed in the FSTF tests M7 and~ MB. These dominant frequencies were also observed to vary during the tests. The load specification frequency ranges, which were selected to conservatively bound these dominant frequency variances, are 4 to 8 Hz and 8 to 16 Hz respectively.~

4.4.12 Revision 2

NEDO21888 Alternate frequency spectra for the 4 to 8 Hz and 8 to 16 Hz frequency ranges have been specified in Figure 4.4.11 and Table 4.4.12. For a plant unique structural evaluation, the structural responses from each 1 Hz band between 0 and 50 Hz would be summed. The 0 to 50 Hz total range analyzed would include only one of the alternate frequency spectra specified for the 4 to 16 Hz range in Table 4.4.12. The alternate frequency spectrum selected to be included in. the.analysis would be that which produces the maximum total response. -

In order to apply the FSTF measured loads to plant unique geometries, two adjustments are made. . The first adjustment is to account for FSI effects in the FSTF data and the second is to account for differences in the ratio of the pool surface area to vent cross sectional area among the Mark I plants.

Structural response effects unique to the FSTF are accounted for by the specification of a baseline rigid wall load which is given as pressure oscilla-tion amplitude as a function of frequency. This load has been derived from the measured FSTF flexible wall load by analysis with a coupled fluidstructural dynamic model of the FSTF torus.

The derivation of the baseline rigid wall load is described below:

a. A finite element coupled fluidstructural model of the FSTF torus was excited at varying frequencies with a unit amplitude pressure source at the vent exits. The torus shell pressure amplitudes relative to the source pressure (amplification factors) were determined as a function of frequency.

b. Using these relative amplitudes (amplification factors), the FSTF vent exit source pressures were derived from the measured torus shell pressures at the various frequencies.

c. The baseline rigid wall load was derived from the computed FSTF vent exit source pressures by applying incompressible potential flow relations (spatial attenuation from the vent exits to the torus wall which is assumed to be rigid).

4.4.13 Revision 2

NEDO21888 The baseline rigid wall load is then adjusted for attenuation due to a plants particular ratio of pool surface area to vent cross sectional area. A multiplication factor to provide this adjustment has been computed from incompressible flow theory for various area ratios (see Figure 4.4.13). Values for the pressure attenuation between sources at the vent exits and the bottom center of the torus were determined from the product of separate two dimen-sional potential flow solutions obtained in orthogonal planes. This method of determining pressure attenuation was proven to be conservative by comparison with a solution from a three dimensional finite element model. Pressure attenuation factors were computed for various area ratios and normalized to the FSTF value to provide the multiplication factors shown in Figure 4.4.13.

Since the FSTF area ratio is smaller than that of niost Mark I plants, the plant unique loads in most cases will be less than or equal to the baseline rigid wall load. The plant unique load is intended to beused in conjunction with a flex-ible wall coupled fluidstructural model in the -plant unique structural evaluation.

The following major assumptions are made in the definition of the condensation oscillation torus load:

a. The frequency range observed in the FSTF data is assumed to be applicable to all Mark I plants. This is justifiable since the con-densation oscillation frequency is primarily controlled by thermo-dynamic and geometric conditions at the vent exit. These conditions are very similar for all Mark I plants and were simulated in the FSTF tests. Minor adjustments to account for vent system and down-comer acoustic frequencies were made to conservatively represent all Mark I plants.

b. Baseline rigid wall pressure oscillation amplitude values can be derived from the flexible wall measurements in the FSTF by the use of fluid coupledstructural dynamic models and incompressible flow theory.

4.4.14 Revision 2

NEJDO21888

c. The effect of the ratio of the pool surface area to vent cross sectional area can be determined since it is assumed that the pressure oscillations will attenuate from the source at the down comer exit to the pool boundary according to incompressible potential flow theory as described in Subsection 4.4.1.1.

4.4.1.2 Load Definition The condensation oscillation load on the torus shell is a plant unique rigid wall load. It is intended to be used in conjunction with a flexible wall coupled fluidstructural model in the plant unique structural evaluation. The plant unique loads are derived from the baseline rigid wall loads as outlined in the following subsections.

4.4.1.2.1 Baseline Rigid Wall Loads Based on the considerations discussed in the previous subsection, the following baseline rigid wall condensation oscillation load definition was derived for the DBA:

Amplitude versus frequency Values given in Table 4.4.12 (Also shown in Figure 4.4.11). The total spectrum to be analyzed is from 0 to 50 Hz, and is to include the one spec-trum from 4 to 16 Hz which produces the maximum total response.

Total Response Resulting responses, from applying the amplitude at each frequency given in the total spectrum to be analyzed, are to be summed.

Spatial Distribution Uniform axially along the torus center-line. Linear attenuation with sub-mergence along the wetted perimeter of the torus cross section as shown in Figure 4.4.12.

4.4.15 Revision 2

NEDO2 1888 For plant unique applications where the ratio of pool area to total downcomer area is greater than that of the FSTF, the magnitude of the condensation oscillation baseline rigid wall load is reduced. A discussion of this adjust-ment is presented in the following subsection.

4.4.1.2.2 Plant Unique Loads Since the dimensions of the torus and the number of downcomers vary from plant to plant, the magnitude of condensation oscillation loads is expected to be different for each plant. To facilitate calculation of condensation oscilla-tion loads for plant unique applications, a multiplication factor has been developed to account for the effect of the pooltovent area ratio. In Figure 4.4.13 this factor is plotted versus the pooltovent area ratio.

To determine the plant unique load, the pooltovent area ratio is computed.

This value may be based on the average value for the entire plant for evalua-tions involving the total net vertical pressure force (e.g., in torus support system load evaluations). For all other evaluations, the smallest value of pooltovent area ratio for any bay in the torus shall be used. Then, from Figure 4.4.13, the multiplication factor for the appropriate calculated pool tovent area ratio is obtained. The plant unique oscillating load on the torus is determined by multiplying the amplitude of the baseline rigid wall load (Section 4.4.1.2.1) by this factor. The resulting plant unique oscillating load is to be applied to the prevailing local static pressures at the various locations on the torus shell at the appropriate times (see Table 4.4.11).

4.4.16 Revision 2

NEDO21888 Table 4.4.11 CONDENSATION OSCILLATION ONSET AND DURATION Onset Time Duration Break Size After Break After On~pt DBA 5 seconds 30 seconds IBA 5 seconds* 900 seconds*

SBA Not Applicable Not Applicable

• For the IBA, the condensation oscillation loading is a continuous sinusoidal function with a peak amplitude and frequency range of that specified for the prechug load in Section 4.5.1.2.

4.4.17 Revision 2

NkDU~i~~

Table 4.4.12 CONDENSATION OSCILLATION BASELINE RIGID WALL PRESSURE AMPLITUDES ON TORUS SHELL BOTTOM DEAD CENTER Alternate Amplitudes to be Analyzed*

Frequency Amplitudes (PSI)

Range to be Analyzed*

(Hz) (PSI) 1 2 3 0.1 0.29 12 0.25 NONE 23 0.32 34 45 1.86 1.20 0.24 56 1.05 2.73- 0.48 67 0.49 0.42 0.99 73 ~-4 c.,u~ 0.59 0.38 0.30 89 0.59 0.38 0.30 910 0.59 0.38 0.30 ZE 1011 0.34 0.79- 0.18

~o ~

1112 0.15 0.45 0.12 1213 0.17 0.12 0.11 1314 ~ 0.12 0.08 0.08 1415 0.06 0.07 0.03 1516 0.10 0.10 0.02 1617 0.04 1718 0.04 1819 0.04 1920 0.27 2021 0.20 NONE 2122 0.30 2223 0.34 2324 0.33 2425 0.16

• Half range (1/2 of peak to peak amplitude) 4.4.+/-8 Revision 2

Table 4.4.12 (Continued)

CONDENSATION OSCILLATION BASELINE RIGID WALL PRESSURE AMPLITUDES ON TORUS SHELL BOTTOM DEAD CENTER Alternate Amplitudes to be Analyzed*

Frequency Amplitudes (PSI)

Range to be Analyzed*

(Hz) (PSI) 1 2 3 2526 0.25 2627 0.58 2728 0.13 2829 0.19 2 930 0.14 3031 0.08 3132 0.03 3233 0.03 3334 0.03 3435 0.05 3536 0.08 3637 0.10 3 738 0.07 NONE 3839 0.06 3940 0.09 4041 0.33 4142 0.33 4 243 0.33 4 344 0.33 4445 0.33 4546 0.33 4647 0.33 4748 0.33 4849 0.33 4950 0.33

• Half range (= 1/2 of peak to peak amplitude) 4.419 Revision 2

ALTERNATE 2 NOTE: ALL AMPLITUDES REPRESENT ONE-HALF OF THE PEAK-TO-PEAK AMPLITUDE ALTERNATE 1 n

2 2

A PSO OF THIS SPECIFICATION (INCLUDING ALTERNATE 2) A YIELDS 90 PERCENT 2 VALUE) OF THE BELOW POWER (psI 25 Hz

-j a.

C A ALTERNATE 3 w w a-I a: 0 D

C I w 3 2:

a: a: a. t~1 a.

H 7 C w

0 J[Ihr~

N)

H a: a: H a.

17ff 0 0, 03 U, 03 w 03 a:

a.

0 i 10 15 5 10 15 I

w FREQUENCY ~I-Iz) FREQUENCY (Hz) FREQUENCY (Hz) 0

-j I

a SELECT ONE ALTER.

w NATE WHICH GIVES THE MAXIMUM E I

a: TOTAL RESPONSE (U

a:

a-0 0 5 10 IS 2-fliTh 20 25 lTHL~-riVJI 30 35 40 11111111 50 4 FREQUENCY (H?)

H Figure 4.4.11. Condensation Oscillation Baseline Rigid Wall Pressure ha Amplitudes on Torus Shell Bottom Dead Center

A = LOCAL PRESSURE OSCILLATION AMPLITUDE WET WELL AIR SPACE AMAX - MAXIMUM PRESSURE OSCILLATION AMPLITUDE (AT TORUS BOTTOM DEAD CENTER)

A FREE SURFACE AMAX -

z (11

-a:- SUPPRESSION POOL 0

-a:-

3 I-I-3 C C

(U 4

H 0,

I-.

0 Figure 4.4.12. Mark I Condensation Oscillation Torus Vertical Cross Sectional ha Distribution for Pressure Oscillation Amplitude

POOL-TO-VENT AREA RATIO POOL FREE SURFACE AREA (INCLUDING VENTS VENT EXIT CROSS SECTIONAL AREA MULTIPLICATION FACTOR = PLANT UNIQUE LOAD AMPLITUDE PROTOTYPICAL RIGID WALL LOAD AMPLITUDE 1.1 1.0 0.9 U.

I-U, U.

0 0.8 I-0 w

N 0.7 4

a:

0 0.6 4=- z IL

-a:- 0 I

U H 0.5 0 C

U.

H ha ha z H 0 03 p4 0.4 C 03 LI

3 a

F

-J 0.3 0.2 0.1 0

15 20 25 30 35 - 40 45 50 55 (U POOL-TO-VEj~T AREA RATIO 4

0, I H

0 Figure 4.4.13. Mark I Condensation Oscillation Multiplication Factor Versus PooltoVent Area ha Ratio for Plant Unique Load Determination

NEDO-2 1888 REFERENCES FOR SECTION 4.4.1 4.4.11 Mark I Containment Program Full Scale Test Program, Final Report, General Electric Company, Report No. NEDO24539, August 1979.

4.4.113/4.4.114 Revision 2

NEDO21%8 4.4.2 Loads on Submerged Structures Due to Main Vent Steam Condensation Oscillations Steam condensation begins after the vent is cleared of water and the drywell air has been carried over into the suppression chamber. The condensation oscillation phase induces bulk water motion and therefore creates drag loads on structures submerged in the pool.

4.4.2.1 Bases and Assumptions The basis of the flow model for condensation oscillation load definition is derived from the work in References 4.4.21 and 4.4.22. The major assump-tions are sunmiarized as follows:

a. The total drag is the sum of standard and acceleration drags (see Ttem g for exceptions).

b. The submerged structures are assumed to be rigid.

c. Condensation oscillation is modeled as a fluid source located at the vent exit, which induces a bulk pool motion that results in drag forces on submerged structures.

d. The actual flow field can be represented bya locally uniform flow field.

e. The presence of boundaries is accounted for by using the method of images. The circufar cross-section of the torus shell is modeled by a rectangular crosssection (Model E specified in Reference 4.3.82).

f. The average strength of the fluid source for condensation oscillation is determined fran wall load measurements (Reference 4.4.2-3) given in Section 4.4.1 by using potential flow theory in an incompressible fluid. A maximum source strength is defined as twice the average source strength.

4.4.21 Revision 2

NEDO- 21888

g. The standard drag forces on circular cylinders are computed by using a conservative drag coeffieient of C = 3 6 which bounds d

the lift force drag coeffieient as experienced in test data for U T/D >2.74, where U is the maximum velocity, T is the period in In of bubble oscillation and D is the cylinder diameter. For U T/D < 2.74, the drag forces shall be computed on the basis in of acceleration drag alone.

h. Drag forces on structures with sharp corners (e.g., rectangular and I beams) are computed by considering forces on an equivalent cylinder of diameter D eq ,f2 Lmax where Lmax is the maximum transverse dimension and is defined as the diameter of a circum-scribed cylinder about the cross-section of the structure.

1.. Long slender structures are modeled in segments of length (L),

which do not exceed its diameter (D or D ). The total force eq on the structure is obtained by integrating the individual seg-ment force over the entire length of the long structure.

j. Interference effects due to the proximity of solid boundaries are considered for each segment that has its center less than 1.5 diameters from the boundary. ?Iultipliers are needed to increase both the acceleration and the standard drag. Similarly, multipliers are needed to include the interference effects be-tween neighboring structures where the centers of the segments are less than 3D (D = (D + D )/2, the average diameter of the 1 2 two structures).

For structures near walls, the multiplier (l+Aw ) shall be used to in-crease the acceleration drag and the multiplier (l+D~)shall be used to increase the standard drag. Bounding expressions for A and D w w 4.4.2la Revision 2

NEDO-21888 are given below as functions of x Cx = r/D - 1/2, where r is w w the distance from the segment center to the boundary and D is the diameter of the structure) 0.05<X <1.0 A = 0.05/X w w w D = O.12/X w w x~ 0.05 A w

= 1.0 ow = 2.4 For structures with neighbors that are less than 3D away and within C 30 of being parallel, the multiplier (l+A )shall be used to increase I

the acceleration drag and the multiplier (l-4-D ) shall be used to I

increase the standard drag. Bounding expressions for and D I are given below as functions of X CX = r /~- 1, where is the I I 12 distance between segment centers) 0.05<x ~2.O A 0.2 D I I D = 0.2/X I I where 0 is the diameter of the structure under consideration and 1

D is the diameter of the neighbor. If more than one neighbor must 2

be considered, th~ A and D values may be summed over the neighbor I I structures. For X1 less than 0.05, the two neighbor structures shall be considered as an effective single structure.

The effects of wall proximity and neighbor structures may be super-imposed in order to compute overall multipliers as follows:

(l+A +~A )

w k 1K (1+ D + ~ D) w kIK 4.4.2lb

NEDO-21888 4.4.22 Load Definition Detailed procedures to calculate drag loads on submerged structures are described in Reference 4.4.21. This reference provides the methodology to utilize the unsteady velocity and acceleration flow fields from condensation oscillation to obtain drag loads on submerged structures, such as pipes, beams, etc. in the suppression pool.

The load definition first requires the establishment of the velocity and accel-eration flow fields within the torus by simulating condensation oscillations at the vent exits. Then, drag loads on the submerged structure or sections of the submerged structure are calculated based on the previously established transient flow fields. The drag loads obtained from the condensation oscilla-tion source frequency range are summed to obtain the total load.

a. Flow Field Establishment Condensation oscillations are described as fluid sources located at downcomer vent exits. The average source strengths are determined from wall load measurements. By using potential flow theory and the method of images to account for the effects of solid walls and the free surface, the velocity and acceleration flow fields within the torus are established. The load definition for each structure is computed on the basis of both the average source at all downcomers (oscillating in-phase) and for the maximum source applied at the nearest dowocomer. The fluidstructure interaction effects are included when the local fluid acceleration is less than twice the boundary acceleration. This may be accomplished by adding the boundary acceleration to the local fluid acceleration.

b. Drag Loads Evaluation The drag force on a submerged structure consists of two components:

the standard drag and the acceleration drag. The standard drag is computed on a locally uniform flow field evaluated at the local velocity. The acceleration drag load is evaluated based on an invis 4.4.22 1~ev1sion 2

NEDO-21888 cid, uniform but unsteady (accelerating) flow field. The sum of these two drag forces gives the total drag load on a sub-merged structure. These loads may be applied quasistatically to submerged structures, only if the highest significant Fourier components occur at frequencies less than half the lowest structural frequency.

4.4.22 a Revision 2

NEDO-2 1888 4.4.2.3 Selection of Key Parameters for Load Evaluation Procedures This section outlines basic criteria for selecting key parameters to be used as input data to the model (Reference 4.4.2-1) so that condensation oscilla-tion loads on submerged structures can be properly obtained.

The plant specific suppression pool geometry should be identified first as follows:

a. Torus shell dimensions

b. Torus water depth

c. Location of submerged structure considered (Figure 4.3.81)

d. Locations of downcoiners (Figure 4.3.81).

Then the condensation oscillation source information calculated from wall load measurements (Reference 4.4.23) is assigned to the downcomer exits.

The submerged structure is divided into the appropriate number of sections for more precise calculation.The coodinates of each section center, orientation of the structure axes and the structure projected crosssectional area should be identified. The acceleration drag volume for each section of the submerged structure should be obtained from Tables given in Reference 4.4.22.

With the input parameters identified and calculated, the submerged structure drag load model is used to evaluate the resultant transient drag forces on each structure element considered. Typical results are shown in Figures 4.4.21 and 4.4.22. These figures show the condensation oscilla-tion drag forces on an S/RV line (with TQuencher arms) in the radial (X) and tangential (Z) directions as a function of time.

4.4.2-3/4.4.24 Revision 2

NEDO21888 CONDENSATION OSCILLATION FORCE ON S/RV LINE AT 5.5 Hz a

0 x

U.

C, 0

LI.

C

-J C

1-0 5-0 0.1 0.2 0.3 OA 0.5 TIME (s.d Figure 4.4.21. Sample Predicted Time History of Total XForce on S/RV Line 4.4.25 Revision 2

NEDO215888 CONDENSATION OSCILLATION FORCE ON S/RV LINE AT 5.5 Hz 0

x w

U 0

U.

N

-J 0

0 0.1 0.2 0.3 0.4 0.5 TIME ~sec)

Figure 4.4.22. Sample Predicted Time History of Total ZForces on S/RV Line 4.4.26 Revision 2

NEDO21888 References for Section 4.4.2 4.4.21 Lasher, L.E., Analytical Model for Estimating Drag Forces on Rigid Submerged Structures Caused by Condensation Oscillations and Chugging Mark I Containments, General Electric Company, Report No. NEDO25070, April 1979.

4.4.22 Moody, F.J., Chow, L.C., and Lasher, L.E., Analytical Model for Estimating Drag Forces on Rigid Submerged Structures caused by LOCA and Safety Relief Valve Air Discharges, Ramshead, General Electric Company, Report No. NEDO21471, September 1977.

4.4.23 J.E. Torbeck, et al., Mark I Containment Program, Full Scale Test Program Final Report, General Electric Company, Report No. NEDO 24539, August 1979.

4.4.27/4.4.28 Revision 2

NEDO21888 4.4.3 Downcomer Dynamic Load The downcomers (D/C) experience loading during the condensation oscillation (CO) phase of a blowdown. The procedure for defining the dynamic portion of this loading for both a Design Basis Accident (DnA) and an Intermediate Break Accident (IBA) is presented in this section. Condensation oscillation loads do not occur for the Small Break Accident (SBA) case. The vent system thrust loads due to the quasistatic pressurization of the containment and the flow in the vent system during the condensation oscillation regime are discussed in Section 4.2.

4.4.3.1 Bases and Assumptions The basis for the DBA and IBA condensation oscillation dynamic load definition is the data obtained from the instrumented D/Gs of the Mark I FullScale Test Facility (FSTF) during testing as described in References 4.4.31 and 4.4.32.

The load definition is developed for, and is directly applicable t6, D/C pairs which are tied (i.e., connected by lateral bracing) or where the down comer to vent header intersection (DC/VH) is stiffened with gussets or other means.

4.4.3.2 Downcomer Dynamic Load (DBA)

The downcomer dynamic load involves two components:

a. Internal pressure load of equal magnitude in each D/C in a pair

b. Differential pressure load between D/Cs in a pair Figure 4.4.31 shows a typical D/C and the pressurization loads. The total dynamic load on the D/C results from F and F YB which are the product of the internal pressure and the unbalanced area at the D/C miter joint. The D/C loading is based on the average internal D/C pressure load, which produces a vertical loading (F ) and the differential pressure observed between the YB downcoiuers in a pair which results in the swinging motion of that pair (F).

4.4.31 Revision 2

NEDO21888 The loading definition accounts for uncertainties in determination of the DC/VH natural frequencies and shifting of the DBA CO dominant driving fre-quency in the 48 Hz range. A second and third harmonic are included. At least 80% of the DC/VH strain response results from the dominant frequency and not more than 15% and 5%from the second and third harmonics, respectively.

The internal pressure load in the D/C was obtained by evaluating the pressure data from the three liquid blowdown DRA tests run in the FSTF. These tests had the largest CO loads and the period of each test showing the highest pressures was used. For each 1sec segment of data, a PSD was obtained from the pressure measurement for each D/C. These PSD pressure values were aver-aged over the total time period selected for evaluation and were then averaged for all 8 0/Cs. This process provided an average D/C pressure for three frequency ranges 08, 816, and 1624 Hz*. These pressures were summed for the three frequency ranges.

