ML20058C367

TMI-1 Reactor Vessel Pressurized Thermal Shock Evaluation

ML20058C367
Person / Time
Site:	Crane
Issue date:	06/30/1982
From:	GENERAL PUBLIC UTILITIES CORP.
To:
Shared Package
ML20058C349	List: ML20058C345 ML20058C367
References
REF-GTECI-A-49, REF-GTECI-RV, TASK-A-49, TASK-OR 77-1130658, 77-1130658-00, NUDOCS 8207260328
Download: ML20058C367 (158)

Text

-_ -.

i GPU NUCLEAR CORPORATION l

Three Mile Island Unit 1

i Reactor Vessel i

Pressurized Thermal Shock Evaluation l

l l

77-1130658 00 June 1982

[

.aN: c n

'I 77-1130658-00 June 1982 I

GPU Nuclear Corporation THREE MILE ISLAND UNIT 1 REACTOR VESSEL PRESSURIZED THERMAL SHOCK EVALUATION I

I

'I I

I I

I I

'I

I ACKNOWLEDGMENT I

GPU Nuclear gratefully acknowledges the invaluable support of B. J.

Short and J. F. Walters of Babcock & Wilcox and their staffs in the preparation of portions of this report in a very limited time under the changing and evolving directions of the Pressurized Thermal Shock Pro-gram.

I

- 111 -

I l

L

ABSTRACT l

This report provides an evaluation of the TMI-1 reactor vessel integrity for induced pressurized thermal shock.

Fracture mechanics results were determined for a plant-specific SBLOCA and overcooling (failure of all turbine bypass valves) transient. Results indicate that the vessel can experience greater than 32 EFPY with no crack initiation for the as-sumed transients.

Fluid mixing occurring in the vessel cold leg and downcomer was deter-mined using the COMMIX-1A computer code for the SBLOCA transient.

1 i

l l

l

- iv -

)

I I

CONTENTS I

Page 1.

INTRODUCTION AND DISCUSSION.

1-1 I

1.1.

Brittle Fracture Concern 1-1 1.2.

Investigations 1-2 1.2.1.

Transient Analysis 1-4 1.2.2.

Multi-Dimensional Fluid Mixing in RCS Cold Leg and Vessel Downcomer.

1-4 1.2.3.

Vessel Wall Thermal Analysis 1-5 1.2.4.

Vessel Material Properties 1-6 I

1.2.5.

Vessel and Weld Fluence Determination.

1-6 1.2.6.

Vessel Fracture Mechanics Analysis 1-6 2.

SUMMARY

AND CONCLUSIONS.

2-1 2.1.

Overall Vessel Integrity Results 2-1 2.2.

Overcooling Transient 2-3 I

2.3.

SBLOCA Transient 2-3 2.4.

Mixing Analysis.

2-3 2.5.

Vessel Material Properties and Fluence 2-4 2.6.

Fracture Mechanis Analysis 2-4 3.

TRANSIENT ANALYSIS 3-1 I

3.1.

Overcooling Transient Experience 3-1 3.1.1.

TMI June 20, 1974.

3-1 3.1.2.

TMI March 30, 1975 3-1 3.1.3.

TMI April 19, 1978 3-2 I

3.1.4.

TMI April 22, 1978 3-2 3.1.5.

TMI April 23, 1978 3-2 3.1.6.

TMI November 7, 1978 3-2 l

3.1.7.

TMI December 2, 1978 3-3 3.1.8.

TMI March 6, 1979.

3-3 3.2.

Overcooling Analysis 3-3 3.2.1.

Introduction 3-3 I

3.2.2.

Description of Applicable Systems 3-4 3.2.3.

Basis for Transient Selection.

3-6 ig 3.2.4.

Analytical Model Description 3-7 jg 3.2.5.

Analysis Assumptions 3-8 l

3.2.6.

Presentation of Transient Results.

3-13 3.2.7.

Summary and Conclusions.

3-14 r

ll 3.3.

Small Break LOCA Analysis 3-35 l =

3.3.1.

Introduction to SB LOCA Transient 3-35 3.3.2.

Definition and Selection of SB LOCA Transient 3-36

-v-

\\I CONTENTS (Cont'd)

Page 3.3.3.