The defined 0/C internal pressure loads for the DBA CO are as follows:

Applied Frequency*

Frequency Pressure Range Dominant 3.6 psi 48 Hz Second harmonic 1.3 psi 816 Hz Third harmonic 0.3 psi 1224 Hz Figure 4.4.32 shows the internal pressure load and the three frequency bands over which they are applied.

The application frequency bands are established as follows. The dominant frequency of the D/Cs will be established by analysis of the plant unique configuration. The dominant frequency is the D/C response mode with the highest participation factor in the 48 Hz range which is the dominant CO range. The specified D/C internal pressure dominant frequency load will be

• The sums for the DBA D/C internal and differential pressure CO loads cover the data ranges 08, 816, and 1624 Hz to sum all the energy in the pressure load. However, the loads are applied over three harmonic ranges, 48, 816, and 1224 Hz, as indicated.

4.4.32 Revision 2

NEDO21888 applied at the dominant frequency of the D/C in the 48 Hz range. The second harmonic D/C internal pressure will be applied at a frequency of twice the dominant frequency. The third harmonic D/C internal pressure will be applied at a frequency three times the dominant frequency. The frequency ranges for application of the dominant, second, and third harmonics are shown in Figure 4.4.32.

Some Mark I plants may not have a significant D/C response in the 48 Hz range after the DC/VH intersections are stiffened. In this case the Architect Engineer may elect to use the frequency with maximum participation in the 816 Hz range for application of the 1.3 psi (second harmonic) DBA CO load. The dominant frequency load of 3.6 psi will be applied at one half the above frequency and the third harmonic load of 0.6 psi at 1.5 times the above frequency.

The data base selected for the D/C differential pressure load definition included data from all 8 D/Cs during the period when the highest differential pressures were observed. These test periods are the same as those used for the D/C internal pressure.

For the selected tests and time periods, time histories were obtained for each of the four D/C pairs (i.e., 12, 34, 56, and 78). PSDs of the D/C differential pressure were generated for each 1sec time period in the overall time period for each test. The power in each of the three frequency bands (08, 816, and 1624 Hz)* was lumped for each 1 second of data. The lumped power was converted into an equivalent single differential harmonic pressure to be applied in each of the three frequency bands. The D/C differential pressure loads for the DBA CO period are as follows:

Applied Frequency Frequency Pressure Range*

Dominant 2.85 psi 48 Hz Second harmonic 2.6 psi 816 Hz Third harmonic 1.2 psi 1624 Hz

• The sums for the DBA DIG internal and differential pressure CO loads cover the data ranges 08, 816, and 1624 Hz to sum all the energy in the pressure load. However, the loads are applied over three harmonic ranges, 48, 816, and 1224 Hz, as indicated.

4.4. 33 Revision 2

NEDO21 888 Application of the dominant second and third harmonic differential pressures is the same as for the internal pressure application previously discussed.

Figure 4.4.33 shows the differential pressure amplitudes and frequency ranges.

Figure 4.4.34 shows how the dynamic loads for the DBA CO defined above are to be applied to the different downcomer pairs on a representative Mark I vent header system. The internal pressure load (Figure 4.4.32) is to be applied to all D/C pairs. For Load Case 1, the differential pressure is to be applied in phase with the internal pressure and in phase on all D/C pairs acting on the inside of the vent header. For Load Case 3, the load application is the same, except that the AP on the D/C pair nearest the centerline of the vent bay is 1800 out of phase with the other D/C pairs. Load Cases 5 and 7 follow and are shown in Figure 4.4.34. Load Cases 2, 4, 6 and 8 are the mirror images of the respective oddnumbered load cases.

The specification of these eight load cases covers the various possible differential pressure cases that could exist in Mark I plants. The load~

applications are applicable to plants with only three fl/c pairs in ~ non.~

vent bay by moving the one D/C pair to the centerline. For Duane Arnold (i.e.,

single vertical D/C) there are no differential pressure loads.

The total response of the DC/VH intersection to the CO dynamic load is the sum of the responses from the internal and differential pressure components plus any other contributions. All eight load cases are to be evaluated and the case with the maximum response is to be used for design.

4.4.3.3 Downcomer Dynamic Load CIBA)

The downcomer dynamic load due to CO during an IBA was established in a manner similar to that used for the DBA. The exceptions were as follows:

a. FSTF tests M4 and MS provided the data base.

b. The dominant frequency load for the IBA CO was established at 610 Hz based on FSTF data.

4.4.34 Revision 2

NEDO21 888 The D/C internal pressure dynamic load due to CO during an IBA is as follows:

Applied Frequency Frequency Pressure Range Dominant 1.1 psi 610 Hz Second harmonic 0.8 psi 1220 Hz Third harmonic 0.2 psi 1830 Hz Figure 4.4.35 shows these D/C internal pressure load values and the range of application.

The D/C differential pressure dynamic load due to CO during an IBA is as follows:

Applied Frequency Frequency Pressure Range Dominant 0.2 psi 6lQ Hz Second harmonic 0.2 psi 1220 Hz Third harmonic 0.2 psi 1830 Hz The frequencies for application of the dnamic IBA load are the same as those for the IBA internal pressure load.

The load cases for the IBA loads are the same as for the DBA loads; there fore, Figure 4.4.34 can be used.

4.4. 35 Revision 2

NEDO21888 REFERENCES FOR SECTION 4.4.3 4.4.31. Mark I Containment Program FullScale Test Program, Final Report, General Electric Company, Report No. NEDO24539, August 1979.

4.4.32. Mark I Containment Program Full Scale TEst Program Evaluation of Supplemental Tests, General Electric Company, Report No.

NEDO24539, Supplement 1, July 1981. I 4.4.36 Revision 2

DICA D/CB P-(P A P B I P-0 PPB P-PB IS

-a:-

(h)

$ z t,1 0

N)

C C

C Fya Fl F

XB (U

4 0,

H SWING MOTION DUE TO AP 0 VERTICAL LOADING DUE TOP El ha Figure 4.4.31. Downcomer Dynamic Load

5 RANGE OF FUNDAMENTAL (F 1) 4--

3.6 psI NOTES:

A 1. THE AMPLITUDES SHOWN ARE HALF RANGE (ONE-HALF OF THE PEAK TO PEAK VALUE).

a 2. F1 IS THE SINGLE FREQUENCY IN THE RANGE 3 OF THE FUNDAMENTAL.

w a: 2:

0 4=- ii, a: N)

L3 a I-I IL C C W 2 C 03 0

U RANGE OF SECOND HARMONIC (2xF1) z 0

o 13 psI RANGE OF THIRD HARMONIC (3xF1) 1 0.5 psi 0

8 12 16 24 (U

4 I (Hz) 0, 0

El ha Figure 4.4.32. Downcomer Pair Internal Pressure Loading for DBA CO

5 LU IL 4

XL 2E5 psI RANGE OF FUNDAMENTAL (F NOTES:

1)

1. THE AMPLITUDES SHOWN ARE HALF RANGE (ONE-HALF OF THE PEAK TO PEAK VALUE).

2- F1 IS THE SINGLE FREQUENCY IN THE RANGE OF THE FUNDAMENTAL.

a:

a 3

~1 z

I zw t-I a:

w 0

U.

U N)

C IL 2 C LU C 0 RANGE OF SECOND HARMONIC I2xF1I U

2 2.6 psI RANGE OF THIRD HARMONIC I3xF1) 1 1 .2 psi (U

4 H

0, 3.

0 (Hz) 16 I

24 0

El tO Figure 4.4.33. Downcomer Pair Differential Pressure Loading for DBA CO

NEDO21888 NOTES

1. . 0/C WITH INITIAL DIFFERENTIAL PRESSURE LOAD.

2. ALL DICs HAVE INTERNAL PRESSURE LOAD IN PHASE WITH DIFFERENTIAL PRESSURE LOAD

3. ANALYZE ALL EIGHT CASES USE WORST RESPONSE CASE VENT BAY VENT BAY CENTERLINE CENTER LINE NON VENT BAY CENTERLINE VENT SAY CASE 3 -~- ~ CASE 4 VENT SAY CENTERLINE / CENTERLINE a.

1 j~

NONVENTBAY VENT BAY CENTER LINE /

CASE5-~- ~NE-CASE6 CENTERLI VENT BAY CENTERLINE

//

U U VENT SAY CENTERLINE Figure 4.4.34. Dowocomer CO Dynamic Load Application 4.4.310 Revision 2

RANGE OF FUNDAMENTAL (F 1)

1.2 NOTES

1. THE AMPLITUDES SHOWN ARE HALF RANGE 1.1 psi (ONE-HALF OF THE PEAK TO PEAK VALUE).

2. F1 IS THE SINGLE FREQUENCY IN THE RANGE OF THE FUNDAMENTAL.

1.0 RANGE OF SECOND HARMONIC (2xF1) a: w

-a:-- a:

U, 0.8 F 0.8 ps U, z H w a:

H a. 0 IL iS w N)

H

-IS 0 06 C U C C--, z C I-A 0 ha 0 0.4 RANGE OF THIRD HARMONIC (3xF1) 0.2 psI 0.2 p.

(U 4

0 0 6 10 12 II 18 20 (Hz)

I I 30 H

0, H

0 El ha Figure 4.4.35. Downcomer Internal Pressure Loading for IBA CO

NEDO21888 4.4.4 Vent System Loads Oscillating loads on the vent system during the condensation oscillation phenomenon are caused by harmonic pressure oscillations superimposed on the prevailing local static pressures in the vent system. The vent system includes main vents, the vent header and downcomers. In the following subsections the definition of these loads is presented on a generic basis, along with a dis-cussion of the bases and assumptions used.

4.4.4.1 Bases and Assumptions The basis for the vent system internal pressure loading due to condensation oscillation is the data from FSTF tests M8, MllB, and M12, the large liquid break tests which simulated the DBA conditions (Reference 4.4.41 and 4.4.42).

An inspection of the PSDs of pressure signals observed at various locations in the vent system indicates the existence of a coimnon dominant frequency. This enables the DBA forcing function for the vent system to be defined as sinusoidal.

The pressure signals from FSTF tests M8, MllB, and M12, for the vent, vent header, and downcomers were analyzed in detail and a load definition was postulated. The postulated load definition was applied to a detailed finite element model of the FSTF vent system and predicted dynamic responses were compared to the measured response (Reference 4.4.43). On this basis, the vent system loads during con-densation oscillation are defined as harmonic pressures superimposed on the prevailing static pressure,.

For the lEA, the data from FSTF tests M4 and M5 (Reference 4.4.41) was analyzed and a load definition postulated.

Major assumptions made in defining vent system condensation oscillation loads include the following:

a. The vent system load values determined from the FSTF data are generic and can thus be applied to all Mark I plants. This is justifiable because the source strength for condensation oscillations is mainly 4.4.41 Revision 2

NEDO21888 controlled by parameters such as the downcomer diameter and the thermodynamic conditions at the downcomer exit which are similar for all Mark I plants.

b. The variation in the dominant frequency observed in the FSTF data is assumed to be applicable to all Mark I plants. This is justifiable since the condensation oscillation frequency is primarily controlled by thermodynamic and geometric conditions at the vent exit. These conditions are similar for all Mark I plants and were simulated in the FSTF tests. Minor adjustments to account for vent system acoustic frequency were made to conservatively represent all Mark I plants.

4.4.4.2 Load Definition Condensation oscillation loads are specified for all three components of the vent system: main vents, the vent header, and downcomers. These loads, as determined from the FSTF data, are generic and are thus directly applicable to all Mark I plants.

Main Vent and Vent Header DBA IBA Amplitude +/-2.5psi +/-2.5psi Frequency Range At frequency of maximum At frequency of maximum response in 48 Hz range response in 610 Hz range Forcing Function Sinusoidal Sinusoidal Spatial Distribution Uniform Uniform Downcomers Amplitude +/-5.5psi +/-2.1psi Frequency Range At frequency of maximum At frequency of maximum response in 48 Hz range response in 610 Hz range Forcing Function Sinusoidal Sinusoidal Spatial Distribution Uniform Uniform The pressures in the downcomer were developed in Section 4.4.3. Since the natural frequency of the downcomer in the hoop mode is very high when compared to the 4.4.42 Revision 2

NEDO2 1888 pressure load frequency, the dominant, second, and third harmonics were summed and shall be applied as the single value indicated above. The pressure loading should be used to calculate only the circumferential structural response (e.g.,

hoop stress) of the downcomers.

The vent system responses to dynamic thrust, or other loads, which are trans-mitted through the downcomers to other components, are evaluated using the dowocomer condensation oscillation dynamic load defined in Section 4.4.3 and the vent system thrust loads in Section 4.2.

In addition to the oscillating pressure defined above, a uniform static pressure should be applied in the vent, vent header, and the downcomer to account for the nominal submergence of the downcomers in the plant being analyzed.

Structural responses from the static and dynamic pressures should be summed.

The condensation oscillation pressure loads should be used for the global vent system response. No separate dynamic load or vent system loads are defined. for the condensation oscillation phase of the blowdown.

REFERENCES FOR SECTION 4.4.4 4.4.41. Mark I Containment Program FullScale Test Program Report, Final Report, General Electric Company, Report No. NEDE24539P, March 1979.

4.4.42. Mark I Containment Program Full Scale Test Program Evaluation of Supplemental Tests, General Electric Company, Report No.

NEDE24539P, Supplement 1, May 1981.

4.4.43. Mark I Containment Program Analysis of the FSTF Vent System for Condensation Oscillation Loads, General Electric Company, Report No. NEDE24838P, July 1981.

4.4.43/4.4.44 Revision 2

NEDO21888 4.5 CHUGGING LOADS Chugging occurs during a postulated lossofcoolant accident (LOCA) when the steam flow through the containment vent system falls below the rate necessary to maintain steady condensation at the downcomer exits. The corresponding flow rates for chugging are less than those of the condensation oscillation phenomenon defined in Section 4.4. During chugging steam bubbles form at the downcomer exits, oscillate as they grow to a critical size (v~ downcomer diameter), and begin to collapse somewhat independently in time. The result-ing load on the torus shell consists of a low frequency component which cor-responds to the oscillating bubbles at the downcomer exits as they grow, and higher frequency components corresponding to the collapsing bubbles.

The loads specified for chugging are based on the FSTF tests. In these tests a prototypical segment (one bay) of a Mark I torus and vent system were sub-jected to ten steam and liquid blowdowns simulating a range of LOCAs.

Duration of the chugging period varies with break size and is specified in Table 4.5.11. Load combinations are given in the bar charts in Section 3.0.

Chugging is a low flow rate phenomenon which is not break size dependent.

Therefore, the loadings presented in the following subsection apply to all the break categories specified.

Loads on the structures within the Mark I Containment system that may be induced by chugging are defined in the following subsections:

4.5.1 Torus shell loads 4.5.2 Loads on submerged structures 4.5.3 Lateral loads on downcomers 4.5.4 Vent system loads.

4. 51/4. 52 Revision 2

NEDO-21888 4.5.1 Torus Shell Loads During the chugging regime of a postulated LOCA, the chugging loads on the torus shell occur as a series of chug cycles, each of which can be organized into two elements: the prechug and the postchug portions. The prechug portion is described as follows. Initially, as the steamwater interface enters the pool, a relatively low frequency (.7 Hz) pressure loading is.,

observed on the torus shell. This low frequency corresponds to the frequency of oscillation of the interface and may persist for several of these low frequency cycles. The interface eventually becomes unstable and breaks up producing a rapid underpressure as the chug occurs. In a chug cycle, the interface breakup occurs in several, but not all of the downcomers and takes place within a time interval usually not exceeding 0.1 seconds. Data from the FSTF tests show that for a given chug cycle chuRs will seldom occur in all of the downcomers, and the chugs in the individual downcomers are never exactly synchronized.

The post chug portion of the chug cycle is a system response (ring out) to the rapid under pressure caused by the break up of the steamwater interface.

The post chug occupies approximately the same time span as the prechug in a given chug cycle.

The resulting response of the torus shell to a chug cycle is a low frequency prechug oscillation prior to the postchug, followed by a higher frequency ring out of the torus shellpool water system in response to the impulsive underpressure after the postchug begins. In the process of the interface collapse, steam bubbles may also be pinched off and collapse producing a very localized short duration pressure load. Since these short duration loads are not transmitted to the torus support system and remain very localized, they are not considered structurally significant and have not been included in these specifications.

4.5.13 Revision 2

NEDO-21888 4.5.1.1 Bases and Assumptions The basis for the torus shell chugging load definition is the chugging data obtained from the FSTF. The parameters which were varied in the FSTF tests include break sizes, dowocomer submergence, initial poo1 temperature, blow down fluid phase (liquid and vapor), and initial wetwell pressure. The load-ing conditions established from the FSTF data and the use of conservative application techniques result in bounding chugging loads for all Mark I plants. The FSTF test matrix and test results are described in Reference 4.5.11.

The FSTF blowdowos which simulated the small steam break accidents produced the most severe chugging loads, and the data from these tests were used as a basis for the chugging load specifications. The FSTF blowdowns which simulated the large steam and liquid break accidents did not exhibit large chugging like behavior: the condensation process observed during these blowdowns appeared to be more continuous.

The onset times and durations specified in Table 4.5.11 are based on FSTF test results and an analytical study discussed in subsection 4.4.1.1. The onset time of chugging for the DBA is based on the time at which condensation oscillations were determined to end (see Table 4.4.11). The onset time for the lEA is consistent with the onset time specified in the condensation oscillation subsection, since~ chugging loads are specified in lieu of conden-sation oscillations. The onset time for the SEA is consistent with the time required to purge the air from the drywell. The durations specified for all break sizes were determined by the analytical modeling methods discussed in subsection 4.4.1.1.

As noted earlier, the chugging load cycles observed in the FSTF tests are divided into a prechug and a postchug portion. Initially, a prechug low frequency sinusoidal oscillation appears on the torus wall. This is related to oscillations of the steamwater interface. The frequency of these oscil-lations is relatively constant in all of the FSTF chugging data, but the fre-quency is related to the vent acoustic frequency, which varies from 4.5.14 Revision 2

NEDO-21S88 plant to plant. This load specification contains a range of frequencies for the application of this initial load which is consistent with the range of vent system acoustic frequencies calculated for Mark I Containments (6.9 to 9.5 Hz). The single frequency which produces the highest system response in this range is specified to be used for evaluation purposes.

The low frequency prechug oscillation is followed by an impulsive load as the rapid condensation event occurs. In the FSTF data, the torus shell loads in this postchug portion of the chug cycle appear as a combination of pressure oscillations at the natural frequencies of the torus shell combined with oscillations at the acoustic frequencies of the downcomers (%45 Hz). This load specification contains a range of frequencies for the application of the postchug load which is consistent with the range of downcorner acoustic fre-quencies for Mark I Containments (40 50 Hz). Figure 4.5.11 shows a typical chug cycle with the prechug and postchug portions identified.

To characterize the prechug and postchug torus shell behavior, Power Spectral Density (PSD) analyses have, been performed on the torus wall pressure traces.

These data have been used to develop the load specification for the torus shell.

The basis of the pressure amplitude specification is several segments of maximum pressure amplitude data from FSTF tests Ml, M4, and M9. Portions of individual tests are used for the two parts of the torus load specification as follows:

• Prechug: Test M9 Entire Test

• Postchug: Tests Ml, 18 Maximum Pressure Amplitude Events M4 and M9 In order to apply the FSTF measured loads to plant unique geometries, an adjustment must be made to account for PSI effects in the FSTF data. Struc-tural response effects unique to the FSTF are accounted for by the specifica.

tion of a rigid wall load (pressure oscillation amplitude as a function of frequency).

4.5.15 Revision 2

NEDO21888 The rigid wall chugging load for Mark I Containments is derived from the FSTF prechug and postchug pressure response data using the following procedure:

a. A finite element analytical model has been developed for the FSTF test facility. This model includes the torus shell support struc-tures and the pool water. Torus shell pressure load amplitudes relative to chug source pressures are determined as a function of frequency using this model of the FSTF. This defines a relationship between pressure loads on the torus wall and a pressure source at the vent exits.

b. The measured prechug and postchug frequency response from the FSTF data is then used with this relationship to develop chugging source definition. The use of the coupled fluidstructure model of the FSTF allows the removal from the chug source those elements of the chug frequency spectrum that are contributed by the structural response of the FSTF.

c. Using the vent source chug definition, rigid wall chug pressures are defined. The rigid wall loads are obtained from the prechug and postchug vent source pressures by applying a spatial attenuation factor to transfer the pressures from the vents to the torus wall.

The resulting distribution around the wetted perimeter of the torus cross section is shown in Figure 4.5.12. This pressure distribution is based on an incompressible flow analysis of the torus geometry.

For the prechug portion of the chug cycle, both symmetric and asymmetric loading specifications have been developed to conservatively account for any randomness in the chugging phenomena. The symmetric loading is based on the average of the FSTF data. The asymmetric loading is based on both low and high amplitude chugging data conservativeby distributed around the torus in order to maximize the asymmetric loading. ~

4.5.16 Revision 2

In order to bound -the postchug portion of the chug cycle, symmetric loads based on the peak average values from the FSTF data have been specified.

Asymmetric loads have not been specified since any azimuthal response would be governed by the asymmetric prechug low frequency load specification.

The frequency range in the pressure amplitude specification has been estab-lished to cover the range of frequencies observed in the FSTF tests and the range expected in all Mark I plants. The vent system acoustic natural fre-quency range for all Mark I plants is 6.9 to 9.5 Hz, and is dominant in the prechug portion of the chug cycle. The downcomer acoustic natural frequency range for all Mark I plants is 40 to 50 Hz, and is dominant in the postchug portion of the chug cycle.

lt is intended that the prechug rigid wall loading be applied to a flexible wall coupled fluid structural model for symmetric and nonsymmetric loading cases. The pressure-distribution for the latter is illustrated on Figure 4.5.13. -

The postchug rigid wall loading is to be applied to a flexible wall coupled fluidstructural model for the symmetric case only.