Major Inputs for Selected Transient 3-37 3.3.4.

Analytical Method.

3-38 3.3.5.

SB LOCA Results.

3-42 4.

FLUID MIXING IN RCS COLD LEG AND VESSEL DOWNCOMER.

4-1 l

4.1.

COMMIX Mixing Analysis 4-2 4.1.1.

The COMMIX Computer Program.

4-2 5

4.1.2.

Modeling Strategy.

4-4 4.1.3.

COMMIX Model Results 4-5 4.1.4.

Qualification of COMMIX Mixing Code.

4-5 4.2.

FLOW-2D, Two-Dimensional Code Simulation 4-6 4.2.1.

FLOW-2D Mixing Model 4-7 4.2.2.

FLOW-2D Mixing Model Results 4-9 4.2.3.

Overcooling Transient Mixing 4-10 5.

VESSEL WALL THERMAL ANALYSIS 5-1 5.1.

Fil.n Heat Transfer Coef ficient and Azimuthal Temperature Distribution 5-1 5.1.1.

Forced and Free Convection HTC 5-1 5.1.2.

Base Metal Azimuthal Temperature Distribution.

5-3 5.1.3.

Vessel Inlet Nozzle Inner Surface Profile.

5-4 5.2.

Wall Thermal Gradient Determination.

5-4 3

5.2.1.

SBLOCA Transient 5-4 3

5.2.2.

Overcooling Transient 5-5 5.3.

Inlet Nozzle Temperature Gradient Determination.

5-5 6.

VESSEL MATERIAL PROPERTIES 6-1 6.1.

Determination of Vessel-Specific Material Properties g

for Input to Linear Elastic Fracture Mechanics l

Analysis 6-1 6.1.1.

Copper and Phosphorus Contents of Welds.

6-1 6.1.2.

Fluence at Weld Locations 6-1 6-2 6.2.

Adjustment of RTNDT.

6.3.

Upper Shelf Requirement for LEFM Analysis 6-2 6.4.

Assumptions and /or Conservatisms in Analysis 6-2 6.5.

Results 6-3 7.

VESSEL AND WELD FLUENCE DETERMINATION.

7-1 7.1.

Analytical Procedure for Fluence Determination 7-1 7.2.

Assumptions 7-2 7.3.

TMI-1 Fluence Results.

7-3 8.

VESSEL FRACTURE MECHANICS ANALYSES 8-1 8.1.

Analytical Model 8-2 8.1.1.

Vessel Beltline Region 8-2 8.1.2.

Vessel Inlet Nozzle.

8-2 8.1.3.

Criteria for Acceptability 8-3

- vi -

)

I CONTENTS (Cont'd)

Page 8.2.

Assumptions and Conservatisms.

8-3 8.3.

Discussion of Fracture Mechanics Results 8-4 8.3.1.

Fracture Mechanics Results - Critical Pressure Vs Time Histories 8-4 9.

REFERENCES 9-1 APPENDIX - CSMP Model for Long-Term Cooling.

A-1 I

List of Tables Table 2-1.

TMI-I Vessel Integrity Results Due to Thermal Shock 2-5 3-1.

Comparison of Turbine Bypass Valve Failure With Other Overcooling Events.

3-15 I

3-2.

Plant Equipment Performance 3-16 3-3.

Operator Actions.

3-17 3-4.

Thermal Shock Overcooling Transient Analysis Sequence of Events 3-18 3-5.

Overcooling Transient Results 3-19 3-6.

SBLOCA Analysis Assumptions 3-46 3-7.

SBLOCA Transient Sequence of Events 3-47 4-1.

COMMIX Input Data for TMI-l SB LOCA Analysis.

4-11 4-2.

Test Conditions for ARC Mixing and Visualization Studies.

4-11 6-1.

Material Properties Used in LEFM Analysis of TMI-l I

Reactor Vessel, 32 EFPY 6-4 6-2.

Material Properties for TMI-l Reactor Vessel Inlet Nozzle 6-5 8-1.

LEFM Results for Thermal Shock Analysis of TMI-l Reactor Vessel.

8-6 I

List of Figures l

Figure 1-1.

Vessel Brittle Fracture Investigation 1-8 I

3-1.

TMI-I RETRAN One-Loop Model 3-20 3-2.

HPI Flow Vs RC Pressure 3-21 3-3.

Feedwater and Steam Flow Vs Time 3-22 3-4.

Hot Leg Pressure Vs Time.

3-23 I

3-5.

RCS Temperatures 3-24 3-6.

System Power Vs Time 3-25 3-7.

RCS Flow Vs Time 3-26

- vii -

I

Figures (Cont'd)

Figure Page 3-8.

OTSG Level Vs Time.

3-27 3

3-9.

Makeup and HPI Flow Vs Time 3-28 3-10.

Pressurizer Level Vs Time 3-29 3-11.

Steam Pressure Vs Time.

3-30 3-12.

P-T Plot TMI-1 TBV Failure Overcooling.

3-31 3-13.

TMI-1 TBV Failure Overcooling Transient 3-32 3-14.

TMI-1 TBV Failure Overcooling Transient 3-33 3-15.

Hot Leg and Head Quality Vs Time.

3-34 3-16.

RCS Flow During Normal System Operation 3-48 3-17.

RCS Flow During Total Loss of Feedwater Event With a Small Break in Pressurizer.

3-49 3-18.

Reactor Vessel Flow During Small Break.

3-50 3-19.

Downcomer Temperature Vs Time, Comparison of Multinode and Eight-Node CRAFT Models, Stuck-Open PORV, Two HPI Pumps, AFW at 40 s.

3-51 3-20.

Downconer Pressure Vs Time, Comparison of Multinode and Eight-Node CRAFT Models, Stuck-Open PORV, Two HPI Pumps, AFW at 40 s.

3-52 3-21.

CRAFT Noding Scheme, Eight-Node Model of RCS.

3-53 3-22.

RV Downcomer Temperature Vs Time, SBLOCA Transient With Operator Throttling of HPI at 3 Hours 3-54 3-23.

RV Pressure Vs Time, SBLOCA Transient With Operator Throttling of HPI at 3 Hours.

3-55 3-24 RCS Loop Flow, Path 3 SBLOCA Transient 3-56 3-25.

Core Outlet Temperature and Subcooling Margin Vs Time SBLOCA Transient With Operator Action at 3 Hours.

3-57 3-26.

Leak Flow Rate Vs Time, SBLOCA Transient With Operator l

Action at 3 Hours 3-58 5

3-27.

HPI Flow Vs Time, SBLOCA Transient With Operator Action at 3 Hours 3-59 3-28.

Pressurizer Level Vs Time, SBLOCA With Break on Top of Pressurizer.

3-60 3-29.

Vent Valve Flow Vs Time, SBLOCA Transient With Operator Action at 3 Hours 3-61 3-30.

Pressurizer Leak and Vent Valve Quality Vs Time 3-62 3-31.

HPI Flow Vs RCS Pressure.

3-63 3-32.

RCS Pressure Vs Temperature for SBLOCA Transient 3-64 g

4-1.

COMMIX-1A Computational Mesh for TMI-1 Cold Leg and g

Vessel Downcomer.

4-12 4-2.

Location of TM1-1 Vessel Welds in 0, Z Coordinates for SBLOCA Transient 4-13 4-3.

Velocity Profile Along J = 6.

4-14 4-4.

Bulk Fluid Temperature From COMMIX Mixing Fbdel for TMI-1 SBLOCA Transient 4-15 4-5.

ARC Mixing Test Section 4-16 4-6.

COMMIX-1A Nodalization Scheme for the B&W Test Facility Mixing Tests 4-17 g

4-7.

COMMIX-1A Temperature Profile - ARC Test Run 2.

4-18 g

4-8.

COMMIX-1A Temperature Profile - ARC Test Run 2.

4-19 4-9.

Finite Difference Mesh With Variable Rectangular Cells 4-20 I

- viii -

I Figures (Cont'd)

Figure Page 4-10.

Location of Variables in Typical Mesh Cell.

4-20 4-11.

FLOW-2D Computational Mesh for Cold Leg Inlet Piping Region.

4-21 I

4-12.

Velocity Field and Interface Between Vent Valve and HPI Streams in Absence of Gravitational Effects and Natural Circulation 4-22 4-13.