For a comparison of the relative strengths of the two portions of a chug cycle, a PSD analysis of the specified loads has been performed. Comparing the power (psi2 value) from a PSD of a 1 Hz band width from the prechug specification to a corresponding value from the entire postchug specification yields the relative proportions of 90% prechug and 10% postchug.

The major assumptions in this definition of the torus chugging loads are:

a. The chugging behavior observed in the prototypical FSTF tests repre-sents the chugging behavior of the Mark I Containment. This is justifiable since the chugging source strength is primarily con-trolled by thermodynamic and geometric conditions at the vent exit.

These conditions are very similar for all Mark I plants and were simulated in the FSTF tests. Minor adjustments to account for vent 4.5.17 Revision 2

NEDO21888 system acoustic frequency were made to conservatively represent all Mark I plants.

b. Rigid wall chugging pressure loads can be derived from the flexible wall pressure measurements from the FSTF data by the use of the coupled fluidstructural model of the FSTF and incompressible flow theory.

c. Incompressible flow theory can be used to determine the torus cross section wetted perimeter pressure distribution from pressure sources at the vent exits. This is justifiable because of the good agreement between the analysis and FSTF data.

4.5.1.2 Load Definition Rigid wall torus chugging load definitions to be applied as a wall load to a structural model of the Mark I Containment are defined below. Because the load definitions are derived as rigid wall forcing functions it is intended that the torus fluid be considered in the structural evaluations. -It ~a1so is intended that the prechug anA postchug ~analyseabe steady state analyses.

PreChug Load Amplitude and Circumferential Two cases shall be evaluated Distribution independently:

Symmetric Distribution

+/-2.0psi uniform axially along the torus centerline at bottom dead center.

Asymmetric Distribution Values shown in Figure 4.5.13.

Vertical Cross Section Linear Attenuation with submergence Distribution along the wetted perimeter as shown in Figure 4.5.12.

4.5.1-8 Revision 2

NEDO-21888 Frequency The frequency producing the maximum response in the range from 6.9 to 9.5 Hz.

PreChug Cycle Duration 0.5 seconds every 1.4 seconds for the appropriate total duration defined in Table 4.5.11.

These loads are to be applied about the local static pressure at the appropri-ate times in the blowdown (see Table 4.5.11).

PostChug Load Amplitude Versus Frequency Values given in Table 4.5.12 (Also shown in Figure 4.5.14).

Total Response Resulting steady state responses from applying the amplitude at each fre-quency given in Table 4.5.12 are to be summed.

Spatial Distribution Uniform axially along the torus center-line. Linear attenuation with submerg-ence along the wetted perimeter at the torus cross section as shown in Figure 4.5.12.

PostChug Cycle Duration 0.5 seconds every 1.4 seconds for the appropriate total duration defined in Table 4.5.11. Prechug and postchug evaluations need not be combined.

These loads are to be epplied about the local static pressure at the appropriate times in the blowdown (see Table 4.5.11).

4.5.19/4.5.110 Revision 2

NEDO21888 Table 4.5.11 CHUGGING ONSET AND DURATIONS Onset Time Duration Break Size After Break After Onset DBA 35 seconds 30 seconds lEA 5 seconds 900 seconds SBA 300 seconds 900 seconds 4.5.111 Revision 2

NEDO2 1888 Table 4.5.12 POST-CHUG RIGID WALL PRESSURE AMPLITUDES ON TORUS SHELL BOTTOM DEAD CENTER Frequency Amp 1itude* Frequency Amplitude*

Range (Hz (PSI) Range (Hz (PSI) 01 0.04 2526 0.04 12 0.04 2627 0.28 23 0.05 2728 0.18 34 0.05 2829 0.12 45 0.06 2930 0.09 56 0.05 3031 0.03 67 0.1 3132 0.02 78 0.1 3233 0.02 89 0.1 3334 0.02 910 0.1 3435 0.02 1011 0.06 3536 0.03 1112 0.05 3637 0.05 1213 0.03 3738 0.03 1314 0.03 3839 0.04 1415 0.02 3940 0.04 1516 0.02 4041 0.15 1617 0.01 4142 0.15 1718 0.01 4243 0.15 1819 0.01 4344 0.15 1920 0.04 4445 0.15 2021 0.03 4546 0.15 2122 0.05 4647 0.15 2223 0.05 4748 0.15 2324 0.05 4849 0.15 2425 0.04 4950 0.15

• Half range (= 1/2 peak to peak amplitude) 4.5.1 12 Revision 2

PRE-CI-4UG PORTION POST-CHUG PORTION w

IL U,

U, LU 2:

IL EL el N)

H C

C C

p..

CYCLE REPEATS ONE CHUG CYCLE 0

4 TIME H

U, H

0 El N) Figure 4.5.11. A Typical Chug Average Pressure Trace on the Torus Shell

A = LOCAL PRESSURE OSCILLATION AMPLITUDE WET WELL AIR SPACE AMAX - MAXIMUM PRESSURE OSCILLATION AMPLITUDE (AT TORUS BOTTOM DEAD CENTER)

A FREE SURFACE AMAX - 0 4=

z U SUPPRESSION POOL 3 0

-A N) 4=

C C

C (U

4 H

0, H

0 Figure 4.5.12. Mark I ChuggingTorus Vertical Cross Sectional Distribution El tO for Pressure Amplitude

180 270 90 NOTE: THE AMPLITUDE SHOWN HERE REPRESENTS ONE-HALF OF THE PEAK.TO-PEAK AMPLITUDE 3

LU 0

-J a

4= 2 tTj U LU a: 0 N)

H I--

U a: C a C I C a:

LU I-z LU 0

0 0 U

In a:

0 I- -I 360 6 (degree)

(U 4

H 0,

H 0

El Figure 4.5.13. Mark I Chugging Torus Asymmetric Circumferential Distribution ha for Pressure Amplitude

NOTE: THE AMPLITUDE SHOWN HERE REPRESENTS ONE-HALF OF THE PEAK-TO-PEAK AMPLITUDE

-a:-

C-Il A LU H 2:

a:

H U, 0 U, 0 LU a: N) a H C

C 0

0 5 JtFh- C 10 15 20 25 30 35 40 45 50 FREQUENCY (Hz)

(U 4

H 0,

H 0 Figure 4.5.14. PostChug Rigid Wall Pressure Amplitudes on Torus El ha Shell Bottom Dead Center

NEDO21888 REFERENCES FOR SECTION 4.5.1 4.5.1-1 J.E. Torbeck, et al., Mark I Containment- Program, Full Scale Test Program Final Report, General Electric Company, Report No. NEDO24539, August 1979.

4.5.117/4.5.118 Revision 2

NEDO-2 1888 4.5.2 Loads on Submerged Structures due to Main Vent Chugging Steam chugging at the downcomers induces bulk water motion and therefore creates drag loads on structures submerged in the pool. The submerged structure load definition method for chugging follows that used to predict drag forces caused by condensation oscillations (Section 4.4.2) except that the source strength for chugging is proportional to the wall load measurement corresponding to the chugging regime.

4.5.2.1 Bases and Assumptions The basis of the flow model for chugging load definition derives from the work in References 4.5.21 and 4.5.22. The major assumptions are summarized as follows:

a) The total drag is the sum of standard and acceleration drags (see Item g for exceptions).

b) The submerged structures are assumed to be rigid.

c) Chugging is modeled as a fluid source located at the vent exit, which induces a bulk pool motion that results in drag forces on submerged structures.

d) The actual flow field can be represented by a locally uniform flow field.

e) The presence of boundaries is accounted for by using the method of images. The circular crosssection of the torus shell is modeled by a rectangular crosssection (Model E specified in Reference 4.5.24).

f) The strength of the fluid source for prechug is determined from wall load measurements (Reference 4.5.23) given in Section 4.5.1 by using potential flow theory in an incompressible fluid. A maximum source strength is defined as twice the average source strength. For postchug, 4.5.21 Revision 2

NEDO21888 the maximum source strength is obtained by using the maximum measured pressure in a Type 1 chug in equation B4 of Reference 4.5.21, with f(r) based on the single nearest downcomer. The local acceleration around a structure is computed on the basis of the two nearest down comers chugging at maximum source strengths. The phasing relation between the two nearest downcomers is established such that the maximum local acceleration results.

g) The standard drag forces on circular cylinders are computed by using a conservative drag coeffieient of C~ 3.6 which bounds the lift force drag coeffieient as experienced in test data for Um T/D >2.74, where Um is the maximum velocity, T is the period of bi.ibble oscillation and D is the cylinder diameter. For U T/D< 2.74, the drag forces shall be computed on the basis m

of acceleration drag alone.

h) Drag forces on structures with sharp corners (e.g., rectangular and I beams) are computed by considering forces on an equivalent cylinder of diameter D =~TL where L is the maximum eq max max transverse dimension and is defined as the diameter of a circum-scribed cylinder about the cross-section of the structure.

i) Long slender structures are modeled in segments of length (L),

which do not exceed its diameter (D or D ). The total force eq on the structure is obtained by integrating the individual seg-ment force over the entire length of the long structure~ Longer seg-ments are used as long as the equivalent uniform flow velocity and ac-celeration are evaluated conservatively for every point on the segment.

j) Interference effects due to the proximity of solid boundaries are considered for each segment that has its center less than 1.5 diameters from the boundary. Multipliers are needed to increase both the acceleration and the standard drag. Similarly, muVipliers are needed to include the interference effects be-tween neighboring structures where the centers of the segments are less than 3D apart, (D (D1 + D9/2, the average diameter of the two structures)

4. 5.2Ia Revision 2

For structures near walls, the multiplier (l+A ) shall be used to in-w crease the acceleration drag and the multiplier (l+D~2 shall be used to increase the standard drag. Bounding expressions for Aw and D w are given below as functions of x w Cxw = r/~ 1/2, where r is the distance from the segment center to the boundary and D is the diameter of the structure):

0.05 ~ X < 1.0 AV = 0.05/XV D = 0.12/X w w

xc 0.05 A = 1.0 w

D = 2.4 V

For structures with neighbors that are less than 3D away and within 300 of being parallel, the multiplier (l+A )shall be used to I increase the acceleration drag and the multiplier (l+D ) shall be used to I

increase the standard drag. Bounding expressions for A and D are I I given below as functions of X1 CX1 r12/~ 1, where r12 is the distance between segment centers):

0.05<X ~2.O AI X I = 0.2 (DlD~ D 2 I

DI 0.2/X I where D is the diameter of the structure under consideration and 1

D 2 is the diameter of the neighbor. If more than one neighbor must be considered, the and D values may be summed over the neighbor I

structures. For less than 0.05, the two neighbor structures shall be considered as an effective single structure.

4.5.2lb Revision 2

NEDO2 1888 The effects of wall proximity and neighbor structures may be super-imposed in order to compute overall multipliers as follows:

(1 + A + ~AIk)

(l+D w +~D k 1k )

k) The fluidstructure interaction is included for any structural segment for which the local fluid acceleration is less than twice the torus boundary acceleration. This is accomplished by adding the boundary acceleration to the local fluid acceleration.

4.5.2.2 Selection of Key Parameters for Load Evaluation Procedure This section outlines basic criteria for selecting key parameters to be used as input data to the model (Reference 4.5.21) 50 that chugging loads on sub-merged structures can be properly obtained.

The plant specific suppression pool geometry should be identified first as follows:

a) Torus shell dimensions b) Torus water depth c) Location of submerged structure considered (Figure 4.3.81) d) Locations of downcomers (Figures 4.3.81)

Then the chugging source information calculated from wall load measurements (Reference 4.5.23) is assigned to the downcomer exits.

The submerged structure is divided into the appropriate number of sections for a more precise calculation. The coordinates of each section center orientation of the structure axes and the structure projected crosssectional area should be identified. The acceleration drag volume for each section of the submerged structure should be obtained from tables given in Reference 4,3,22 4.5.22 Revision 2

With the input parameters identified and calculated, the submerged structure drag load model is used to evaluate the resultant transient drag forces on each structure element considered. The drag loads obtained from the postchug source frequency range are summed to obtain the total load. For prechug, these loads may be applied quasistatically to structures only if the highest significant Fourier components occur at frequencies less than half the lowest structural fre-quency. F or postchug, these loads shall be applied dynamically unless the lowest structural natural frequency times the duration of the spike in the source strength is greater than 3. Typical results are shown in Figures 4.5.21 and 4.5.22. These figures show the chugging drag forces on an S/RV line (with TQuencher arms) in the radial (X) and tangential (Z) directions as a function of time.

4.5. 22a Revision 2

NEDO2 1888 CHUGGING (PRECHUG) FORCE ON S/RV LINE AT 5.5 HZ 0

0 0

LL

-J iJJ C

I-i-

-J I.-

C I-0.00 0.20 0.40 0.60 TiME (SEC)

Figure 4.5.21 Sample Predicted Time History of Total XForces on S/RV Line 4.5.23 Revision 2

NEDO21888 CHUGGING (PRECHUG) FORCE ON S/RV LINE AT 5.5 HZ U-m

-J U)

U-,

C.)

C IL-

-J C

0.00 0.20 0.40 0.60 TiME (SEC)

Figure 4.5.22 Sample Predicted Time History of Total ZForces on S/RV Line 4.5.24 Revision 2

NEDO2 1888 References for Section 4.5.2 4.5.21 Lasher, L. E., Analytical Model for Estimating Drag Forces on Rigid Submerged Structures Caused by Condensation Oscillations and Chugging Mark I Containments, General Electric Company, Report No. NEDO25070, April 1979.

4.5.22 Moody, F. J., Chow, I. C., and Lasher, L. E., Analytical Model for Estimating Drag Forces on Rigid Submerged Structures caused by LOCA and Safety Relief Valve Air Discharges, Ramshead, General Electric Company, Report No. NEDO21471, September 1979.

4.5.23 J. E. Torbeck, et al., Mark I Containment Program, Full Scale Test Program Final Report, General Electric Company, Report-No.

NEDO24539, August 1979.

4.5.24 Submerged Structures Model Evaluation Report, Company, Report No. NEDO21983, September 1979.

(2eneral Electric I

4.5.25/4.5.26 Revision 2

4.5.3 Lateral Loads on Downcomers During the chugging phase of a postulated Loss of Coolant Accident (LOCA),

vapor bubbles are formed at the downcomer end which collapse suddenly and inter-mittently to produce lateral loads on the downcomer. The procedure for defining the dynamic portion of this loading for a Design Basis Accident (DBA),

lntermediate Break Accident (IBA) and Small Break Accident (SBA) is presented in this section. The vent system thrust loads due to the quasistatic pres-surization of the containment and the fluid flow in the vent system during the chugging regime are reported separately in Section 4.2.

4.5.3.1 Bases and Assumptions The basis for the chugging lateral load definition is the data obtained from the instrumented downcomers of the Mark I Full Scale Test Facility (FSTF) during testing as described in Reference 4.5.31. The load definition is developed for, and is directly applicable ta, downcomer pairs which are untied, i.e., not connected by lateral bracing. Based on FSTF observations, this load definition is also applicable to braced downcomers.

The FSTF down-coiner lateral loads are defined ..aa.Resultmnt Static Equivalent Loads (RSEL) which, .when .applied statically to the -end .of the -downcoiner, will reproduce at any given time the measured bending response near the downcomer/

vent header junction. The RSEL were obtained from the meisured downcomer bending moments using conversion factors determined from static calibration tests performed on the downcomers, Reference 4.5.31. These RSEL are, there-fore, an actual representation of the structuralhydrodynamic interaction which occurs during chugging.

The maximum design loads for individual plants are obtained by scaling the maximum measured RSEL from the FSIT, The scaling factors are derived on the basis of a comparison of the dynamic characteristics of the downcomers of the individual plants and the FSTF.

4.5.31 Revision 2

NTDO-21888 For fatigue evaluation, the number of load reversals, i.e., PSEL reversals, during the chugging phase were counted and presented in the form of REEL reversal histograms. These histograms are defined on the basis of a statis-tical loading with a 95 percent probability of nonexceedance. Since the chugging loads may be oriented in any lateral direction, these histograms were obtained for each of eight angular sectors around the downcomer end as shown in Figure 4.5.3-2. The FSTF REEL reversal histograms are scaled into a plant unique set of histograms by first scaling the maximum REEL reversal on the same basis as was done for the maximum design loads. Then the nmnber of RSEL reversals are scaled by the ratio of the chugging duration specified for the plants to that of the FSTF. For fatigue evaluation of the downcomer/

vent header junction, the plantunique set of RSEL reversals must be con-verted to a set of stress reversals at the downcomer/vent header junction which result from application of the RSEL reversals.

During the chugging phenomenon, lateral loads (i.e., RSEL) are imposed on the ends of the dowocomers which are random in both magnitude and direction.

Because of the random nature of these chugging forces, there is a small but finite probability that the RSEL on two or more downcomers will align in the same direction to produce a resultant axial loading on the vent system which is larger in magnitude than the maximum RSEL observed to act on a single downcomer.

When two or more downcomers chug synchronously the event has been referred to as a pool chug. The FSTF RSEL were statistically combined to determine the probability of lateral forces on two or more downcomers exceeding specific values during pool chug synchronization.

The procedure for determining chugging lateral loads on downcomers includes the following assumptions:

a. The vent header and downcomers can be considered rigid when compared to the flexible downcomer/vent header junction. Therefore the down comer responds dynamically in any given direction as a single degree of freedom system. This assumption has been verified by analytical studies (Reference 4.5.31).

4.5.32 Rev~!~on 2

NEDO21888

b. The lateral hydraulic load experienced by the FSTF downcomers during chugging is representative of the load that would be experienced by other Mark I downcomers for the same condition (Reference 4.5.31).

c. The FSTF RSEL reversal histograms obtained for the chugging phase of the LOCA are representative of an actual plant behavior over the same duration (Reference 4.5.31). Thus, for the fatigue evaluation of an individual plants downcomer, the FSTF RSEL reversal histogram data must be scaled by the ratio of the chugging duration in the plant to the duration simulated in the FSTF.

d. The loading on any given downcomer as determined from the measured downcomer bending response is predominantly due to a chug occurring on that downcomer. The effect that chugs on neighboring dowocomers have on the measured bending response can be neglected. This assump-tion applies to excitations transferred through the vent system and is based on test observations from the FSTF, Reference 4.5.31. The structuralhydrodynamic phenomenon which takes place during multi vent chugging is however, inherent in the FSTF data and the RSEL, thus determined, include this effect.

e. The chugging loads are approximated as triangular pulse loads for the purpose of dynamic scaling. This is based on observation of the FSTF chugging data (Reference 4.5.31).

f. The load development procedure is applicable for loadings where the measured bending strains on the dowocomer remain within the elastic range. These strains were observed to be well within the elastic range for all of the FSTF tests, Reference 4.5.31.

The procedure for determining the probability of the lateral loads on groups of downcomers exceeding specific values during poo1 chug synchronization requires the following assumptions in addition to those mentioned above.

4.5.33 Revision 2

NEDO21888

g. The maximum chugging load on an individual down-corner can occur in any arbitrary direction with equal probability, as observed from the FSTF test results (Reference 4.5.31).

h. All downcomers experience chugging in phase, i.e., the randomness of chugs in time is conservatively neglected.

i. The magnitude and direction of the chugging load on a downcomer is statistically independent from all other down-comers, as observed from the FSTF test results (Reference 4.5.31).

4.5.3.2 Evaluation Procedure The loads associated with chugging obtained from the FSTF data will be scaled to determine the loads for other Mark I plant down-comers. For each plant, the maximum downoomer design load, histograms of load reversals and the maximum vent system loading produced by synchronous chugging of the down-comers may be determined from the FSTF loads.

The load used for determining the maximum design chugging load on an individual Mark I plant down-corner is based on the peak chugging load observed in the FSTF tests.

The maximum chugging design load P for an individual Mark I plant down-comet max is obtained by scaling the maximum measured FSTF load P The scaling law which incorporates the dynamic characteristics of the FSTF down-corners and the down--

corners of an individual plant is given in Reference 4.5.31 as:

p-::

max l~DLF I

_I where DLF and DLF are the dynamic load ~Tactors- of The untied plantunique and 1

FSTF down-corner/vent-header structures, -respectively. -:

4.5.34 Revision 2

The dynamic load factors depend upon the natural frequencies of the single degreeoffreedom systems representing the downcomer/vent header structures.

For the range of -the Mark I plant dowocomer geometries, the ratio of the dynamic load factors between the FSTF and a plantunique down-coiner is essen-tially independent of :the direction of load application, Reference 4.5.31.

Therefore, in this procedure the dynamic load factors used in the scaling relationship shall be determined on the basis of the down-corner frequencies in the EastWest direction. The natural frequency in the EastWest direction can be determined once the rotational stiffness of the downcomer/vent header junction and the mass moment of inertia of the downcomer in this principal direction have been ascertained. The mass moment of inertia must include the added mass of the water for the submerged portion of the dowocomer.

A maximum load of~3,046 lbs was determined from the FSTF chugging data.

The direction of this load was observed from the FSTF tests to be random in nature. The maximum design load P therefore must be applied to the end of max the plantunique down-corners in a direction such as to maximize the stresses at the down-coiner/vent header junction.

For fatigue evaluation of the down-corners, the required stress reversals at the down-corner/vent header junction can be obtained from the FSTF RSEL reversal histograms. Only RSEL reversals which are greater than a threshold of 5% of the maximum load range were counted for the histograms, since cycles of smaller magnitude do not contribute significantly to fatigue usage. The junction stress reversals are obtained by first scaling the FSTF RSEL reversals into a plant unique set of RSEL reversals. This scaling procedure is a twofold process.

First, the maximum FSTF RSEL reversal is scaled on the same basis as was done for the maximum design loads. Then, the total number of reversals must be scaled by the ratio of the chugging duration specified for the plants to that of the FSTF. Chugging durations for the DBA, IBA, and SEA are specified in Table 4.5.11 of Section 4.5.1, and are applicable to all plants. Since iden-tical durations for the lEA and SBA are specified, only the DBA and lEA need be con--

considered. Based upon actual FSTF chugging durations, for a DBA fati6ue evalua-tion of the down-corners the total number of reversals must be scaled by 0.0586.

For an lEA evaluation, this scaling factor is 1.76.

4.5.35 Revision 2

NEDO21888 The plantunique set of RSEL reversals may now be transformed into a set of stress reversals at the downcomer/vent header junction. One method which can be used to transform the loads to stresses at a particular point A near the downeomer/vent header intersection, Figure 4.5.33, is through a loadstress transformation matrix. This matrix may be obtained from a detailed static finite element analysis of the downcomer/vent header structure. First, a unit load F would be applied at the downcomer end in the North direction as shown.

N Figure 4.5.33 to obtain a = KNFN where o is a representative stress measure to be used in fatigue analysis.

a Similarly, a unit load F would be applied at the downcomer end in the East E

direction to obtain a similar stress measure 0b as KF EE which upon combining stresses yields o=o a b = [KNKE] [~]

For an arbitrary loading P having components PN and P E , the resultant stress is o = KP where the transformation matrix K will, in general, be different for different locations of stress determination.

The total number of stress reversals at a location selected for fatigue evalu-ation is obtained by summing the stress reversals produced at that location by the chugging RSEL reversals in each sector.

4.5.36 Revision 2

NEDO-21888 For the case of pool chug synchronization, the probability of exceeding a given force magnitude at least once during multidowncomer chugging is deter-mined from probability of exceedance curves derived from the FSTF data. The load per FSTF downcomer can be obtained from the curves for various numbers of downcomers once an acceptable probability level of exceedance has been es-tablished. A probability of exceedence of l0~ per LOCA is conservative for this application. The resultant load in any direction due to pool chug syn-chronization may now be determined by multiplying the number of downcomers being considered by the load per downoomer. It is necessary to scale this resultant FSTF load by the same scale factor used previously for scaling the maximum FSTF chugging load to determine the resultant lateral load on a plant-unique Mark I vent system.

For downcomer pairs connected by lateral bracing, the stress in the bracing is evaluated by assuming one of the two tied downoomers is subjected to a dynamic load of triangular shape, with an amplitude of:

p F 1 max vrf td where P is the maximum measured RSEL of 3,046 lbs for an untied downcomer 1

during chugging, f is the lowest natural frequency of vibration of an untied plantunique downcomer, and td is the duration of the chugging load, assumed to be .003 seconds. The load direction shall be taken as that (in the hori-zontal plane) which results in the worst loading condition for the tie bar and its attachments to the downcomers.

4.5.37/B Revision 2

NEDO21888 Figure 4.5.31. Deleted 4.5.39 Revision 2

700 1248 600 2100 540 500 1800 Co

-J 400 U,

a:

LU

-I: 2:

IL C-i U.

0 C a: 300 N)

H LU 0, H 0 C 247 C z C 200 108 100-49 7m~L~i28 6 3 8 3 4 2 1 3 (1 ~ ~

0 10 20 30 40 50 60 70 80 90 100 4

~3.

0, RSEL (PERCENT) a El ha Figure 4.5.32. Distribution of Chugging RSEL Reversals for a Typical Sector

NEDO-21888 N

Figure 4.5.33. Notation Used for Transforming RSEL Reversals into Stress Reversals at a Fatigue Evaluation Location A 4.5.311/4.5.312 Revision 2

NEDO21888 REFERENCES FOR SECTION 4.5.3 4.5.31 Mark I Containment Program, Development of Downcomer Lateral Loads from Full Scale Test Facility Data Task Number 7.3.2, General Electric Company, Report NO. NEDO24537, August 1979.

4.5.313/4.5.314 Revision 2

NEDO21888 4.5.4 Vent System Loads Pressure loadings are experienced by the vent system as a result of chugging.

These vent system loads can be separated into the following three components:

a. A gross vent system pressure oscillation consisting of pressuri-zation during the prechug portion and depressurization during the postchug portion of each chug cycle.

b. An acoustic vent system pressure oscillation which is excited as a result of the pressurization and depressurization of the vent system.

c. An acoustic downcomer pressure oscillation which is excited as a result of the rapid depressurization at the downcomer exits.

The first component *of pressure loading is applied over a relatively long loading cycle (1 to 2 seconds) which corresponds to the time between chug cycles. The second and third pressure load components are related to the acoustic response frequencies in the vent system and downcomer, and are defined as a periodic load with components at the acoustic frequencies of the vent system .(including -the downcomers) and of the downcomers themselves.

4.5.4.1 Bases and Assumptions The Mark I vent system consists of the main vents, the vent header and the downcomers. The vent system load definition is based on data obtained from the Mark I FSTF. The test facility, test matrix and test results are described in Reference 4.5.41. Two tests were selected for the bases of the vent system chugging loads. Test Ml with the small steam break is the basis for both the 0.7 Hz gross vent system pressure oscillation and the 40 to 50 Hz acoustic downcomer pressure oscillation load specifications.

Test M4 with the small steam break and overpressure is the basis for the 4.5.41 Revision 2

NEDC-21888 6.9 to 9.5 Hz acoustic vent system pressure oscillation load specification.

These tests produced the largest pressure oscillation amplitudes for these components.

The frequency range basis for the acoustic vent system and downcomer pressure oscillations are as defined in subsection 4.5.1.1. The frequency specified for the gross vent system pressure oscillation is based on the mean of the FSTF chugging data.

The characteristics of the total vent system pressure load and the role of each of the pressure load components exhibit some change over the period of the blowdown during which chugging occurs. Early in the chugging period, both the acoustic downcomer pressure oscillation and the gross vent system pressure oscillation contributions are relatively small compared to the contribution of the acoustic vent system pressure oscillations which are dominant. Later in the chug cycle, a higher frequency postchug acoustic downcomer pressure oscillation begins to dominate the downcomer pressure loads. The contribution of the acoustic vent system pressure oscillation lessens during this portion of the blowdown. The relative contribution of the gross vent system pressure oscillation generally becomes more apparent in the vent header and main vents.

The vent system data simultaneously contained contributions of some or all of the individual load components. In order to simplify the specification, the chugging data have been analyzed to extract the maximum of each load-ing component so each vent system pressure load component can be speci-fied as if it were acting alone. This was done by selecting from the chugging data those time segments during the blowdown where each loading component was clearly dominant. The peak load is defined for each as if it alone were contributing to the observed pressure loads. The frequency range specified for these loads is based on a linear acoustic analysis of the spectrum of Mark I vent system and downcomer geometries which included all of the varia-tions in the Mark I Containment vent system designs. The result is a conserva-tive definition of the vent pressure loads.

4.5.42 Revision 2

NEDO21888 The major assumptions made in this definition of the vent system pressure loads are:

a. The vent system pressure loads due to chugging observed in the prototypical FSTF tests represent the pressure loads that would be expected during chugging in a Mark I Containment. This is justi-fiable because the source strength for chugging is mainly controlled by thermodynamic and geometric conditions at the downcomer exits.

These conditions are very similar for all Mark I plants and were simulated in the FSTF tests.

b. Vent system acoustic response frequencies for the various Mark I vent system and downcomer geometries can be defined by a linear acoustic analysis. This is justified by the good agreement between FSTF analysis and data.

c. The response of the vent system to the load definitions given here will remain linear to permit the definition of bounding single frequency components of the vent system pressure loads. This is justified by the relatively low magnitude of the loading components.

4.5.4.2 Load Definition The vent system chugging load definition is summarized in Table 4.5.41. For the duration of application refer to Table 4.5.11. Because of the manner in which they were developed, these loads are to be applied individually, not combined with each other. They are to be applied about the local pressures at the appropriate times in the blowdown depending on the size of the break (Table 4.51).

The chugging load specified for the downcomers should be used to calculate only the circumferential structural response (e.g., hoop stress) of the downcomer, not system responses to lateral, thrust, or other loads which are 4.5.43 Revision 2

NEDO-21888 transmitted through the downcomers to other components. The lateral loads that result from the unbalanced pressure between the vent header and down comer exits are included in the chugging lateral load definition (see Section 4.5.3), and vent system thrust loads are included in Section 4.2.

For other loads which may occur in combination with the chugging loads see the bar charts in Section 3.0.

4.5.44 Revision 2

NEDO21888 Table 4.5. 41 VENT SYSTEM LOAD AMPLITUDES AND FREQUENCIES FOR CHUGGING Amplitude (psi Frequency Main Vent Load Type (Hz) Vents Header Downcomers Gross Vent System Use wave form in +/-2.5 +/-2.5 +/-5.0 Pressure Oscillation Figure 4.5.41 (0.7 Hz)

Acoustic Vent System Sinusoidal with +/-2.5 +/-3.0 +/-3.5 Pressure Oscillation frequency varying between 6.9 to 9.5 Hz Acoustic Downcomer Sinusoidal with N/A N/A +/-13.0 Pressure Oscillation frequency varying between 40 to 50 Hz 4.5.45 Revision 2

LU IL 2:

LU 1:21 4=- IL a C C-fl N) 4= H C

0 C C

TIME (U

4 H

0, 0

El ha Figure 4.5.41. Chugging Wave Form for Gross Vent System Pressure Oscillation Load

REFERENCES FOR SECTION 4.5.4 4.5.41 J. E. Torbeck, et al., Mark I Containment Program, Full Scale Test Program Final Report, General Electric Company, Report No. NEDO24539, August 1979.

4.5.47/4.5.48 Revision 2

SECTION 5 SAFETY/RELIEF VALVE DISCHARGE LOADS Revision 2

NED 021888 5.0 SAFETY/RELIEF VALVE DISCHARGE LOADS This section specifies the methodology to be used to define loads due to S/RV actuations. Section 5.1 gives an introduction to the S/RV phenomena and loadings. Section 5.2 describes the loads due to S/RV actuations through TQuencher discharge devices. Section 5.3 describes the loads due to S/RV actuations through Ramshead discharge devices.

5.01/5.02 Revision 2

NEDO21888

5.1 INTRODUCTION

When a S/RV actuates, pressure and thrust loads are exerted on the S/RVDL piping and discharge device (Ramshead or TQuencher). In addition, the expul-sion of water and then air into the suppression poo1 through the discharge device results in pressure loads on the submerged portion of the torus shell and drag loads on submerged structures. This section describes procedures for determination of loads* resulting from water/air discharge through both TQuencher and Ramshead devices following S/RV actuation. Procedures are presented for the definition of the following S/RV discharge loads:

• S/RVDL pressure and temperature

• Thrust loads on S/RVDL piping

• Thrust loads on TQuencher arms

• TQuencher internal pressure

• Water jet loads on submerged structures

• Torus shell pressure distribution

• Air bubble induced drag loads on submerged structures In addition, methodology for determining the S/RVDL water reflood height and timing for input to subsequent S/RV actuation analyses is provided.

A series of analytical models and hand calculational procedures will be used to develop S/RV line unique loads. Flowcharts depicting the loads/parameters evaluated by each analytical model or procedure and how each methodology inter-acts with the others are presented in Figures 5.11 and 5.12.

• All loads are applicable to S/RVDLs ending in either Ramshead or TQuencher discharge devices, unless specifically indicated otherwise.

5.11 Revision 2

NEDO21888 These loads/parameters are to be evaluated under each of the following initial conditions:

A. First actuation B. First actuation leaking S/RV (Ramshead only)

C. Subsequent actuation Section 5.2 presents phenomenological descriptions, load definition proce dures, assumptions and bases, for each load resulting from a S/RV discharge through a TQuencher device. Section 5.3 presents similar information for loads resulting from S/RV discharges through Ramshead discharge devices.

5.12 Revision 2

I OBTAIN S/RVOL AND TORUS

_____ SIRVDL CLEARING MODEL CALCULATES:

SIRVOL PRESSURE MARK I LOAD DEFINITION REPORT PROVIDES CALCULATIONAL PROCEDURES FOR:

GEOMETRY AND OPERATING SIR VOL THRUST LOADS CONDITIONS FOR THE CASE S/RVDL TEMPERATURE TO BE EVALUATED T-OUENCHER PRESSURE WATER JET DRAG LOADS ON SUBMERGED STRUCTURES WATER CLEARING VELOCITY WATER CLEARING THRUST WATER CLEARING ACCELERATION LOADS ON T-QUENCHER ARMS DISCHARGE PRESSURE 2:

RI a

U N H

03 1.) 03 03 SIRVOL REFLOOD MODEL CALCULATES:

S/RVOL REF LOOD HEIGHT AND TIMING p I TORUS SHELL LOAD MODEL (FOR SUBSEQUENT S/RV CALCULATES:

ACTUATION ANALYSES)

AIR BUBBLE PRESSURE AIR BUBBLE DRAG MODEL AIR BUBBLE RADIUS S CALCULATES:

AIR BUBBLE VERTICAL POSITION V~

AIR BUBBLE INDUCED DRAG TORUS SHELL PRESSURE LOADS ON SUBMERGED DISTR IBUTION STRUCTURES (U

4 H

0, H

0 El ha Figure 5.11. TQuencher Load Definition Scheme

MARK I LOAD SIRVDL CLEARING MODEL DEFINITION REPORT CALCULATES:

SIRVDL PRESSURE OBTAIN SIRVDL AND TORUS SIRVOL THRUST LOADS PROVIDES CALCULATIONAL GEOMETRY AND OPERATING PROCEDURE FOR:

CONOIT IONS FOR THE CASE WATER CLEARING ACCELERATION TO BE EVALUATED AND VELOCITY S WATER JET LOADS ON SUBMERGED STRUCTURES DISCHARGE PRESSURE 0 HORIZONTAL POSITION OF BUBBLES

_________________ T 2:

RI S/RV AIR BUBBLE PRESSURE MODEL CALCULATES:

S/RVDL REFLOOD MODEL 4=- 03 AIR BUBBLE PRESSURE CALCULATES:

5/AVOL REFLOOD HEIGHT ~ AIR BUBBLE RADIUS AND TIMING (FOR SUBSE-QUENT ACTUATIONS) AIR BUBBLE VERTICAL HEIGHT AND TIMING POSITION I

S/RV AIR BUBBLE PRESSURE AIR BUBBLE DRAG MODEL ATTENUATION MODEL CALCULATES:

CALCIJt.ATES: AIR BUBBLE INDUCED DRAG LOADS ON SUBMERGED TORUS SHELl. PRESSURE STRUCTURES (U

4 DISTRIBUTI.iN H

U, H

0 El ha Figure 5.12. Ramshead Lpad Definition Scheme

NEDO21888 5.2 TQUENCHER LOADS 5.2.1 S/RV Discharge Line Clearing Transient Loads 5.2.1.1 Load/Parameter Descriptions

a. S/RVDL and TQuencher Pressure When a S/RV opens, the pressure within the S/RVDL undergoes a transient prior to reaching a steady state value. A transient pressure wave travels back and forth in the line as the pressure continues to increase until the inertia of the water slug in the submerged portion of piping is overcome. During the water clearing transient, the pressures within the discharge pipe and the TQuencher device reach their maximum values. Following expulsion of the water slug, the peak pressure in the discharge pipe decreases to a quasisteady state value which is a function of the S/RV steam flow rate and friction along the line upstream of the entance to the TQuencher device. Similarly, the TQuencher internal pressure increases and then decreases to a quasisteady state value which is a function of the steam flow rate and pressure losses resulting from flow through the holes in the TQuencher device. Figures 5.2.11 and 5.2.12 present sample analytical model predictions of S/RVDL and TQuencher internal pressure.

b. Thrust Loads on the S/RVDL The high flow of steam into the discharge line when an SIR? opens results in the development of a pressure wave at the entrance to the line.

This pressure wave travels to the air/water interface where it is reflected and returns to the entrance of the line. The wave con-tinues to travel back and forth in the discharge line until a quasi steady state condition is reached as the pressure differential across the wave approaches zero. The pressurization of the S/RV line results in an acceleration and expulsion of the water in the submerged portion of piping.

5.2.11 Revision 2

NEDO21888 During the early portion of this transient, a substantial pressure differential exists across the pressure wave. Therefore, when the wave is within a S/RV pipe segment between a pair of elbows there exists a substantial difference in the pressure applied to the inte-rior surface of the elbows on each end of the segment. This pressure differential, plus momentum effects from steam (or water in initially submerged pipe runs) flowing around elbows in the line, results in transient thrust loads on the SIR? discharge pipe segments. These loads should be considered in the design of S/RV pipe restraints, the connection of the S/RV to the main steam line, and the TQuencher support system. Figures 5.2.13 and 5.2.14 presents sample analytical predictions of the transient thrust loading on S/RV dis-charge pipe segments initially filled with gas and water, respectively.

Figure 5.2.15 presents the positive sense of the thrust loading on a pipe segment.

c. S/R? Air Bubble Charging Pressure Following an S/R? actuation, the airsteam mixture initially in the S/RVDL is compressed prior to discharge into the pool through the TQuencher device. The pressure at which the air is discharged to the pool is proportional to the maximum pressure which occurs at the air/water interface. Therefore, the maximum predicted pressure (predicted by the S/RVDL clearing model) at the air/water interface is used as input to the evaluation of S/R? air bubble and shell pressures.

d. Mass Flow Rate and Acceleration of Water During S/RV Discharge As the pressure increases in the S/RVDL following SIR? actuation, the water slug in the submerged portion of piping is expelled into the suppression pool. The mass flow rate and acceleration of the water slug during this transient are evaluated using the S/R? discharge line clearing model. One or both of these transient parameters are used as input for evaluation of water jet drag loads and thrust loading 5.2.12 Revision 2

NEDO21888 5.2.1.3 Load Definition Procedure A summary of key parameters which influence S/R? discharge line clearing loads is shown below.

a. S/RVDL initial water leg

b. S/RVDL air volume

c. S/RVDL diameter

d. SIR? steam flow rate

e. S/RVIJL initial temperature

f. S/RVDL configuration and hydrodynamic losses

g. SIR? main disk stroke time.

Using the assumptions presented in Section 5.2.1.2, unique values for these parameters are established for each S/RVDL in the plant under initial conditions A (First Actuation) and C (Subsequent Actuation).

These values are then used as input to the S/RVDL line clearing analytical model (Reference 5.2.12) to obtain each transient S/RVDL clearing load and parameter.

5.2.15/5.2.16 Revision 2

I LU IL nU, U, 2:

LU RI IL U 2 ha N) 2C H I- IL C I w C

.4 I C 2

-J 0

IL U,

0 0.1 01 0.3 0.4 (U TIME FROM VALVE ACTUATION heel 4

H 0,

H 0

El N) Figure 5.2,11. Sample Prediction of S/RVDL Internal Pressure Transient

U LU U,

U)

LU 0~ z 131

-~ e 4

l~3 Uj H I-. Iz oz~

LU C.,

z LU 9-0 TIME FROM VALVE ACTUATION (secl I-..

0 Figure 5.2.12. Sample Prediction of TQuencher Internal Pressure Transient

if 0

3C 2:

U a-. RI to 0 ha D 0 a:

I N)

I I- H

.0 03 03 03 0.4 (U

4 TIME FROM VALVE ACTUATION (see)

H 0,

H 0

El Figure 5.2.13. Sample Prediction of Thrust Loading on an S/RV Pipe Segment Initially Filled with Gas ha

NEDO-21888 0

4:

0 4

U, I

I-TIME FROM VALVE ACTUATION (sod Figure 5.2.14. Sample Prediction of Thrust on S/RV Pipe Run Between the Discharge Device and the First Upstream Elbow (Pipe Run Initially Filled with Water) 5.2. 110 Revision 2

TO SIRV THE TRANSIENT THRUST LOAD IS THE NET LOAD ON THE PIPE SEGMENT TO AND IS POSITIVE IN THE DIRECTION SHOWN DISCHARGE DEVICE Figure 5.2.15. Sense of Thrust Loading 5.2.111 Revision 2

E

.0 LU I

4 z131

.3 0 4 0 U.

LU H 9-0.3 TIME FROM VALVE ACTUATION Ca I-.

0 13 Figure 5.2.16. Sample Prediction of Mass Flow Rate of Water Exiting TQuencher

.3

N hi -~

F N) -o Z RI I 2 C I-)

IL H C-fl LU C

-J C ha LU C U

U I C IL 4= LU I-C TIME FROM VALVE ACTUATION (U

4 H

0, H

0 El Figure 5.2.17. Sample Prediction of Water MaBs Acceleration I-,

NEDO21888 REFERENCES FOR SECTION 5.2.1 5.2.1-1 R. A. Asai, et al., Mark I Containment Program Final Report Monticello TQuencher Test, General Electric Company, Report No. NEDO21864, June 1979.

5.2.1-2 A. J. Wheeler, Comparison of Analytical Model for Computing Safety/Relief Valve Discharge Line Transient Pressures and Forces to Monticello TQuencher Test Data, General Electric Company, Report No. NEDO23749l, September 1978.

5.2.13 ASME Boiler and Pressure Vessel Code,Section III, Subsection NB7000. Available from the American Society of Mechanical Engineer, New York, New York.

5.2.115/5.2.116 Revision 2

NEDO21888 5.2.2 Torus Shell Pressure 5.2.2.1 Load Description Prior to the initial actuation of an SIR? caused by a normal operational transient, the S/RV discharge lines contain air at atmospheric pressure and suppression pool water in the submerged portion of piping. Following SIR? actuation, steam enters the S/RVDL compressing the air within the line expelling the water slug and discharging the air into the pool. The compressed air charges bubbles which expand resulting in an outward motion of the surrounding pool water. The outward momentum of the pool water causes the pressure within the bubbles to drop below the ambient pool pressure. The negative bubble pressure slows and reverses the bubble expansion and the pool water begins to move inward. The inward momentum of the water results in a compression of the bubbles to a pressure above ambient.

The expansion and compression of the air bubbles continues until they rise and break through the pool surface. The positive and negative dynamic pressures developed within these bubbles result in an oscillatory, attenuated pressure loading on the torus shell.

Figure 5.2.21 is an example of the analytically predicted torus shell pressure transient resulting from a S/R? discharge through a TQuencher device. Fig-ures 5.2.22 and 5.2.23 are examples of analytically predicted torus shell longitudinal and radial pressure distributions, respectively.