Velocity Fields With Gravitational Effects and Turbulence 4-23 4-14.

Temperature Contours Associated With Velocity Field in Figure 4-13.

4-24 4-15.

Temperature Contours Across Downcomer at Lower Lip of I

Inlet Pipe.

4-25 4-16.

Bulk Fluid Temperature From FLOW-2D Mixing Model for SBLOCA Transient 4-26 5-1.

Thermal Gradient in Reactor Vessel During Normal and Abnormal Conditions 5-7 5-2.

Wall Convective Heat Transfer Coefficient Vs Transient Time for SBLOCA Transient 5-8 5-3.

Wall Convective Heat Transfer Coefficient Vs Time for overcooling Transient 5-9 5-4.

Base Metal Temperature Vs Time for SBLOCA Transient I

at Location of WF-70 Weld 5-10 5-5.

Base Metal Temperature Profile Vs Time for Overcooling Transient 5-11 5-6.

Vessel Inlet Nozzle Base Metal Temperature Vs Time I

for SBLOCA Transient at Knee of Nozzle 5-12 5-7.

Inside Surf ace of Reactor Vessel Weld Locations, TMI-1.

5-13 5-8.

SBLOCA Transient - Wall Temperature Profiles for Location of WF-70 Weld Using COMMIX.

5-14 5-9.

Wall Temperature Profile for Overcooling Transient 5-15 5-10.

Vessel Inlet Nozzle Finite Element Model.

5-16 I

5-11.

SBLOCA Transient - Inlet Nozzle Temperature Profiles at Plane of Nozzle Knee 5-17 5-12.

Overcooling Transient - Vessel Inlet Nozzle Base Metal Temperature Profile Vs Time at Knee of Nozzle 5-18 I

7-1.

Fast Flux Attenuation Through Pressure Vessel Wall.

7-4 7-2 Typical Azimuthal Flux Profile Normalized to Peak Flux Location.

7-5 I

7-3.

Longitudinal Weld Locations to Azimuthal Fluence Profile.

7-6 7-4.

Reductions in Peak Reactor Vessel Fluence in TMI-1.

7-7 8-1.

Flaw Geometry Used in Vessel Thermal Shock Analysis 8-7 I

8-2.

Pressure History for SBLOCA Transient, WF-70 Weld at 32 EFPY 8-8 A-1.

Schematic Model Diagram of Reactor Coolant System and Steam Generation for Long-Term Transient Calculation.

A-5 I

e K -

'll

1.

INTRODUCTION AND DISCUSSION The investigations described in this report were performed to evaluate the concern of brittle fracture in the Three Mile Island Unit 1 (TMI-1) reactor pressure vessel as a result of severe cooldown and repressurization transients.

These transients can be the result of coolant discharge from the primary sys-tem or various events that cause the steam generators to remove more energy than is being generated in the primary system.

1.1.

Brittle Fracture Concern The reactor vessel integrity concern as a result of design basis events -

loss-of-coolant accident (LOCA), main steamline break (MSLB) - has been ad-dressed previously during the original licensing process. During the past two years, as a' result of the TMI-2 accident, concerns have been expressed over reactor integrity during moderate frequence overcooling and small break LOCA (SBLOCA), high-pressure injection events. Conservative analyses were begun and completed for the B&W Owners Group in 1980 for the SBLOCA event.1 This generic, conservative analysis showed some vessels to be acceptable for only a few additional years.

In March 1981, the first of several industry /NRC meetings was held to discuss the issue of reactor vessel (RV) integrity fol-lowing postulated overcooling transients, as well as SBLOCA thermal shock.

I This initial meeting was followed by meetings between industry and Staff in April, July, and September. Utility submittals were made to the NRC in May and October, the latter of which was in response to an August 21, 1981 request by the Staff for information. During 1981, B&W began, at the request of the B&W Owners Group, a detailed investigation into the various factors affecting brittle fracture cracking of reactor pressure vessels. The first plant-spe-cific analysis completed was submitted by Duke Power Company to the USNRC on January 15, 1982.

The results of the analysis for GPU Nuclear's (GPUN) TMI-1 plant are provided in this report.