5.2.2.2 Bases and Assumptions The _torus shell~ pressure, loading is evaluated jasi~g..afirst principles analytical model (Torus Shell Load Model).. The model incorporates information obtained from extensive full scale inplant TQuencher testing (Reference 5.2.21) as well as parametric testing of a 1/4 scale TQuenche? discharge system (Ref erence 5.2.22). The model has been shown to conservatively predict avail-able full scale shell pressure data and to correctly predict the parametric trends measured at 1/4 scale. The analytical and/or empirical bases for the Torus Shell Load Model are documented in Reference 5.2.23.

5.2.21 Revision 2

NEDO2 1888 To ensure that a conservative load definition is obtained, the same assumptions are employed when applying the Torus Shell Load Model for design as were used with the S/RVDL Clearing Model (See Section 5.2.1.2) with the following additions:

a. For multiple S/R? actuation, it is assumed that the pressure loading at a given location on the torus shell is equal to the absolute sum of the peak pressure loads predicted to occur at that location for each individual SIR? actuation. In the event that the -combined peak tOru8 shell pressure exceeds 1.65 times the local predicted peak bubble pressure due~zo a~single valve actuation, the resultant torus shell peakpres.urs will be taken .at~ the lower valveA All bubbles~

are assumed to oscillate inphase.

In addition, the following conservative assumptions/methids are applied when evaluating the torus shell loading for each single valve actuation prior to combining the results to obtain multiple valve loads.

(1) Under test conditions, the analytical models conservatively pre-dict available measured data (single valve actuations).

(2) The design S/RV flow rate is 1.225 times the ASME rated flow.

(3) The pool water level is assumed to be at its maximum value allowed by technical specifications.

(4) The pool temperature is assumed to be at its maximum value for the condition under evaluation.

b. The predicted frequency of the torus shell pressure waveform is accurate to +/-25percent for first actuations and +/-40percent for leaking valve and subsequent actuations. Therefore, the Frequency range for first actuation shall be D.75 times tbe -minimum predicted frequency to L23 .11t+/-mesThe max1iiuu~ predicted frequency, and fot~

leaking valve and~ subsequent -actuation the frequency range shall be 0.6 tI.mes the minimi~ predicted frequency to 1.4 times the maximum predicted frequency,,.

5.2.22 Revision 2

NEDO21888 These uncertainties are based on consideration of scatter observed in full scale testing (Reference 5.2.21) and accuracy with which the Torus Shell Load Model has been shown to predict shell loading frequencies over the range of S/RVDL geometries and operating conditions (Reference 5.2.23).

c. For subsequent actuations the pressure amplitude predicted for first actuation is used in conjunction with the bubble frequency and frequency range for subsequent actuations.

d. The SIR? discharge line water leg length s~alLbe..less..-than or equal -~

to 13.5 feet. .1n the-event thewater leg length for a particular~

discharge line exceeds 13.5 feet, the load prediction for a 13.5 feet water leg length shall be used.

5.2.2.3 Load Definition Procedure The following are key parameters which influence the torus shell pressure loading.

a. S/R? air bubble charging pressure (from S/R? discharge line clearing model)

b. Suppression poo1 temperature

c. Suppression pool geometry

d. TQuencher location in the suppression pool.

The assumptions in Section 5.2.2.2 are used to establish plant, SIR? line, and initial condition unique values for these parameters. These values are then input to the Torus Shell Load Model (Reference 5.2.23) to evaluate the torus shell pressure distribution and transient pressure waveform for a single valve actuation for initial conditions A and C. The peak pressure values pre dicted at each shell location (due to single valve actuations) are added by the SRSS method to ~obtain theit~grated~pruB shell J.oadi.n8~j,or eacb of the load case/initial condition combinations shown in Table 3.01.

The structural evaluation is performed using the predicted pressure waveform within the appropriate uncertainty range (stretched or compressed time scale) which is expected to result in the maximum structural response.

5.2.23 Revision 2

NEDO2 1888 The SRV load assessment procedure for plants equipped with quencher devices may be based on a series of singlevalve, cold discharge inplant tests, performed on the SR? discharge line expected to produce the highest loads.

The pressure waveform and structural response measurements of the inplant tests shall be used to calibrate a coupled loadstructure analytical model.

The load prediction portion of the model previously described shall be calibrated at test conditions and then applied at design conditions.

5.2.24 Revision 2

ha I

LU a:

a. H 4

C-P LU C U,

(U 4

H TIME AFTER VALVE ACTUATION (mteel I

ha Figure 5.2.21. Sample Prediction of Torus Shell Pressure Loading Transient

2:

U U

0 ha ha H

ia 03 03 0 C A

~:1 (U

4 H

0, H

0 El ha Figure 5.2.22. Sample Prediction of Torus Shell Longitudinal Pressure Distribution

NEIDO21888 Figure 5.2.23. Sample Prediction of Torus Shell Radial Pressure Distrib.ution at Section AA in Figure 5.2.22 5.2.27/5.22..8 Revision 2

NE DO 21888 REFERENCES FOR SECTION 5.2.2 5.2.21 R. A. Asai, et al., Mark I Containment Program Final Report Monticello TQuencher Test, General Electric Company, Report No. NEDO21864, June 1979.

5.2.22 C. T. Sawyer, Mark I Containment Program Final Report 1/4 Scale TQuencher Test, General Electric Company, Report No. NEDO24549, June 1979.

5.2.23 P. Valandani, Mark I Containment Program Analytical Model for Computing Air Bubble and Boundary Pressures Resulting from a SIR?

Discharge Through a TQuencher Device, General Electric Company, Report No. NEDO21878, September 1979.

5.2.24 E. A. Buzek, et al., Final Report InPlant Safety/Relief Valve Discharge Load Test Monticello Plant, General Electric Company, Report No. NEDC21581, June 1977.

5. 2. 29/5. 2. 210 Revision 2

NEDO21888 5.2.3 S/RVDL Reflood Transient 5.2.3.1 Description of Phenomena Following closure of a S/RV the steam pressure in the S/RVDL decreases rapidly as the steam flows into the suppression pool. At a sufficiently low steam pressure, pool water will reenter the S/RVDL. A rapid depressurization of the line then occurs as steam in the line is condensed by the inf lowing water.

This depressurization causes the water to reflood into the S/RVDL and the vacuum breaker valve on the S/RVDL to open, allowing drywell atmosphere to enter the line. The reflooding water may rise in the line to a level somewhat above its initial preS/RV actuation level before equilibrium is reestablished.

The actual reflood water level depends primarily on the size of the S/RVDL vacuum breaker valve to allow a rapid repressurization of the line.

A subsequent actuation may occur following closure of the SIR?. The loads developed during the subsequent actuation will depend on the conditions in the S/RVDL at the time of the actuation. Subsequent actuation loads on the S/RVDL and discharge device are proportional to the initial water level in the S/RVDL.

Therefore, the peak reflood water level predicted to occur at a point in time after the minimum predicted time* between SIR? actuations is used as the initial water leg in the S/RVDL for subsequent actuation load calculations.

An example of analytically predicted S/RVDL reflood behavior is illustrated in Figure 5.2.31, which presents water column length in the S/RVDL as a function of time. It is seen that the water column length increases continuously from the time the reflood transient begins until the first peak is reached. The delay between valve closure and the initiation of reflood is due to pressure decay in the S/RVDL following S/R? closure.

• Prediction made by plant unique analysis. If an analysis is not utilized to determine timing between actuations, the subsequent actuation is assumed to occur at the time corresponding to the maximum predicted reflood.

5.2.31 Revision 2

NEDO-218 88 5.2.3.2 Bases/Assumptions The S/RV discharge line reflood transient is evaluated using a first principles analytical model (Ref lood Model). The model has been shown to conservatively predict the maximum S/RVDL reflood measured during full scale inplant testing.

The analytical bases and appropriate model data comparisons for the reflood model are documented in Reference 5.2.31.

The following assumptions will ensure that a conservative reflood prediction is obtained:

a. The S/RVDL vacuum breaker is located in the drywell at a point above the maximum S/RVDL reflood level.

b. The suppression pool water level is at the maximum value allowed by technical specifications. This assumption maximizes the pool pressure (elevation head) at the discharge device elevation resulting in the maximum net force driving the water slug up the S/RVDL.

c. The delay time between complete S/R? closure and the beginning of the reflood transient is a linear function of the entire SIR? discharge line volume. The delay time used for the evaluation of reflood timing is based on measured inplant S/RVDL depressurization transients following S/R? closure (Reference 5.2.32). This delay time is mul-tiplied by the ratio of the S/RVDL volume to the volume of the line tested to provide line unique delay times.

5.2.3.3 Procedure for Evaluating the S/RVDL Reflood Transient A summary of key parameters which influence the S/RVDL reflood transient is shown below:

a. Drywell and wetwell pressure

b. S/RVDL vacuum breaker effective flow area

d. Vacuum breaker opening time 5.2.32 Revision 2

NEDO21888

d. Vacuum breaker set point

e. S/R?DL geometry and hydrodynamic losses

f. Vacuum breaker location along the S/RVDL

g. Whether air or steam enters the vacuum breaker

h. Condensation heat transfer coefficient on the interface of the reflooding water slug

i. Pool water temperature

j. Submergence depth of the discharge device.

The assumptions presented in Section 5.2.3.2 are used to establish SIR? dis-charge line unique values for these parameters. These values are then used as input to the S/R?DL reflood analytical model (Reference 5.2.31) to evaluate the discharge line reflood transient.

Once the reflood transient predictions have been obtained, a delay time is added at the beginning of each transient to account for time required for the S/RVDL to depressurize following SIR? closure and therefore allow the reflood transient to begin. This delay time is determined using the following relation:

Delay Time = C V sec where Volume of S/RVDL under consideration, ft3 C = 0.008, -~-~ determined from Monticello full scale TQuencher ft3 tests (Reference 5.2.31)

The maximum S/R?DL reflood, which is predicted to occur at a time in the transient after the minimum predicted time between SIR? actuations (based on plant unique analyses), is used as the initial water leg for subsequent S/R?

actuation evaluation. Optionally, if plant unique analyses are not performed, the maximum reflood occurring at any time during the ref lood transient is to be used as the initial water leg for such evaluations.

5. 2. 33/5. 2. 34 Revision 2

4 0

Ln ni

.3 0 z

I-., r~J H

LU U 4 U

LU

.3 I.-

4 U

CD TIME AFTER SIRV CLOSURE hoc) 4 I-.

a, 0

Figure 5.2.31. Sample Prediction of S/RVDL Ref lood Transient I...,

NEDO21888 REFERENCES FOR SECTION 5.2.3 5.2.31 A. J. Wheeler, Mark I Containment Program, Analytical Model for Computing Transient SIR? Discharge Line Ref lood, General Electric Company, Report No. NEDO23898, August 1979.

5.2.32 R. A. Asai, et al., Mark I Containment Program Final Report Monticello TQuencher Test, General Electric Company, Report No. NEDO21864, June 1979.

5. 2. 37/5. 2. 38 Revision 2

NEDO21888 5.2.4 TQuencher Water Jet Loads on Submerged Structures 5.2.4.1 Load Description When an S/R? is actuated, water initially contained in the submerged portion of the S/R? discharge line (S/RVDL) is forced out of the TQuencher arm through the arm holes forming orifice jets. Some distance downstream the orifice jets will merge to form coli.min jets. Further downstream, the column jets will merge to form the quencher arm jets. As soon as the water flow through the arm holes ceases, the quencher arm jet velocity will decrease rapidly and the jet will penetrate a limited distance into the pool. Figure 5.2.41 shows the phases of quencher jet formation and decay. These TQuencher water jets will create drag loads on nearby submerged structures which are within the jet path.

5.2.4.2 Bases/Assumptions The bases for establishing the water jet dynamics are empirical mathematical.

models (Reference .5.-2.42) obtained from steadystate submerged jet theory at various Jetzones.-(orific~e jet, column jet, _quencher_arm jet and jet penetration). Reference 5.2.41 provides the first principles methodology from which the jet properties for water jets from the quencher arms and endcap holes can be conservatively determined.

To ensure that a conservative TQuencher water jet load definition is obtained, the following assumptions are employed when applying the TQuencher water jet model.

a. The quencher arm is divided into six sections along the axis of the arm. Each section assumes a uniform hole pattern. For arms with endcap holes, there is an additional endcap section (see Figure 5.2.42).

5.2.41 Revision 2

NEDO21888

b. As the air/water interface reaches the first column of holes on the arm, the water acceleration through all the holes is assumed constant and equal to the value when the interface is at the first column of holes.

c. The maximum value of the water velocity through the quencher arm holes found in a section of the quencher arm is conserva-tively used to calculate velocity characteristics for the water jet from that section.

d. The above water velocity is further conservatively assumed constant with time and applied in the calculation throughout the clearing period of that section.

Higher water velocities through the arm holes result in higher momentum in the water jet and larger drag loads on the structures.

5.2.4.3 Load Definition Procedure A summary of the key parameters which influence the TQuencher water jet loads is shown below:

a. Maximum water velocity through the quencher arm holes.

This quantity depends on the S/R?DL clearing transient at the initial conditions to be analyzed. The S/RVDL water clearing model is used to determine the water jet velocity and acceleration up to the time that the first column of holes is uncovered. The water jet momentum depends on this velocity.

b. Distance from the structure to the quencher arm.

The velocity~t~t1i~structure location depends on this distance because the water jet. velocity decreases rapidly with the distance from the arm. The drag load on the structure is proportional to the square of-~the.io~a1 welocity at the structure.

5.2.42 Revision 2

NEDO21888

c. Structure geometry.

The drag load is proportional to the structures projected area and the drag coefficient which are functions of the structure geometry.

5. 2. 43/5. 2. 44 Revision 2

NEDO2 1888 QUENCHER ARM a ORIFICE JETS b COLUMN JETS c. QUENCHER ARM JETS 0

d. JET PENETRATION Figure 5.2.41. Phases of Quencher Jet Formation and Decay 5.2.45 Revision 2

U 4=

I ha 0 H C

03 C

SECTION SECTION SECTION SECTION SECTION SECTION NO.1 NO.2 NO.3 NO.4 NO.5 NO.6 NOTE: FOR T-OUENCHER ARM WITH NO ENDCAP HOLES THERE IS NO SECTION NO. 7 (U

4 H

0, H

0 El ha Figure 5.2.42. Jet Sections Along the Quencher Arm on Torus Plan View

NEDO2 1888 REFERENCES FOR SECTION 5.2.4 5.2.4.1 Analytical Model for TQuencher Water Jet Loads on Submerged

-1 Structures, Prepared for General Electric Company by Nuclear Services Corporation and Southwest Research Institute, Report No. NEDO25090, May 1979.

5.2.42 F. J. Moody, Analytical Model for Liquid Jet Properties for Predicting Forces on Rigid Submerged Structures, General Electric Company, Report No. NEDO21472, April 1978.

5.2.47/5.2.48 Revision 2

NEDO-21888 5.2.5 TQuencher Bubble- Induced Drag Loads on Submerged Structures Oscillating bubbles resulting from an S/RV actuation create an unsteady three dimensional flow field and therefore induce acceleration and standard draq forces on the sukinerged structures in the suppression pool.

This section addresses the load definition procedures for loads on sub-merged structures due to TQuencher S/RV bubbles. The following subsections discuss the methodology to define the loads; the bases and assumptions on which the analytical models are developed and verified; and the selection of key parameters for which the design load on submerged structures can be obtained.

5.2.5.1 Bases and Assumptions The bases of the analytical model are presented in Reference 5.2.51. The major assumptions are summarized as follows:

a. The total drag onsuhnerged structures is the sum of the acceleration and standard drags (See item e. for exceptions).

b. The flow fields caused by T-Quencher S/RV bubbles are described by fluid sources. The presence of boundaries is accounted for by using the method of images. The circular crosssection of the torus shell is modeled by a rectangular crosssection (Model E specified in Reference 4.3.82).

c. Source strengths are determined from the TQuencher bubble dynamics model (Reference 5.2.5-2). The determination of the charging, formation, and rise of the oscillatory bubbles is subject to the same conservative factors that are used for the quencher torus shell pressure loads, as described in Reference 5.2.52.

d. The drag loads on the quencher arms and the S/RV discharge line are computed on the basis of asymmetric bubble dynamics, i.e., two full strength bubbles acting on one side of the quencher arm with ambient pressure acting on the other side.

5.2.51 Revision 2

NEDO21888

e. The standard drag forces on circular cylinders are computed by using a conservative drag coefficient of Cd = 3.6 which bounds the lift force drag coefficient as experienced in test data, for Li T/D> 2.74, where U is the maximum velocity, T is the period m m of bubble oscillation and D is the cylinder diameter. For U m T/D~ 2.74, the drag forces shall be computed on the basis of acceleration drag alone.

f. Drag forces on structures with sharp corners (e.g., rectangular and I beams) are computed by considering forces on an equivalent cylinder of diameter D = ~7L where L is the maximum eq max max transverse dimension and is defined as the diameter of a circum-scribed cylinder about the crosssection of the structure.

g. Long slender structures are modeled in segments of length CL),

which do not exceed its diameter CD or D ). The total force eq on the structure is obtained by integrating the individual segment force over the entire length of the long structure.

h. Interference effects due to the proximity of solid boundaries are considered for each segment that has its center less than 1.5 diameters from the boundary. Multipliers are needed to increase both the acceleration and the standard drag. Similarly, multi-pliers are needed to include the interference effects between neighboring structures where the centers of the segments are less than 3D (~ = (D1 + D2)/2, the average diameter of the two structures).

For structures near walls, the multiplier (l+A ) shall be used to increase the w

acceleration drag and the multiplier (l+D w ) shall be usedAto increase the standard drag. Bounding expressions for A and fl are given below as functions of X w w w (X = r/D 1/2, where r is the distance from the segment center to the boun-dary and D is the diameter of the structure):

0.05 < X < 1.0 A = 0.05/x w w Dw O.12/X w 5.2.51a Revision 2

NEDO2 1888 X <0.05 A = 1.0 V

D = 2.4 w

For structures with neighbors that are less than 3D away and within 300 of being parallel, the multiplier (1+A I

) shall be used to increase the accelera tion drag and the multiplier (1+D1) shall be used to increase the standard drag. Bounding expressions for A and D are given below as functions of I I (X = r IDi, where r is the distance between segment centers):

I 12 12 D

<20 A 02 2 0.05 (X1 I X~D +D~

1 1 2 D = O.2/X I I where D is the diameter of the structure under considerzaion and D is the 1 2 dkmeterof the neighbor. If more than one neighbor must be considered, the A and D values may be summed over the neighbor structures. For X less I I I than 0.05, the two neighbor structures shall be considered as an effective single structure.

The effects of wall proximity and neighbor structures may be superimposed in order to compute overall multipliers as follows:

(1 + A + ~ A1~)

(1+D +~DIk) 5.2.5.2 Load Definition Detailed procedures to calculate drag forces on submerged structures are described in Reference 5.2.51 which provides the methodology of utilizing the unsteady flow field generated by the TQuencher SIR? bubbles in the 5.2 .51b Revision 2

NEDO21888 suppression pool, to obtain drag loads on submerged structures, such as pipes, beams, etc.

The load definition first requires the establishment of the unsteady flow field within the torus by simulating the S/R? bubbles by point sources with strengths determined from the TQuencher SIR? bubble dynamics model (Reference 5.2.52). Then, drag loads on submerged structures are calcu-lated based on the previously established unsteady flow field.

a. Flow Field Establishment An oscillating TQuencher SIR? bubble is considered as a fluid source with source strength determined from the ~TQuencher.

SIRV1~ubb1eidynami mod 5.2~52). The TQuencher S/RV bubbles ~jiLhe Mark I ,tor~s.~are simulated by multiple sojirces.

in a finite~pool. By using potential flow theory and the method of images to account for the effects of solid walls and the free surface, the flow field within the torus is established.

b. Drag Loads Evaluation The drag force on a submerged structure consists of two compo-nents: the standard drag and the acceleration drag. The standard drag load is computed on a locally uniform flow field evaluated at the local velocity. The acceleration drag load is evaluated based on an inviscid, uniform but unsteady (accelerat-ing) flow field. The sum of these two drag forces gives the total drag load on a submerged structure.

5.2.5.3 Selection of Key Parameters for Load Evaluations This section outlines the key parameters which influence the magnitude and characteristics of the TQuencher S/R? bubble drag loads on submerged struc-tures. The procedure used to specify the input for the submerged structure drag load model is also discussed.

5.2.52 Revision 2

NEDO21888 The plant specific suppression pool geometry should first be identified as follows:

a. Torus shell dimensions

b. Torus water level

c. Relative locations of S/R? discharge lines in the suppression pool.

Then, the initial conditions of each SIR? discharge line are specified as follows:

a. Initial pipe pressure

b. Initial volume of airsteam mixture.

Next, the S/RVDL Clearing Model and the TQuencher SIR? Bubble Dynamics Model are applied as discussed in Sections 5.2.1 and 5.2.2 to determine the following:

a. S/RVDL discharge pressure

b. Clearing time

c. TQuencher bubble pressure and bubble radius history.

After dividing the structure into appropriate sections for more precise calculation, information on the structure section locations, orientations, acceleration drag volumes, projected areas and the standard drag coefficients are obtained.

With the above parameters identified, the dimensions of the pool model, the coordinates of the initial locations of the TQuencher S/R? bubbles, and the structure sections are calculated.

5.2.53 Revision 2

NEDO-21888 With the input parameters identified and calculated, the submerged structure drag load model is used to evaluate the resultant transient drag forces on each structure sections being analyzed. Typical results are shown in Figures 5.2.51 and 5.2.52. These figures show the TQuencher bubble drag forces on a downcomer in the radial CX) and tangential (Z) directions as a function of time.