I

]

l 1-1

\\

l

The brittle fracture concern arises from the postulated assumption that a

" crack" (i.e., flaw) of critical size exists in a vessel that might undergo a severe overcooling transient. As a result of this transient, a large ther-mal gradient may be obtained across the pressure vessel wall.

In addition, if this vessel has undergone significant irradiation by fast neutrons, the vessel toughness has decreased and it could be susceptible to brittle frac-ture cracking. Also, in the case of non-break overcooling transients the emergency core ccoling system (ECCS) high-pressure injection (HPI) system is activated.

This system adds cold water to the reactor coolant system (RCS) to maintain inventory as a result of the oveccooling transient and, if not terminated by operator action, the HPI flow could repressurize the RCS.

The brittle fracture concern then arises from a vessel that has a critical sized flaw, a significant period of irradiation, a severe overcooling transient, several hardware or control system failures, and no operator action to throt-tle HPI flow to prevent repressurization of the RCS.

The extent of the brittle fracture concern applies to all RVs but is most severe for vessels having significant irradiation and welds with a high cop-per/ phosphorus content, which are subject to greater radiation damage.

1.2.

Investigations Various engineering disciplines and computer codes and programs are necessary to place the brittle fracture concern in perspective.

Results of independent programs have contributed to the evaluations herein and have resulted in significant reductions in the number of overcooling and SBLOCA events, as well as a better understanding of system response once an event is postulated to occur. These programs include the following:

1.

Improvements in the primary and secondary system instrumentation and controls.

2.

Improved reliability of emergency feedwater (EFW) initiation and control.

3.

Improved reliability of controls, indication, and operation of the power-operated relief valve (PORV) as well as its block valve.

4.

Improved understanding of system response through improved analyti-cal codes.

5.

Improved understanding of the RV materials properties through the ongoing (integrated materials) surveillance program.

1-2

6.

Enhanced examination techniques associated with beltline region welds, which are most susceptible to irradiation embrittlement.

7.

Operator guidance for minimizing thermal shock to the vessel.

The integrated reactor vessel materials surveillance (RVMS) program was imple-mented in 1977.

It included irradiation of specifically designed test and specimen capsules from operational vessels. Programs were also undertaken to improve understanding of vessel weld chemistry, strength properties, fluence levels, crack initiation, growth, and arrest. Also addressed are shif t of RT and decrease in upper shelf energy. This multiyear program will con-g tinue into 1995.

The investigation of the brittle fracture concern is divided into six techni-cal areas:

1.

Transient analysis.

2.

Fluid mixing in RCS cold leg and vessel downcomer.

3.

Vessel wall thermal analysis.

4.

Vessel material properties.

I 5.

Vessel and weld fluence determination.

6.

Vessel fracture mechanics analysis.

These areas, as used in the vessel brittle fracture investigation for the B&W Owners Group, are explained further in the following sections. The flow of I

information between the various engineering disciplines is illustrated in Fig-ure 1-1.

The final results of the investigation determine when the vessel's toughness has degraded to a point where a flaw could initiate and arrest with-ing 1/4T (per the acceptance criteria of section 8) as a result of the severe overcooling transient. The results are reported in section 2 in terms of ef-fective full power years (EFPY).

The transients analyzed were chosen to be realistic and probable transients.

The transients were thought to be of sufficient probability to warrant an analysis on their results. There are multiple plant failures assumed in these transients other than the mechanistic failure that initiates the transients.

This report utilizes a TMI-1 plant-specific SBLOCA analysis that has been de-termined to be realistic, yet conservative. The report also addresses an over-i cooling transient that repressurized.

This transient is the result of overcool-us ing the primary system as a result of some fault on the secondary side of the plant. The overcooling transient analysis was performed by GPU Nuclear (GPUN) 1-3

based specifically on the design of the TM1-1 plant.

The RV analysis is based on plant-specific design features of the TMI-l vessel, supplemented as necessary with data obtained through the ongoing B&W Owners Group Integrated RVMS Program.

This report provides an evaluation of the potential for RV failure due to severe overcooling transients and SBLOCAs (extended HPI) for the TMI-1 reactor vessel.

1.2.1.

Transient Analysis 1.2.1.1.

Overcooling This transient evolves from a secondary side upset (e.g., a small steamline break) and causes excessive heat removal from the primary side. This over-cooling causes shrinkage in the RCS inventory (pressure reduction) and large thermal gradients in the vessel wall.

S ince the pressure decrease causes initiation of the HPI system, the RCS will be repressurized if operator action is not taken.

Further discussion is provided in section 3.2.

1.2.1.2.

SBLOCA Consideration of.varic.,us requirements for system response and its ef fect on thermal shock, as outlined in section 3.3, resulted in the selection of a transient that unfolds as a result of a loss of all feedwater with the pres-surizer safety valve f ailing open. This results in the core being cooled in the once-through HPI cooling mode of operation.

The cold HPI water is in-jected in the cold leg, enters the downcomer, and could result in large ther-mal gradients in the vessel wall.

Further discussion of the SBLOCA analysis is provided in section 3.3.

W 1.2.2.

Multi-Dimensional Fluid Mixing in RCS Cold Leg and Vessel Downcomer The mixing of the cold HPI fluid and the hot vent valve fluid in the upper vessel and cold leg regions has been investigated to determine to what extent the cold HPI fluid is heated before it reaches the reactor vessel. Two inde-pendent approaches to the mixing aspect of this problem have been taken with the expectation that each will conclude similar results, thus supporting the final result that significant mixing of the cold HPI fluid occurs before it reaches the vessel downcomer.

I 1-4

I I

1.2.2.1.

Three-Dimensional Mixing Analysis 2

Using the COMMIX Code,a The COMMIX code was uses to assess the fluid temperatures and velocities exist-I ing in the TMI-l plant geometry consisting of a 90-degree section of the ves-sel with cold leg piping, internal vent valves, and vessel downcomer. COMMIX, developed by Argonne National Laboratory and modified by B&W for the thermal shock work, is a three-dimensional computer code which solves the continuity, momentum, and energy equations. The code predictions have been benchmarked aga inst the results of a series of fluid mixing tests at B&W's Alliance Re-search Center (ARC). COMMIX-1A produced results that agree reasonably well with the ARC data and flow visualization studies.

See sections 2.4 and 4.1 for further discussion.

1.2.2.2.

Two-Dimensional Mixing Analysis Using the FLOW-2D Code I

A two-dimensional fluid mechanics digital computer code (FLOW-2D, a deriva-tive of the SOLA-V0F code) was used to model the B&W nuclear steam system I

(NSS) cold leg, upper vessel, vent valves, and downcomer regions so the re-sulting fluid mixing in this region could be analyzed. Two-dimensional stud-ies were completed to evaluate the important effects of the physical parameters such as buoyancy and turbulent mixing. These studies defined the effects of these variables on overall mixing of the cold HPI flow and warm vent valve flow.

These FLOW-2D results have been used in the evaluation of the vessel inlet nozzle in this report. The FLOW-2D model has been benchmarked to the test data obtained from the one-fifth scale model CREARE test facility. The data were obtained as part of an EPRI-sponsored research program in the sum-I mer of 1981.

See section 4.2 for further discussion.

1.2.3.

Vessel Wall Thermal Analysi_s The wall film heat transfer coefficient (HTC) is determined as the maximum of either the pure forced or free convection HTC (see section 5.1). Once the film HTC and wall surface temperature are determined, the temperature gradient in the vessel material is determined in the r and 0 coordinates at selected I

axial weld locations as a function of time for the SBLOCA transient. This two-dimensional vessel wall temperature gradient is calculated using the ANSYS finite element code.5 For the overcooling transient, a two-dimensional ver.sel 1-5 I

temperature gradient is determined using a two-dimensional axisymmetric finite element model and the ANSYS code (see section 5.2).

1.2.4.

Vessel Material Properties The material properties used as input to the LEFM analysis for this thermal shock analysis are based on the material certification and weld metal qualifi-cation records where available. These properties and data are available in BAW-1439.6 For cases where the properties are not available, they were es-tablished in accordance with BAW-10046A, Rev. 1.7 These results and the re-sults of the integrated RVMS program indicate that the materials are behaving in a predictable manner. This surveillance program will allow continued eval-uation and predictions for the TMI-1 materials as they are affected by their service environment.