5 .2 . 54 Revision 2

U, LU 0

0 U.

C

.4 4

0 I-TIME AFTER BUBBLE ENTERS POOL (sac)

Figure 5.2.51. Sample Predicted Time History of Total XForces on Downcomer 5.2.5s Revision 2

NEDO21888 TIME AFTER BUBBLE ENTERS POOL (sac)

Figure 5.2.52. Sample Predicted Time History of Total ZForces on Downcamer 5.2.56 Revision 2

NEDO-2 1888 REFERENCES FOR SECTION 5.2.5 5.2.5-1 F.J. Moody, et al., Analytical Model for Estimating Drag Forces on Rigid Submerged Structures Caused by LOCA and Safety Relief Valve Raxnshead Air Discharges, Supplement for TQuencher Air Discharges, General Electric Company, Report No. NEDO-21471-2, April 1979.

5.2.5-2 P. Valandani, Mark I Containment Program, Analytical Model for Computing Air Bubble and Boundary Pressures Resulting from a SIR? Discharge Through a T-Quencher Device, General Electric Company, Report No. NEDO-21878, September 1979.

5.2.57/5.2.58 Revision 2

NEDO21888 5.2.6 Thrust Loads on TQuencher Arms 5.2.6.1 Load Descriptions

a. Thrust Loads Along Axis of TQuencher Arms Following S/R? actuation the pressurization of the S/RVDL causes the water initially in the TQuencher and submerged portion of piping to be accelerated and expelled through the TQuencher arm holes into the suppression pool. The redirection of flow of water 90 degrees out the holes and the internal pressure of the arms results in thrust loads on the arm endcaps. By assuming a difference in the water flow rate into each arm of the TQuencher a net thrust load is obtained acting along the axis of the TQuencher device.

One arm of the TQuencher may have holes in the endcap. For quenchers of this design, an additional component of endcap thrust is produced due to differences in momentum of the fluid (water or steam) leaving the device.

An example of the predicted individual end cap thrust loads and the net thrust loading on a TQuencher with endcap holes is shown in Figure 5.2.61.

b. Thrust Loading Perpendicular to the TQuencher Arms As the water is expelled from the TQuencher device, irregularity in the shape of the gas/water interface (e.g. the interface may not be perpendicular to the quencher arm) may result in a net thrust load perpendicular to the arm due to momentum imbalance. An example of the predicted thrust loading perpendicular to the quencher arm is shown in Figure 5.6.22.

5.2.61 Revision 2

NEDO-21888 5.2.6.2 Bases and Assumptions

a. Thrust Loading Along Axis of TQuencher Arms The thrust load on the TQuencher along the arm axis is determined by performing a momentum balance (in the direction of the arm axis) on each arm separately to obtain the individual endcap thrust loadings.

The difference in the endcap loads is then computed to determine the net thrust load on the TQuencher device as a whole.

The analyses is separated into three phases:

(1) Before the gas/water interface reaches the first row of holes (2) After the gas/water interface reaches the first row of holes (3) After water clearing.

During the first phase the S/RVDL Clearing Model (see Section 5.2.1) is used to determine the water velocity and acceleration as it enters the TQuencher arms.

The velocity and acceleration of fluid flow from the holes during this period is determined using conservation of mass and an assumed flow split between the arms.

The inputs required to perform a momentum balance on each quencher arm during the second phase of the analysis are determined using the following assumptions:

(1) Slug flow is assumed.

(2) The acceleration of the water flowing from the quencher holes is assumed to remain constant in time at the value predicted 5.2.62 Revision 2

NEDO-21888 when the interface reaches the first row of quencher holes.

The gas driving the interface will begin to escape through the holes into the pool at this point. Therefore, the driving pressure will decrease, which results in a reduction in water clearing acceleration from the holes.

(3) In evaluating the internal pressure on the endcaps, the pressure in the arm is assumed to remain constant at the value predicted when the interface reaches the first row of holes. This is a bounding assumption since the pressure will actually drop significantly as the gas escapes through the quencher holes.

(4) The water velocity through the holes is the same for all holes for any point in time.

(5) The water acceleration through the holes is the same for all holes for any point in time.

(6) Following water expulsion, the steam flow through the end cap holes is assumed to be choked.

b. Thrust Loading Perpendicular to the Quencher Arms The thrust loading perpendicular to the TQuencher arms is determined by performing a momentum balance on each arm in the direction perpendicular to the arm axis. The analysis is performed using the following assumptions:

(1) The v4ocity and acceleration of water out the holes is computed using the assumptions in Section 5.2.6.2(a).

(2) As the gas/water interface travels down the quencher arm, a small length of the quencher arm has water exiting one side while gas exits the other side of the arm. This assumption is based on evaluation of movies taken of water/gas discharges from TQuencher devices during the 1/4 scale TQuencher Test Program (Reference 5.2.61).

5.2.63 R~vi5~in,, 9

(3) The gas momentum flow through the holes in the region of the arm immediately behind the gas/water interface is neglected.

As the interface travels down the arm the gas exiting the holes immediately behind the interface will not have sufficient time to establish a choked flow condition before water clears the holes directly opposite them. By assuming the gas velocity equals zero, a minimum gas momentum out one side of the arm is obtained and the net thrust loading is thus maximized.

5.2.6.3 Load Definition Procedure A summary of key parameters which influence the thrust loads is given below.

a. TQuencher arm geometry (i.e., with or without endcap holes)

b. Watermass flow rate and acceleration as a function of time

c. Time at which the air/water interface reaches the holes.

These parameters are evaluated for initial condition C (see Table 3.01) and the quencher arm geometry under consideration.

a. Thrust Loads Along TQuencher Arm Axis Since the gas/water interface location is being followed in the transient, the time at which the interface reaches the first row of holes should be determined from the S/R? discharge line clearing model output. This time separates the two phases of the method-ology. The water velocity and acceleration up to this time are readily available from the model output (see Section 5.2.1) and can be applied in the calculation of the thrust load for the before phase. The water velocity and acceleration predicted at the time the interface reaches the holes becomes the initial conditions for the after phase. The water velocity through the holes, the interface location, velocity and acceleration, and the 5.2. 64 Revision 2

NEDO 21888 average water acceleration in the arm are determined as a function of time as the water clears the arm. Following water expulsion, steam is assumed to flow through the end cap holes. The thrust load on each endcap and the net thrust load on the device is then determined as a function of time. Finally the three phases are combined to obtain the entire thrust load transient on the TQuencher device along the arm axis.

b. Thrust Loads Perpendicular to the TQuencher Arms Using the assumptions in Section 5.2.6.2(h), the net momentum flow per-pendicular to the arm is evaluated as the gas/water interface travels down the arm. From this, the net thrust load application point is deter-mined as a function of time.

5 .2. 65/5. 2. 66 Revision 2

NEDO21888 40.,..

NET THRUST ON T.OUENCHER HOLES 30.000 20.000 10,000 I I I C

70,000 60.000 50.000

- 40.000 0

4 0

-j 30.000 I-U)

~ 20.000 I

I-10,000 70.000 THRUST LOAD ON END CAP WITHOUT HOLES 60,000 50.000 (52% FLOW) 40,000 30.000 20.000 10.000 0

U 0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 TIME FROM S/RV OPEN (sac)

Figure 5.2.61 Thrust Loads on Arm Endcaps 5.2.67 Revision 2

NEDO21888 8

7 Cd)

LU

.4 60 I

IL 4 0 LU I 0 0

2 I-U, LU 9 5 ~

I-0 0 I-U-

4 LU

.4 0 2

C.) 4 I-0 U, 2

LU 0 0.

I-LU 2 0.

0 0.

4 2 0 0

.4 I- 3I-4 U) U

.4 Z 0.

I- 0.

4 0

4 0

.4 2

0 0 0.06 0.10 0.16 0.20 0.25 0.30 TIME FROM S/RV OPEN Iseci Figure 5.2.62. Generalized Shape of Thrust Loading Transient and Application Point With Time Revision 2 5.2.68

NEDO-21888 REFERENCES FOR SECTION 5.2.6 5.2.61 C. T. Sawyer, Mark I Containment Program, 1/4 Scale TQuencher Test, General Electric Company, Report No. NEDO24549, June 1979.

5. 2. 69/5. 2. 610 Revision 2

NEDO21888 5.2.7 Maximum S/RVDL and Discharge Device Pipe Wall Temperature The S/R? discharge pipe and discharge device are subjected to thermal expansion loading during a 51KV discharge which bounds any other thermal loading resulting from normal or accident conditions. The maximum (bulk pipe wall) temperature at any axial location along the discharge pipe is a function of the steady state internal pressure at that location during steam flow. Figure 5.2.71 presents an example of the predicted SIRVDL steady state temperature distribution during an SIR? discharge.

5.2.7.1 Bases/Assumptions The maximum S/R? discharge pipe temperature (upstream of the discharge device) at a given axial location along the line is assumed to be the saturation temperature corresponding to the steady state pressure predicted by the S/RITDL clearing model at that location. The analytical and empirical bases for the S/RvDL clearing model are documented in Reference 5.2.71.

The maximum discharge device temperature is assumed to be the saturation temperature corresponding to the maximum steady state stagnation pressure of steam expected to occur within the device.

The maximum steam stagnation pressure is assumed to be that whick would be required to develop the maximum expected S/RV steam flow (for all flark I S/R?s) through the discharge device exit holes under choked flow conditions.

The steam pressure density product required to do this was evaluated using the following relation:

k+l

~o~o k84 O.6A (i~r) 2(kl) } 2 lbflbm 2

P0 = stagnation steam pressure in discharge device, lbf/ft p0 = stagnation steam density in discharge device, lbm/ft3 5.2.71 Revision 2

NEDO21888 0.6 = exit hole discharge coefficient Ae = total exit hole area, ft2 k = ratio of specific heats

= gravitational constant, (lbmft)/(lbfsec2) m = steam flow through SIR? throat, lbm sec Comparison of the steady state SIRVDL temperature distribution and the dis-charge device temperature measured during extended SIR? discharges performed during full scale inplant testing, Reference 5.2.72, indicates that this approach provides a bounding thermal loading definition.

5.2.7.3 Load Definition Procedure The S/R? discharge line clearing model is used to evaluate the steady state pressure distribution along the discharge pipe applying procedures presented in Section 5.2.1.3. This pressure distribution is then converted to a temperature distribution using the steam tables and the assumption that the steam is saturated. The resulting temperature distribution is applied from the SIR? exit to the discharge device entrance.

The discharge device temperature is defined generically as follows:

TQuencher 3700F Ramshead 3900F 5.2. 72 Revision 2

440 aU.

LU a:

I 4

a:

LU a

LU I 2:

U a:

U-, RI I

ha U 0 2

I LU ha H

hO 03 I C U

a C ha z C

4 -J 0

4=-

IL U,

110 (U DISTANCE FROM S/RV EXIT ((tI 4

H 0,

H 0

El Figure 5.2.71. Example of Predicted S/RVDL and Discharge Device Temperature Distribution I)

NEDO21888 REFERENCES FOR SECTION 5.2.7 5.2.71 A. J. Wheeler, Comparison of Analytical Model for Computing Safety/

Relief Valve Discharge Line Transient Pressures and Forces to Monticello TQuencher Data, General Electric Company, Report No. NEDO23749, Addendum 1, September 1978.

5.2.72 R. A. Asai, et al., Mark I Containment Program Final Report, Monticello TQuencher Test, General Electric Company, Report No. NEDO21864, June 1979.

5. 2. 75/5. 2. 76 Revision 2

NEDO21888 5.2.8 Fatigue Cycles The actuation of S/R\Ys is an expected occurrence in plant operation. During the plant lifetime the containment will be subjected to numerous S/R? actuation transients. Therefore, fatigue loading due to SIR? discharge must be considered in the structural evaluation of the components affected.

The number of fatigue cycles for each of the various SIR? discharge loads dis-cussed in the previous sections is a function of the number of SIR? actuations assumed to occur during the plant lifetime. The total number of S/R? actuations is a function of the number of reactor transients which occur and the number of S/RV actuations for each.

The total number of transients from nonaccident operating conditions shall be based on extrapolation of plant operating history. In addition, SIR? actua tions due to accident (SBA or IBA) conditions shall be considered.

5.2.81/5.2.82 Revision 2

NEDO-21888 5.3 RANSHEAD LOADS The following sections define the S/RV discharge loads for a Ramshead discharge device. Since many of the loads are similar to those previously defined for a TQuencher discharge device, some of the Ramshead sections will refer to the appropriate TQuencher section.

5.3-1/5.3-2 Revia ion 2

5.3.1 S/RV Discharge Line Clearing Transient Loads The analytical model for S/Ky discharge line clearing calculates S/RVDL loads resulting from S/RV actuation. In addition, it calculates various parameters for use as input to other S/Ky loading analyses. The analytical basis for this model is documented in Reference 5.3.11 and verified against test data contained in Reference 5.3.12.

5.3.1.1 Load/Parameter Descriptions

a. S/RVDL Pressure When an S/RV opens, the pressure within the S/RVDL undergoes a transient prior to reaching a steady state value. A transient pressure wave travels back and forth in the line as the pressure continues to increase until the inertia of the water slug in the submerged portion of piping is overcome. During the water clear-ing transient, the pressures within the discharge pipe reach their maximum values. Following expulsion of the water slug, the peak pressure in the discharge pipe (the peak S/RVDL pressure occurs at the S/K? exit) approaches a quasisteady state value which is a function of the SIR? steam flow rate and friction along the line from the S/K? to the upstream entrance to the Ramshead device. Figure 5.3.11 presents a sample of a pre-diction of the S/RVDL pressure transient at the SIR? exit.

b. Thrust Loads on the S/RVDL When a SIR? opens, the high flow rate of steam into the discharge line results in the development of a pressure wave at the entrance to the line. This pressure wave travels to the air/water inter-face where it is reflected back to the entrance to the line.

This wave continues to travel back and forth in the discharge line until a steady state condition is reached when the pressure differential across the wave approaches zero.

5.3.11 Revision 2

NEDO2 1888 During the early portion of this transient, a substantial pressure differential exists across the pressure wave. There-fore, when the wave is within a S/K? pipe segment there exists a substantial difference in the pressure applied to the interior surface of the elbows on each end of the segment. This pressure differential, plus momentum effects from steam flowing around elbows in the line, results in transient thrust loads on the S/Ky discharge pipe segments. These loads are considered in the design of pipe restraints, the connection of the SIR? to the main steam line, and the Ramshead support system. Figure 5.3.12 presents a sample analytical model prediction of the thrust loading on a typical SIR? discharge pipe segment.

As the water slug in the submerged portion of the S/R\TDL is ex-pelled following a S/K? actuation, the redirection of water flow in the Ramshead device results in a thrust load transient on the last pipe segment. Figure 5.3.13 presents the location and posi-tive sense of this loading. Figure 5.3.14 presents sample pre-diction of the water clearing thrust load transient on the last pipe segment.

c. Ramahead Discharge Pressure Following a S/R? actuation, the air/steam mixture initially in the S/RVDL is compressed prior to discharge into the pool through the Ramshead device. The predicted pressure after water clearing at the Ramshead discharge is used as input to the analysis of SIR?

air bubble drag loads and torus shell pressures.

d. Mass Flow Rate of Water During Discharge When the S/R?DL and Ramshead are modeled in the S/RVDL clearing model, the mass flow rate of water through the Kamshead exit is calculated as a function of time.

5. 3.12 Revision 2

NEDO2 1888 The calculated mass flow rate of water is used as input to water jet load analyses as well as analysis of the water clearing thrust loads on elbows in the S/RVDL. Figure 5.3.15 presents an example of a water mass flow rate transient.

e. Mass Flow Acceleration of Water Through the Ramshead The mass flow acceleration through the Ramshead exit is computed by the S/RVDL clearing analytical model in a manner similar to the mass flow rate as outlined in Item d. above.

The mass flow acceleration is used as input to the analysis of water clearing thrust loads on S/RVDL elbows. Figure 5.3.16 presents an example of the water mass acceleration transient.

5.3.1.2 Bases and Assumptions The bases and general assumptions that are applicable to the line clearing analytical model for Ramshead discharge devices are identical to those for the TQuencher device. See Section 5.2.1.2 for details.

5.3.1.3 Load Definition Procedure

a. Outline of Procedure The following is a general outline of the procedure involved in evaluating the loads and parameters calculated by the line clear-ing analytical model.

Inputs for the line clearing model are prepared using Initial Conditions A through C for each S/RVDL under consideration. See Table 3.01. The initial water leg in the discharge line

5. 3.13 Revision 2

NEDO21888 is an input to the line clearing model. The water leg must be determined by use of the reflood analytical model (see Section 5.3.3) prior to using the line clearing model for subsequent actuation cases.

Next, the line clearing model is used to obtain transient values for the following parameters or loads: 1) S/RVDL pressure,

2) thrust loads on S/RVDL pipe segments, 3) Ramshead discharge pressure, 4) water slug mass flow rate, and 5) water slug mass acceleration.

The values obtained for Ramshead discharge pressure and/or mass flow rates and acceleration are used as input to evaluate the:

1) Torus Shell Pressure Distribution (Section 5.3.2), 2) Air Bubble Drag Loads (Section 5.3.5), 3) Water Jet Loads (Sec-tion 5.3.4), and 4) Maximum Discharge Pipe and Ramahead Environmental Temperature (Section 5.3.6).

b. Key Inputs The key inputs for Ramahead line clearing loads are identical to those for the TQuencher device. See Section 5.2.1.3 for details.

5.3

• 14 Revision 2

I-.

x LU a:

U, U I-C z

U LU RI IL NJ U LU H a: C a C

-J C C

z IL LU I

z

-J 0

a:

In TIME AFTER VALVE ACTUATION (see)

(U 4

H 0,

H 0

El ha Figure 5.3.11. Sample Prediction of the S/RVDL Internal Pressure Transient

U U LU 2:

I O U. RI 0 LU C ha H

I- 03 zLU 03 C,

LU U,

(b 4

H 0,

H TIME AFTER VALVE ACTUATION (sec) 0 El ha Figure 5.3.12. Sample Prediction of Thrust Load Transient at a Typical S/RVDL Segment

NEDO21888 TO S/RV S/RV DISCHARGE PIPE SEGMENT THE TRANSIENT THRUST LOAD IS THE NET LOAD ON THE PIPE SEGMENT TO AND IS POSITIVE IN THE DIRECTION SHOWN DISCHARGE DEVICE Figure 5.3.13. Sense of Thrust Loading 5.3.17 Revision 2

NEDO21888 20 10 10 20 30 0

40 C

n o 50 4

0

.4

~-60 zI-

-70 2

~-80

.4 0

LU 9Q 4

-100 110

-120 130 140

-1~u 0 0.1 0.2 0.3 TIME AFTER VALVE ACTUATION (sac)

Figure 5.3.14. Sample Prediction of Water Clearing Thrust Load On the Last Pipe Segment of the S/RVDL 5.3. 18 Revision 2

(2 0

X nE LU z I-4 RI U a: 0 U 0-J N)

I U. H

.0 a: C C

w I C C

4 0 0.05 0.10 0.15 0.20 (U

4 H TIME AFTER VALVE ACTUATION (see) 0, H

0 El ha Figure 5.3.15. Sample Prediction of Mass flow Rate of Water Exiting Ramshead

NEDO2 1888 U) 0 C

(NC)

E

.0 2

2I-4 LU

.4 LU 0

0 4

LU I-4 TIME AFTER VALVE ACTUATION (sac)

Figure 5.3.16. Sample Prediction of Water Mass Acceleration During S/R? Discharge 5.3.110 Revision 2

NEDO2 1888 REFERENCES FOR SECTION 5.3.1 5.3.11 Mark I Containment Program Analytical Model For Computing Transient Pressures and Forces in the Safety/Relief Valve Discharge Line, General Electric Company, Report No. NEDO23749, February 1978.

5.3.12 E. A. Buzek, et al., Final Report InPlant S/RV Discharge Load Test Monticello Plant (Kamshead), General Electric Company, Report No. NEDC21581, August 1977.

5.3.111/5.3.112 ReviSion 2

NEDO21888 5.3.2 Ramshead Torus Shell Pressures Loads on the torus shell resulting from air discharge from a Ramshead device are calculated using two separate analytical models. The bubble dynamics model calculates the air bubble pressure, radius and location as a function of time. The torus shell load model calculates the maximum and minimum pressures and the spatial pressure distribution on the torus shell.

The torus shell load definition procedure involves determining the wall pressure amplitude and frequency of an actual system, and applying these values to an experimentally measured wall pressure/time history that has been normalized. The procedure to conservatively predict loads on the torus shell involves the application of experimentally determined multipliers to the analytical prediction of the peak pressures from the torus shell pressure model. The adjusted magnitude of the shell load is then applied to an experimentally determined torus shell pressure/time history. Additionally, the frequency of the experimental pressure/time history is adjusted to account for line unique variations of the actual bubble frequency.

5.3.2.1 Load Description The actuation of an SIR? results in the discharge of water from the S/Ky discharge line followed by air and then steam. Subsequent to the water discharge, the air discharge from the S/RVDL forms two high pressure oscil-lating bubbles. The bubbles will rise as a result of buoyancy effects.

The positive and negative dynamic pressures developed within these bubbles are transmitted to the torus shell. Figure 5.3.21 presents an example of the torus shell pressure transient resulting from an S/R? discharge through a ramshead device. Figures 5.3.22 and 5.3.23 present examples of torus shell radial and longitudinal pressure distributions, respectively.

5.3. 21 Revision 2

NEDO21888 5.3.2.2 Bases and Assumptions The theoretical basis for the mathematical models which predict the torus shell pressure loads is described in Reference 5.3.21. Additionally, experimental verification and empirical adjustmeiit of the predicted torus shell pressure loads are used to justify the model and load definition procedure (See Reference 5.3.21).

When evaluating inputs for the bubble dynamics and torus shell pressure distribution analytical models the following general assumptions apply:

a. The bubble distance from the centerline of the Ramshead discharge device is 6.25 ft. This was empirically determined from the measured pressure distributions (see Reference 5.3.21).

b. The Mark I torus suppression poo1 is modeled (for the torus shell pressure distribution model) as an equivalent rec-tangular cross section. This equivalent geometry is based on conservation of the pool water volume and pool depth. The results of thic procedure are in good agreement with Monticello test data as reported in Reference 5.3.21. Reference 5.3.21 (Appendix D) contains the detailed basis and justification for the equivalent rectangular geometry assumption.

c. The pool water level is at the technical specification high water level. This assumption results in the maximinn initial water leg in the S/RVDL which in turn results in the highest loads on the S/RVDL and discharge device.

d. The S/R?DL vacuum breaker does not leak, i.e., for first actuations, the initial water leg in the S/K?DL is determined by the pressure differential between the S/K?DL and wetwell air space. It is assumed equal to the drywell to wetwell A?

minus the vacuum breaker set point. By assuming the 5.3.22 Revision 2

NEDO21888 vacuum breaker does not leak, a lower value of S/KVDL to wetwell L~P is calculated which results in a longer initial water leg in the dis-charge line. Since longer water legs result in higher loads, the assumption of a non leaking vacuum breaker is conservative.

e. The maximum S/RVDL reflood, which is predicted to occur at a time in the transient after the minimum predicted time between SIR?

actuations (based on plant unique analyses), is used as the initial water leg for subsequent S/Ky actuation evaluation. Optionally, if plant unique analyses are not performed, the maximum reflood occurring at any time during the ref lood transient is to be used as the initial water leg for such evaluations.

5.3.2.3 Load Definition Procedure

a. Outline of Procedure The following is a general outline of the steps involved in obtaining the pressure loads on the torus shell. Figure 5.3.24 presents the flowchart delineating this procedure.

First, the necessary inputs to the line clearing analytical model are prepared for Initial Conditions A through C (Table 3.01) for each S/RVDL under consideration (see Section 5.2.1.3).

Ndxt, the line clearing analytical model is used (for each Initial Condition) to obtain the discharge pressure at the Kamshead after water clearing.

The necessary inputs are then prepared for each of the Initial Conditions (for each S/R?DL) for the bubble dynamics model.

General assumptions used in preparing the inputs are given in Section 5.3.2.3(b).

5.3.23 Revision 2

NEDO2 1888 Next, inputs are prepared and the torus shell pressure distribution model is used to obtain the torus shell pressure transient at selected locations on the shell for a valve actuation from each S/RVIYL dis-charge device for each Initial Condition.

It is then necessary to compute a multiplier* to be applied to the torus shell pressures, which are output from the torus shell pres-sure distribution model, to adjust for condition dependent (first actuation, leaky valve or subsequent actuation) and location dependent (azimuthal attenuation) effects. See Reference 5.3.21 for the values of the multipliers.

The adjusted maximum positive and maximum negative shell pressures are then applied to the amplitude scale of the appropriate empirically derived pressure time history traces (see Figures 5.3.21, 5.3.25 and 5.3.26).

The next step is to adjust the time scale nf the appropriate condi-tion dependent pressure history to account for the actual bubble frequency due to S/RVDL differences. This is accomplis~hed ~by determining a time scaling parameter ( ~).

The time scaling parameter (T

5) to be applied to the appropriate empirical pressure/time history (see Figures 5.3.21, 5.3.25, and 5.3.26) to stretch or compress the time scale is computed from the following equations.

= ~R (time scaling parameter)

• The basis and justification for the multipliers are contained in Reference 5.3.21.

5. 3. 24 Revision 2

NEDO2 1888 where 6.9 Hz (from equation below using Monticello test data).

The frequency prediction for the given S/RVDL is:

1 3kg P0j144)

K2 It should be noted that the air bubble frequency prediction is derived from the Rayleigh equation assuming adiabatic compression and expansion. See Reference 5.3.21 for a description of this equation.

The calculated frequency of the torus shell pressure transient is assumed accurate to +/-15percent for first actuation, leaky valve, and subsequent actuation cases. This uncertainty is based on consideration of data scatter observed in full scale testing (Reference 5.3.21), and the accuracy with which the above frequency equation has been shown to predict torus shell pressure loading frequencies over the range of operating conditions. Due to the above uncertainty, the structural evaluation is performed using the calculated frequency within the specified uncertainty range, which is expected to result in the maximum structural response.

Each load case is then evaluated under the appropriate Initial Condition given in Table 3.0-1. For multiple valve actuations, the peak positive and negative shell pressures predicted at a given location for each S/K? discharge device are to be combined by calculating the square root of the sum of the squares (SRSS) of the peak pressure loads predicted to occur at that location for each individual S/R? actuation (see Section 5.2.2.2).

Following evaluation of the torus shell pressure distributions for the Load Case/Initial Conditions shown in Table 3.01, the pres-sure loadings are to be combined with other torus shell loads occurring during the same time period per the load combination bar charts presented in Section 3.0.

5.3.25 Revision 2

NEDO-21888

b. Key Inputs There are several parameters that must be evaluated in order to apply the bubble dynamics analytical model. These include the thermo-dynamic state of the gas in the S/RVDL and geometric considerations.

Such parameters include:

• Pressure in the discharge device during air discharge

• Initial air mass in the S/RVDL

• Initial gas temperature and pressure in the S/RVDL

• Initial gas volume in the S/RVDL

• Suppression pool temperature and pressure at the discharge device exit

• Suppression pool geometry

• Configuration of Ramaheads in the suppression pool Similarly, the key inputs that must be evaluated to apply the torus shell pressure distribution model are as follow~:

• Torus shell geometry

• Configuration of Ranisheads in the suppression pool

• Maximum bubble radius

• Minimum bubble radius

• Maximum bubble pressure

• Minimum bubble pressure

• Bubble oscillation frequency

• Bubble azimuthal location 5.3.26 Revision 2

LU IL LU IL a

-J

-J LU 2:

z RI U,

U a

LU Cd N

3 H ha C C C

-4 C IL 0

z (U

4 TIME FROM VALVE ACTUATION (see)

H 0,

H 0

Figure 5.3.21. Normalized Shell Pressure Transient for any Location on the Pool Wall Due to an S/RV Discharge for a Cold Pipe (Monticello Test Data)

NEDO21888 TO PRESSURE VESSEL

• NORMAL WATER LEVEL (NWL)

TORUS SHELL Figure 5.3.22. Normalized Positive Pressure Distribution at the Torus CrossSection Coincident with the Ramshead ~. Ramshead Located on Torus Centerline 5.3.28 Revision 2

1.~~

a -

-J

-J LU U x z RI U 0.6 N

t N)

.0 C H C

a: C 0 03 2 0.4 LU 02 0 5 I 10 I 15 I 20  :~s 35 45 DISTANCE FROM RAMSHEAD ~ (ftl H

0, H

Figure 5.3.23. Sample Prediction of Torus Shell Longitudinal Pressure Distribution ha Resulting from an S/RV Discharge Through a Ramshead Discharge Device

DISCHARGE MAXIMUM BUBBLE U 2:

RI U

N)

N)

I-A o PEAK POSITIVE AND C SHELLPRESSURES.P C 03 SCALED TORUS SHELL PRESSURE TIME HISTORIES (U

4 H

U, H

0 Figure 5.3.24. Procedural Flowchart ha

1.0 LU IL w

a: Z a.

U U W I NJ N) (I~ H I 0 03 LU C I N C

3 C

IL 0

z 1.,

(U 4 TIME FROM VALVE ACTUATION (see)

H 0,

H 0

El Figure 5.3.25. Normalized Shell Pressure Transient for any Location on the Pool Shell Due to an S/RV Discharge Leaky Valve (Monticello Test Data)

LU a:

In In LU a:

a

-J U X z RI U o 0 ha LU NJ I N H

3C H C ha ~ C a: C 0

z TIME FROM VALVE ACTUATION (see)

(U 4

H 0,

H Normalized Shell Pressure Transient for any Location on the Pool Figure 5.3.26.

ha Shell Due to a Subsequent Valve Actuation NWL (Monticello Test Data)

NEDO2 1888 REFERENCES FOR SECTION 5.3.2 5.3.21 Safety Relief Valve System Analytical Models for Use with Ramshead Discharge Devices, General Electric Company, Report No. NEDO23803, August 1979. I 5.3.213/5.3.214 Revision 2

NEDO-21888 5.3.3 S/RVDL Reflood Transient (Ramshead The ref lood analytical model is used to determine the transient water rise in an S/RVDL following SIR? closure. Outputs from the reflood model are used to provide inputs to the models which calculate loads during a sub-sequent S/K? actuation. For details of the procedure used to calculate reflood parameters refer to Section 5.2.3.

5.3.3.1 Bases The ref lood procedure and analytical model for the Ramshead discharge device are identical to those used for the TQuencher discharge device.

The only differences are model input parameters dependent on the geometry of the discharge device (e.g., loss coefficients, volumes, inertial lengths, and average areas). The analytical model is described in detail in Reference 5.3.31.

This reference also shows that the analytical model conservatively predicts test data which is documented in Reference 5.3.32.

5.3.31/5.3.32 Revision 2

NEDO21888 REFERENCES FOR SECTION 5.3.3 5.3.31 A. .3. Wheeler, et al., Analytical Nodel for Computing Water Rise in a Safety/Relief Valve Discharge Line Following Valve Closure, General Electric Company, Report No. NEDO23898, August 1979.

5.3.32 E. A. Buzek, et al., InPlant S/RV Discharge Load Test Monticello Plant (Ramshead), General Electric Company, Report No. NEDC21581, August 1977.

5. 3. 33/5. 3. 34 Revision 2

NEDO-21888 5.3.4 Ranshead Water Jet Loads on Submerged Structures 5.3.4.1 Load Description An S/RV actuation causes the discharge of a water column into the suppression pool. As the water column leaves the Ramshead discharge device and enters the suppression pool, it forms a jet which may load structures inside the torus which are fully or partially intercepted by the discharge. The calculation of these loads is described in this section.

Jet loading lasts from the time that the jet reaches the structure until the time when the last particle of the water slug passes the structure. Fig-ure 5.3.41 presents a sample predition of this load.

5.3.4.2 Bases for Methodology The calculation procedure described to obtain Ramshead jet loads is based on the analytical model described in Reference 5.3.41. S/RV line unique dis-charge jet velocities and accelerations are obtained from the S/RVDL clear-ing model as described in Section 5.3.1. The major assumptions included in the methodology are as follows:

a. When a structure is engulfed in the jet path, the force on the structure can be calculated using a standard velocity squared drag equation. If a structure partially or fully intercepts the jet, a momentum balance is used to define a proportionality factor to be used in place of the standard drag coefficient.

b. The jet is completely dissipated when the last fluid particle leaving the Ramshead reaches the jet front; the dissipated jet provides no more drag.

5.3.4.3 Load Definition Procedure The procedure used to calculate forces on submerged structures in the Ramshead jet path is presented below.

5.3.41 Revision 2

NEDO21888 First, from the S/RVDL clearing model described in Section 5.3.1 the time dependent discharge velocity and acceleration are obtained. The parameters involved are:

-r = An identifier equal to the time at which a jet particle is discharged from the Ramshead, sec V (r) = Jet discharge velocity at time -r, ft/sec D

dv Cr)

Cr) di = Jet discharge acceleration at time , ft/sec2 Next the maximum jet penetration distance is determined by V 2 ~-r D max ADCr max )~ft where -r denotes the time at which the water jet clears the Ramshead. If max i < L 5, there is no jet load on the structure. If £ > L5, there will be jet loads on the structure. L5 = distance (x) from the Ramshead discharge to the structure of concern.

From the Ramshead clearing data, a plot of the jet front position versus time is constructed. This is done by plotting the position (denoted as x.) versus J

time (t being the elapsed time following the start of the S/RV discharge history of a number of fluid particles which exit the Ramshead during the clear-ing transient). This timespace plot (t,x) is based on the following equation:

V Qr) (t r), ft D

Whenever a line of constant T overtakes an earlier line, the earlier line ceases to extend, and the overtaking line continues. The timespace plot enables one to enter a location x. read the corresponding value of ~r,and then obtain associated jet velocity from the Ramshead clearing.

5.3.42 Revision 2

NEDO21888 For a structure at position x., values of -r and t can be obtained from the U

time space plot. Then the equation below and the Ramshead clearing data is employed to determine the corresponding jet area. The jet area is given by:

A AV (r) 2 A(x.,t) 1 D DD

_____ ft 1 VDCr) ADCr) (t-r) 1 (vD~)) A 13Qr)x.

where A is the discharge area of the S/RV line.

D The area of the jet front is approximated by the area the jet possesses one initial discharge diameter (d = (4A/~) 1/2 ) removed from the front. An additional computation of the jet radius determines which structures are overlapped by the jet, and to what degree overlap occurs.

The standard drag force on the submerged structure is based on the normal component of velocity intercepting the structure, the projected area inter-cepted by the normal component of velocity, and the jet area profile.

If a structure is fully submerged inside the jet boundary at time t, a velocity squared drag calculation is used.

(VCr) cos 6)2 F~(t) = CD A~ 2

• lbf Where A represents the projected area intercepted by the jet.

p If a structure fully or partially intercepts the jet (no jet flow around structure) the following equation applies:

F(t) = K~A~ (VCr) 2gcos 6)2 VCr), lbf c

Where K = 2 for momentum stoppage, or K = 4 if the structure turns the jet

.3 .3 back on itself.

5.3.43/5.3.44 Revision 2

NEDO2 1888 20.000 15,000 10,000 U.

5,000 0

0.18 0.19 0.20 0.21 TIME AFTER JET CONTACTS STRUCTURE (sec)

Figure 5.3.41. Sample Prediction of a Typical Structure Loading History 5.3.45/5.3.46 ReviBion 2

NEDO21888 REFERENCES FOR SECTION 5.3.4 5.3.41 F. .3. Moody, Analytical Model for Liquid Jet Properties for Predicting Torus on Rigid Submerged Structures. General Electric Company, Report No. NEDO21472, April 1978.

5.3.47/5.3.48 Revision 2

NEDO2 1888 5.3.5 Ramshead BubbleInduced Drag Load on Submerged Structures Oscillating bubbles resulting from S/RV actuation create an unsteady three dimensional flow field and therefore induce acceleration and standard drag forces on the submerged structures in the suppression pool.

This section addresses the load definition procedures for loads on sub-merged structures due to Ramshead S/RV bubbles. The following subsections discuss the methodology to define the loads; the bases and assumptions on which the analytical models are developed and verified; and the selection of key parameters from typical plants for which the design load on sub-merged structures can be obtained.

5.3.5.1 Bases and Assumptions The bases and assumptions of the analytical model are derived from the work in Reference 5.3.51. The major assumptions are summarized as follows:

a. The total drag on submerged structures is the sum of the acceleration and standard drags.

b. The flow fields caused by Ramshead S/RV bubbles are described by fluid sources. The presence of boundaries is accounted for by using the method of images.

c. Source strengths are determined from the Ramshead bubble dynamics model (Reference 5.3.52).

5.3.5.2 Load Definition Detailed procedures to calculate drag forces on submerged structures are described in Reference 5.3.51 which provides the methodology of utilizing the unsteady flow field generated by the Ramshead S/RV bubbles in the suppression pool to obtain drag loads on submerged structures such as pipes, beams, etc.

5.3.51 Revision 2

NEDO2 1888 The load definition first requires the establishment of the unsteady flow field within the torus by simulating the S/RV bubbles by point sources with strengths determined from the Ramshead S/RV bubble dynamics model (Reference 5.3.52). Then, drag loads on submerged structures are calculated based on the previously established unsteady flow field.

a. Flow Field Establishment An oscillating Ramshead S/RV bubble is considered as a fluid source with a source strength determined from the Ramshead S/RV bubble dynamics model (Reference 5.3.52). The Ramshead S/RV bubbles in the Mark I torus are simulated by multiple sources in a finite pool. By using potential flow theory and the method of images to account for the effects of solid walls and the free sur-face, the flow field within the torus is established.

b. Drag Loads Evaluation The drag force on a submerged structure consists of two components:

the standard drag and the acceleration drag. The standard drag load is computed on a locally uniform flow field evaluated at the local velocity. The acceleration drag load is evaluated based on an inviscid, uniform but unsteady (accelerating) flow field. The sum of two drag forces gives the total drag load on a submerged structure.

5.3.5.3 Selection of Key Parameters for Load Evaluations This section outlines those key parameters which influence the magnitude and characteristics of the Ramshead S/RV bubble drag loads on submerged struc-tures. The procedure used to specify the inputs for the submerged structure drag load model is also discussed. The plant specific suppression pool geometry should first be identified as follows:

a. Torus shell dimensions

b. Torus water level 5.3.52 Revision 2

NEDO-21888

c. Relative locations of the S/RV discharge lines in the suppression pool.

Then, the initial conditions of each S/RV discharge lines in the suppression as

a. Initial pipe pressure

b. Initial volume of airsteam mixture.

Next, the S/RVDL clearing model and the Ramshead S/RV bubble dynamics model are applied as discussed in Sections 5.3.1 and 5.3.2 to determine the following inputs:

a. Discharge pressure

b. Clearing time

c. Ramshead bubble pressure and bubble radius history.

After dividing the structure into appropriate sections for more precise calculation, information on the structure section locations, orientations, acceleration drag volumes, projected area and the standard drag coefficients should also be obtained.

With the above parameters identified, the dimensions of the pool model, the coordinates of the initial locations of the Ramshead S/RV bubbles and the structure sections can be calculated.

With the input parameters identified and calculated, the submerged structure drag load model is used to evaluate the resultant transient drag forces on each structure sections being analyzed. Typical results are shown in Fig-ures 5.3.51 and 5.3.52. These figures show the Ramshead bubble drag forces on a downcomer and a vent header support column in the radial (X) and tangential (Z) directions as a function of time.

5. 3. 53/5 .3. 54 Revision 2

NEDO 21888 LU C-)

0 U.

x

-J 0

TIME AFTER BUBBLE ENTERS POOL (sac)

Figure 5.3.51. Sample Predicted Time History of Total XForces on Downcomer and Vent Header Support Column 5.3.55 Revision 2

NEDO21888 (0

LU U

0 IL N

-J C

I-0 TIME AFTER BUBBLE ENTERS POOL (sac)

Figure 5.3.52. Sample Predicted Time History of Total ZForces on Dowucomer and Vent Header Support Column 5.3.56 Revision 2

NE DO- 21888 REFERENCES FOR SECTION 5.3.5 5.3.51 F. J. Moody, et al., Analytical Model for Estimating Drag Forces on Rigid Submerged Structures Caused by LOCA and Safety Relief Valve Ramshead Air Discharges, General Electric Company, Report No.

NEDO21471, September 1977.

5.3.52 Mark I Containment Program, SafetyRelief Valve System Analytical Models for Use with Ramshead Discharge Devices, General Electric Company, Report No. NEDO23803, August 1979.

Revision 2

NEDO21888 5.3.6 Maximum S/RVDL and Discharge Device Wall Temperature The procedure to determine the thermal loads associated with the S/RV dis-charge line and the Ramshead device is identical to that for the TQuencher device. See Section 5.2.7 for details of the procedure.

I 5.3.61/5.3.62 ReviBion 2

NEDO-2 1888 5.3.7 S/RVDL Fatigue The procedure to determine the fatigue usage factor for lines terminating in a Ramshead discharge device is identical to that for a TQuencher discharge device. See Section 5.2.8 for details of the procedure.

5.3.71/5.3.72 Revision 2

NEDO21888 SECTION 6 OTHER CONS IDERAT IONS Revision 2

NEDO21888 6.0 OTHER CONSIDERATIONS Section 6 discusses LOCA and S/RV loads not addressed in Sections 4 and 5.

These loads are of secondary importance because they are negligible when compared to other loads defined for the structures affected. Each potential loading condition is described and justification is given for its classifica-tion as negligible.

6.01/6.02 Revision 2

NEDO21888 6.1 SEISMIC SLOSH LOADS In the event of an earthquake, the Mark I torus will be subjected to both horizontal and vertical oscillatory motions which will produce waves on the surface of the suppression pool (seismic slosh). These waves will generate pressure loads on the torus shell. Unsteady motions in the pool water will also generate drag loads on submerged structures.

Scale model tests of the Mark I torus were conducted to determine the seismically induced slosh amplitude and slosh pressures (Reference 6.11). The tests were performed over the range of Mark I Final Safety Analysis Report (FSAR) seismic criteria. Calculated slosh amplitudes for the full scale Mark I plants are small, as shown in Table 6.11. These results are conservative since the seismic response spectrum used in the seismic slosh tests is an envelope of the individual responses for all the Mark I plants. The magnitude of the maximum local slosh pressures on the full scale torus shell was found to be less than 0.8 psi. This pressure is insignificant when compared to the other torus wall loads occurring during a postulated LOCA.

The drag loads on submerged structures due to the vertical pool motion can be conservatively evaluated by using the following equation:

p = (o.ooosn2 cos (l.684t) fcos (l.684t)J 0.001 nDsin (l.684t)) (#-) sin e where P net drag load on structure (psi). Applied normal to the structures longitudinal axis.

n = maximum slosh amplitude, in.

t time, sec 6.11 Revision 2

NEDO- 21888 D = diameter of cylinder circumscribing the structure, in.

r = distance from center of torus to object, ft R = Initial water level plus the maximum slosh amplitude, ft 0

6 = angle between the vector normal to the structures? longitudinal axis and a line drawn from the center of the torus to the point of application (see Figure 6.11), rad.

The above equation was developed assuming sinusoidal motion and used the maximum drag coefficients for oscillatory flow from Reference 6.1.2. It is assumed to apply to any structure below the height defined by the addition of the torus high water level and the maximum slosh amplitude. Large struc-tures should be divided into smaller sections for force calculations.

Figure 6.11 illustrates the application of the above equation.

6.12 Revision 2

Table 6.11 MARK I CONTAINMENT SEISMIC SLOSH RESULTS Maximum Slo~h Amplitudes at Torus Wall Plant (inches Browns Ferry 1, 2, 3 21.0 Brunswick 1, 2 15.5 Cooper 19.8 Dresden 2, 3 21.2 Duane Arnold 11.2 Fermi 2 16.7 Fitzpatrick 15.5 Hatch 1, 2 14.9 Hope Creek 1, 2 20.8 Millstone 16.4 Monticello 11.1 Nine Mile Point 1 10.2 Oyster Creek 20.7 Peach Bottom 2, 3 12.6 Pilgrim 14.3 Quad Cities 1, 2 25.0 Vermont Yankee 12.9

6. 13 Revision 2

NEDO21888 INITIAL POOL SURFACE TORUS CENTER Figure 6.11. Application of Seismic Slosh Drag Load 6.14 RevisiOn 2

NEDO21888 REFERENCES FOR SECTION 6.1 6.11 5. Arain, Seismic Slosh Evaluation, General Electric Company, Report No. NEDC23702, March 1978.

6.12 T. Sarpkaya, Forces on Cylinder Near a Plane Boundary in a Sinusoidally Oscillating Fluid, JournaZ of Fluids Engineering, September 1976.

6.15 Revision 2

NEDO21888 6.2 POST POOL SWELL WAVES Following a LOCA, random pool surface waves are present in the torus as a result of two sequential wave formation mechanisms:

a. The fallback of the bulk pool swell after the DBA and the subsequent pool motion which is generated and dies out as the vent flow reaches a quasisteady state condition

b. The pool surface disturbance resulting from steam blowdown through the vent system during the subsequent condensation oscillation and chugging.

The basis for the evaluation of the wave behavior during the first regime is observed wave loads from the QSTF, Reference 6.21. The loading from these waves during and immediately after fallback is included in the pressure data directly incorporated into the torus shell pressure histories (Section 4.3.2).

Although the QSTF results are not scaled precisely for system behavior following bubble break through, they do give the correct order of magnitude for wave behavior. By review of the pool swell vertical loading generated during this time period, it is concluded that the wave loading is insignificant when compared with other vertical loads.

Loads resulting from waves generated during the second regime have been evaluated based on FSTF data (Reference 6.22). As FSTF torus shell pressure transducers measured these loads during condensation oscillation and chugging, they are inherently included in the load definition for these phenomena.

6.21/6.22 Revision 2

NEDO21888 REFERENCES FOR SECTION 6.2 6.2-1 .3. M. Humphrey, Mark I Containment Program, 1/4Scale Two Dimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

6.2-2 .3. E. Torbeck, Mark I Containment Program Full Scale Test Program Final Report, General Electric Company, Report No. NEDO24539, August 1979.

6.23/6.24 Revision 2

NEDO2 1888 6.3 ASYMMETRIC VENT PERFOR.MANCE During a LOCA, geometric effects in the Mark I vent system may cause differences in pool swell behavior around the circumference of the torus. As discussed in this section, these variations are already included in the loads presented in this report.

The circumferential variations in pool swell loads were studied in the 1/12 scale threedimensional tests performed at EPRI (Reference 6.31). Because all plants have similar downcomer spacing, one model was used to define the variations in longitudinal pool swell behavior for all plants. In that model1 one specific plant vent system and torus were accurately modeled to include the effects of nonuniform downcomer spacing and the flow distribution between dowcomers. It is the difference in the fL/D between downcomers which causes the variation in flow rate, and the test facility was designed to accurately simulate the fL/D values of the prototype vent system.

Near a downcomer pair, the pool behavior is determined by the expanding bubbles.

The pool water is pushed upward more rapidly directly over the bubbles, locally generating higher pool swell velocities and pool swell heights. At locations where the downcomer spacing is close, the resulting poo1 swell velocity and height are higher than at those locations where the spacing is further apart.

The EPRI test results were used as a basis for including the longitudinal variation of pool swell behavior. The following pool swell loads have been determined including that behavior:

a. Torus shell pressure histories (Section 4.3.2)

b. Vent system impact and drag loads (Section 4.3.3)

c. Impact and drag loads on structures above the pool (Section 4.3.4)

d. Froth impingement loads (Section 4.3.5)

e. Pool fallback loads (Section 4.3.6).

6. 31/6. Revision 2

NEDO 21838 REFERENCES FOR SECTION 6.3 6.3-1 The Electric Power Research Institute, ThreeDimensional Pool Swell Modeling of a Mark I Suppression System, EPRI NP906, October 1978.

6.33/6.34 Revision 2

NEDO21888 6.4 DOWNCOMER AIR CLEARING LATERAL LOADS During the initial phase of a postulated Loss of Coolant Accident (LOCA),

the rapid clearing of air through the downcomers causes them to be sub-jected to lateral loads. The predominant load in this phase has been reported in Section 4.2 under Vent System Thrust Loads. Additionally, a minor dynamic air clearing lateral load was observed in the Mark I Full Scale Test Facility (FSTF) data. This secondary load is bounded by the condensation oscillation lateral load defined in Section 4.4.3, and therefore should not have to be considered separately. The procedure for defining it, however, is presented in this section for completeness.

6.4.1 Bases and Assumptions The bases and assumptions used for defining air clearing lateral loads on downcomers are the same as those given in Section 4.4.3.

6.4.2 Evaluation Procedure A maximum downcomer lateral load of 2,172 lbs was determined from the dynamic portion of the FSTF Design Basis Accident (DBA) air clearing data to occur within +/-22~l/2c of the EastWest direction, i.e., in the plane of the dowucomer pairs as identified by sectors 4 and 5 in Figure 6.41. The scaling procedure used to obtain the maximum design load for an individual Mark I plant dowucomer is the same as that described in Section 4.4.3.

6.4l/6.4...2 Revision 2

NEDO21888 SECTION OF DowNcoMER/

HEADER STRUCTURE N

0 -

292.50 67.5~

E 2700 900 -~

247.50 112.50 202.50 i8O~

Figure 6.41. Sectors Used to Define Directions of Lateral Loads on Dowucomer s End Revision 2 6 .43/6. 44

NEDO-21888 6.5 SONIC WAVE With the postulated instantaneous rupture of a steam or recirculation line, a pressure wave traveling at sonic velocity would expand from the break location into the drywell atmosphere and through the vent system. The wave amplitude, at the break location and at the time of the break initiation, would be substantially below the reactor operating pressure and would attenuate rapidly as it expanded out from the break location into the large drywell volume. The wave amplitude after entering the vent system would be nearly uniform, but greatly attenuated from its initial value at the break location.

The insignificance of the sonic wave was determined from the tests performed on the FSTF (Reference 6.42) and QSTF (Reference 6.51).

FSTF data traces show some evidence of a shock wave at break initiation but this is due in part to the arrangement of the discharge pipe configuration for the simulated break flow. The sonic wave produced by the rupture disc was not proto-typical since the rupture disc led into a long, empty line in which a strong pres-sure wave would be expected. In addition to this, the break flow is discharged to one location in the cylindrical FSTF drywell which has no- other structures contained in it. In the Mark I plants the large drywell volume would greatly attenuate any shock wave, and scattering of the shock wave due to the equipment and piping located inside the drywell would further reduce the effects of any sonic wave.

The FSTF had a rupture disc arrangement upstream of the drywell which simulates the instantaneous rupture of the steam or recirculation pipe. In the FSTF dis-charge pipe section, the rupture disc was located downstream of the orifice which simulates the break area. It should be noted that in the QSTF discharge pipe arrangement, the rupture disc was located just upstream of the orifice used to simulate the break area. Further, in FSTF the critical flow area of the rupture disc which generates the shock wave was much larger (by a factor of 3) than the 6.51 Revision 2

NEDO21888 throat section of the orifice (simulated break area). Therefore, the initial shock wave generated in FSTF is expected, but is not prototypical for DBA conditions.

Additionally, the FSTF arrangement as discussed earlier, did not allow for the prototypical attenuation of this sonic wave as would occur in the Mark I plants.

The QSTF data confirms that the sonic wave generated in FSTF was not prototypical.

Examination of the QSTF data shows no evidence of a sonic wave in the drywell, vent system, or wetwell. Since in QSTF the rupture disc was located upstream of the flow orifice which simulates the break area, the strength of the sonic wave is better simulated than in FSTF. Note also that QSTF also had an empty discharge line downstream of the orifice and a drywell with no structures to scatter the wave.

Considering the above discussion, it is concluded that the loading on the vent system and torus due to sonic waves is negligible.

6.52 Revision 2

NEDO21888 REFERENCES FOR SECTION 6.5 6.51 .3. M. Humphrey, Mark I Containment Program 1/4 Scale TwoDimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO21944, August 1979.

6.52 .3. E. Torbeck, Mark I Containment Program Full Scale Test Program Final Report, General Electric Company, Report No. NEDO24539, August 1979.

6.53/6.54 Revision 2

NEDO21888 6.6 COMPRESSIVE WAVE Thc rapid bulk pressurization of the drywell immediately following a postulated DBA prior to vent clearing would theoretically result in a weak compressive wave being generated in the downcomer water leg that would travel across the pool to the torus shell.

The insignificance of the compressive wave was determined from tests performed on both the QSTF and the FSTF, see References 6.61 and 6.62, respectively.

If the wave phenomena were significant, its presence would have been observed in the QSTF, because the drywell pressure response and wetwell pressure response were accurately scaled. Additionally, the pressure transducers used in the QSTF were capable of measuring response at frequencies up to 200 Hz, which would have been sufficient to detect a compressive wave. Examination of wetwell submerged pressure test data indicates no evidence of a compressive wave. The same observation has been made in data obtained from the FSTF which also has instrumentation capable of detecting the presence of such a wave.

The energy of the presumed compressive wave is dissipated in accelerating the water slug in the downcomers into the pool. It is also attenuated as the wave travels from the downcomer exit to the torus wall. Therefore, the resulting pressure on the torus wall generates negligible loads as verified by the QSTF and FSTF tests.

Even though the presence of the compressive wave has not been specifically observed, its effect has already been included in the load definition for both the torus net vertical loads (Section 4.3.1) and the torus shell pressure histories (Section 4.3.2) because the QSTF was built and operated in accordance with the pool swell scaling laws. Thus, the potential compressive wave behavior is accurately included in :he test results.

6.61/6.62 Revision 2

NEDO21888 REFERENCES FOR SECTION 6.6 6.61 .3. H. Humphrey, Mark I Containment Program, 1/4 Scale Two Dimensional Plant Unique Pool Swell Test Report, General Electric Company, Report No. NEDO-21944, August 1979.

6.62 .3. E. Torbeck, Mark I Containment Program Full Scale Test Program Final J Report, General Electric Company, Report No. NEDO24539, August 1979.

6.63/6.64 Revision 2

NEDO21888 6.7 SAFETY/RELIEF VALVE STEAN CONDENSATION LOADS 6.7.1 TQuencher Discharge Following opening of an S/RV and the expulsion of the air and water initially in the line, a steady flow of steam enters the suppression pool through the TQuencher exit holes. As steam is condensed within the pool, highfrequency, lowmagnitude pressure loads are produced on the torus shell.

Steady state steam discharges through TQuencher devices were performed during the Monticello TQuencher Test, Reference 6.7.11. These tests were performed at two discrete mass fluxes through the quencher holes (150 lbm/secft2 and 22 lbm/secft2). During these tests the maximum steam condensation loading on the torus shell over a local pool temperature range of 50 to 1150F was

~.O.6 psid with frequencies ranging from 75 to 230 Hz. The peak dynamic extreme fiber principle stress determined from strain data measured on the torus shell during steady steam discharge was negligible (less than 0.6 ksi).

The results of additional testing of the steam condensation performance of a cross quencher device are documented in Reference 6.7.12. During these tests the vertical and horizontal spacing, as well as the diameter of the exit holes, were identical to that on the TQuencher. These parameters are of primary importance in the condensation performance of a quencher device.

Sonic steam discharges through cross quencher arms resulted in no discernible changes in steam condensation pressure magnitude over a local pool temperature range of 90 to 2120F.

The low magnitude and high frequencies (higher than those producing the major contribution to torus shell and support response) of these loads, indicate that they are of relatively minor importance in the structural assessment of the Mark I torus and support sytem.

6.7.11/6.7.12 Revision 2

NEDO-21888 REFERENCES FOR SECTION 6.7.1 6.7.11 R. A. Asai, et al., Mark I Containment Program Final Report Monticello TQuencher Test, General Electric Company, Report No. NEDO21864, June 197.

6.7.12 T. Y. Fukushima, et al., Test Results Employed by GE for BWR Containment and Vertical Vent Loads, General Electric Company, Report No. NEDO21078, October 1975.

6.7. 13/6.7. 14 Revision 2

NEDO2 1888 6.7.2 Ramshead Discharge Steady state steam discharges through Ramshead devices were performed during the Monticello Ramshead Test (Ref. 6.7.21). These tests were performed at a 2

mass flux of 200 lbm/secft through the Ramshead exit areas. During these tests the maximum steam condensation loading on the torus shell over a local pool temperature range of 76 to 930F was +/-5.0psid. The frequency of these pressure oscillations ranged between 80 and 125 Hz. The peak dynamic extreme fiber principle stress determined from strain data measured o~i the torus shell during steady steam discharge was small (less than 3.0 ksi).

Additional data (Reference 6.7.22), provides the basis for the establishment of a steam condensation stability limit of 1700F (local poo1 temperature) for 2

Ramsheads when the mass flux exceeds 40 lbm/secft . This data indicates that smooth condensation of steam from a Ramshead discharge can be expected for local pool temperatures below 1600F.

Operation of Mark I plants within the technical specification limits on pool temperature, and prudent operator action following any postulated transients or accidents, will ensure that local temperatures above 1600F concurrent with mass fluxes above 40 lbm/secft 2 will not be reached. Therefore, steam condensation will remain smooth.

6.7.21/6. 7.22 Revision 2

NEDO21888 REFERENCES FOR SECTION 6.7.2 6.7.21 E. A. Buzek, et al., Final Report In Plant Safety/Relief Valve Discharge Load Test Monticello Plant, General Electric Company Report No. NEDO21581, August 1977.

6.7.2-2 J. R. Rollins, Memorandum Report, 1700 Pool Temperature Limit for S/RV Ramshead Condensation Stability, General Electric Company, September 1977.

6.7.23/6.7.24 Revision 2

NEDO21888 EXTERNAL DISTRIBUTION Utility I&SE cc Mr. M. H. OConnor R. F. Thibault Mr. N. Celia (TES)

Boston Edison Co General Electric Co. Mr. W. E. Cooper (TES) 800 Boylston St. 1 Washington St. Mr. C. W. Sullivan (EPRI)

Boston, MA 02199 Wellesley, MA 02181 Mr. P. D. Hedgecock (NUTECH)

Mr. M. G. Zaalouk Mr. C. J. Brossier Mr. H. E. Painter (TIE)

Carolina Power & Light General Electric Co.

P. 0. Box 1551 P. 0. Box 10117 Raleigh, NC 27602 Southport, NC 28461 Mr. R. Mirochna Mr. R. A. Kopriva Mr. H. W. Massie Commonwealth Edison Co General Electric Co. (NUTECH)

P. 0. Box 76734 First 2015 Spring Rd.

National Plaza Oak Brook, IL 60521 Chicago, IL 60690 Mr. D. F. Lehnert Mr. C. M. Johnson Mr. A. Higgenbotham Detroit Edison Co M/C 391 (NUTECH) 318 ECT Project Manager 3331 West Big Beaver Rd. Fermi 2 Project Troy, MI 48084 Mr. J. R. Jordan Mr. J. D. Johnson Mr. Ozem Batum (SCS)

Georgia Power Co General Electric Co. Mr. R. A. Glasby 24th Floor, 3565 Piedmont Rd., N.E. (BECHTELGAITHERSBURG) 230 Peach Tree St. P. 0. Box 4659 Mr. P. Hsu (SCS)

Atlanta, GA 30302 Atlanta, GA 30302 Mr.. S. A. White (SCS)

Mr. W. Rehrauer Mr. D. L. Butcher Mr. J. W. Rasp (NUTECH)

Iowa Electric Light (REPEAT) Mr. T. Shugart (IE)

& Power General Electric Co.

200 First St., S.E. 2015 Spring Road Iowa Electric Building Oak Brook, IL 60521 Cedar Rapids, IA 52401 Mr. R. W. Wulf (2) Mr. G. C. Nelson Mr. R. M. Weiner (MPR)

GPU Nuclear General Electric Co. Mr. J. J. Sturkey (SDT) 100 Interpace Parkway 6 Vreeland Rd.

Parsippany, NJ 07054 Florham Park, NJ 07932 Mr. W. H. Rushron (4) Mr. R. H. Lube Mr. D. Ogden (KAISER)

Nebraska Public Power General Electric Co. Mr. J. W. Angeli (EDS)

District Swanson Building, P. 0. Box 499 Suite 122 Columbus, NE 68601 8401 West Dodge Road Omaha, NE 68114 1 Revision 2

NEDO21888 EXTERNAL DISTRIBUTION (Continued)

Utility I&SE cc Mr. M. G. Mosier Mr. J. M. Morris Mr. N. Celia (TES)

Niagara Mohawk Power General Electric Company Corp P. 0. Box 4841 300 Erie Blvd. West Syracuse, NY 13221 Syracuse, NY 13202 Mr. R. M. Smart Mr. R. F. Thibault Mr. N. Celia (TES)

Northeast Utilities (REPEAT)

Service Co General Electric Co.

107 Shelden St. 1 Washington St.

Berlin, CT 06037 Wellesley, MA 02181 Mr. G. H. Neils Mr. G. H. Scott Mr. T. N. Vogel (NUTECH)

Northern States General Electric Co.

Power Co 5353 Gamble Drive 414 Nicollet Mall Minneapolis, MN 55416 8th Floor Minneapolis, MN 55401 Mr. R. H. Logue (2) Mr. H. T. Ayres Mr. R. Elias Philadelphia Electric Co General Electric Co. (BECHTEL-SAN FRANCISCO) 2301 Market St. 2 Penn Center Plaza Philadelphia, PA 19101 Philadelphia, PA 19102 Mr. John Hoddy Mr. J. M. Morris (REPEAT) Mr. N. Celia (TES)

Power Authority of the General Electric Co.

State of New York P. 0. Box 4841 J. A. Fitzpatrick Nuclear Syracuse, NY 13221 Power Plant P. 0. Box 41 Lycoming, NY 13093 Mr. P. R. H. Landrieu, Mr. J. C. Larrew H/C 391 Mr. H. E. Morris T 17A Project Manager (BECHTEL-SAN FRANCISCO)

Public Services Hope Creek Project Mr. D. A. Gerber Electric & Gas (NUTECH) 80 Park Plaza Mr. C. W. Churchman Newark, NJ 07101 (PSE&G) T2lA Mr. J. C. Carter W9C163 Mr. K. D. Fagan Mr. J. A. Raulston (TVA)

Tennessee Valley General Electric Co.

Authority 832 Georgia Avenue 400 Coimnerce Ave. Chattanooga, TN 37402 Knoxville, TN 37902 2 Revision 2

NEDO21888 EXTERNAL DISTRIBUTION (Continued)

Utility I&SE cc Mr. John Hoffman Mr. R. F. Thibault Mr. N. Celia (TES)

Yankee Atomic Electric (REPEAT)

Co. General Electric Co.

Turnpike Rd., Route 9 1 Washington St.

Westboro, MA 01581 Wellesley, MA 02181 Mr. A. L. Armitage (6)

GETSCO Office Roppogni, Japan Mr. T. Umezu Tokyo Electric Power Co., Inc Suite 720 1901 L Street N.W.

Washington, D.C. 20036 Mr. P. F. Riner (2) Mr. R. A. Canalas GETSCO ZURICH (TRANSMITTAL LETTER 34 Bellerivestrasse ONLY)

CH8034 Zurich Switzerland Mr. Juan Atucha Mr. W. J. Zarella GETSCO SPAIN (TRANSMITTAL LETTER International Division ONLY)

Apartado 700 Madrid, Spain Mr. P. D. Hedgecock (PUT ALL NUTECH MAIL NUTECH TOGETHER IN ONE PACKAGE 145 Martinvale Lane ADDRESSED TO San Jose, CA 95119 MR. P. D. HEDGECOCK) 3 Revision 2

NEDO21888 INTERNAL DISTRIBUTION Name M/C C. W. Cullen (transmittal letter only) 602 F. P. Fel mi 880 D. M. Gluntz 584 G. S. Gupta (3) 126 S. A. Hucik 194 R. M. Hunt (transmittal letter only) 855 R. F. Pratt (2) 094 A. E. Rogers 879 P. P. Stancavage 879 M. E. Tanner 855 D. B. Thomas 849 H. E. Townsend 186 P. Valandani 146 G. E. Wade 154 J. P. Zilinskas 843 Customer Service File 881 NEBG Library (5) 328 VNC Library (2) VOl TIE (5) Bldg 81, Rm A133 SCH TSS (2) 742 4 Reviaion 2

NUCLEAR ENERGY DIVISIONS

• GENERAL ELECTRIC COMPANY SAN JOSE, CALIFORNIA 96125 GENERAL ELECTRIC TECHNICAL INFORMATION EXCHANGE TITLE PAGE AUTHOR SUBJECT TIE NUMBER 8lNED 282 DATE 730 November 1981 TITLE GE CLASS Mark I Containment Program I Load Definition Report GOVERNMENT CLASS REPRODUCIBLE COPY FILED AT TECHNICAL NUMBER OF PAGES SUPPORT SERVICES. R&UO. SAN JOSE.

CALIFORNIA 96125 (Mail Code 211) 440

SUMMARY

This document provides the methodology and definition of the thermalhydraulic loads produced on the pressure suppression containment system o~f GE Mark I containments during a postulated lossofcoolant accident, safety/

relief valve discharges, and related dynamic events.

Information and guidance has been provided to assist in evaluating the design conditions for the various structures of the containment system. This document has been prepared for use by the Mark I Owners in performing plant unique structural evaluations.

This document was prepared by personnel of the Nuclear Power Systems Engineering Department.

By wtting out this rectangle and folding in half, the aboys informadon can be fitad into a staidard card file.

DOCUMENT NUMBER NEDO21888. Rev~ini, 2 INFORMATION PREPARED FOR Nuclear Fuel and Services Division

~CTION I~h,,-1~v Training and Technical Services BUILDING AND ROOM NUMBER 1887 2007 MAIL CODE 886 NEO.914 (6/77)