A detailed study was conducted in 1978 to ascertain the representative chemi-cal composition of all vessel beltline region weld metals for B&W 177-fuel assembly (FA) vessels.' This chemistry, as well as the copper and phosphorus contents, was determined and reported in BAW-1511P.' Section 6 includes a discussion of the results.

1.2.5.

Vessel and Weld Fluence Determination Vessel and weld fluence were determined for each weld in the vessel beltline region. The fluence was determined from data for the TMI-1E capsule presented 6

in BAW-1439 using analytical procedures described in BAW-1485." Section 7 includes a discussion of the results, and fluence information is provided in Table 6-1.

1.2.6.

Vessel Fracture Mechanics Analysis The vessel stress analysis uses linear elastic fracture mechanics (LEFM) eval-uations for analysis of the beltline region of the vessel.Section XI, Appen-

=

dix A, of the ASME Code is used as the basis for all calculations. The anal-ysis evaluates flaw sizes from 1/40T to 30/40T using an aspect ratio of 6:1 with an adjusted RT fr m Regulatory Guide 1.99, Rev. 1, and the ASME Code NDT equation for K using an upper limit of 200 kni /in. The cladding is consid-IC ered when determining the thermal gradients in the wall but is neglected in the critical pressure determination.

I 1-6

i The vessel inlet nozzle was analyzed to verify integrity since the cold HPI water injected into the cold leg during these transients could result in cool-1 I

down rates greater than the design rates. Since the fluence is low (on the

)

16 2

order of 10 n/cm at 32 EFPY) the resultant stress is due primarily to the thermal gradient in the nozzle as a result of the plant transient. See sec-tion 8 for further discussion.

I I

I I

I I

I I

I I

I I

I 1-7 I

Figure 1-1.

Vessel Brittle Fracture Investigation TMI-1 PLANT-SPECIFIC RCS SBLOCA PRESSURE TRANSIENT HISTORY COMPARE ALLOWABLE ANALYSIS PRESSURE Vs ACTUAL Section 3 l

^

TMI-1 OVERC00 LING RCS FLUID VESSEL WALL MECHANICS REPRESSURIZATION PRESSURE /

MIXING THERMAL VESSEL ANALYSIS

--*- L I F E TEMPERATURE ANALYSIS ANALYSIS i

PLANT-SPECIFIC

+

( Allow. Press.)

(EFPY)

TRANSIENT ANALYSIS HISTORY Section 3 Section 8 7

d cn VESSEL MATERIAL l

PROPERTIES Section 6 VESSEL AND WELD FLUENCE DETERMINATION Section 7 i

i O

O

2.

SUMMARY

AND CONCLUSIONS The following is a summary of the results and conclusions for the TMI-1 PTS evaluation:

2.1.

Overall Vessel Integrity Results

. Vessel interrity is maintained with no defect growth for the two PTS scenarios analyzed, as shown in Table 2-1.

Assumed flaws up to 1/4T (i.e., 1/4 distance through vessel wall) at the most critical areas in the vessel will remain stable during a SBLOCA or an overcooling event n

(stuck-open TBV) at 32 EFPY and longer.

. The criteria used in selecting the two overcooling scenarios used in I

this report were that the events had to be:

- Mechanistic, i.e.,

the transient had to be physically possible from the standpoint of plant-specific equipment and systems.

- Scmewhat probable, i.e.,

the transient had to have some relative possibility of occurring. The multiple failures chosen were based on careful consideration of the operating history of the plant and the industry in general and on the various classes of events exam-ined in the PRA analysis in the Oconee report.

- A PTS sequence of events, i.e.,

the scenario had to go in a direc-tion where there was overcooling with repressurization.

The results of the generic and very conservative PTS evaluations --

BA1-1628 and BAW-1648 - showed approxiamtely 4.8 EFPY, whereas the TM2-1 results indicate no defect growth even at 40 EFPY. The reasons for these differences are accounted for in the various bounding inputs in the two generic reports that include HPI flows, pressurizer safety and vent valve capacities, conservative mixing assumptions, weld loca-tions, fluence levels, chemistry, etc. versus the plant-specific data used in this report. Thus, the results dictate the need to input and account for plant-specific designs.

The use of a realistic mixing model (COMMlX-1A) coupled with the ac-tual plant design (e.g., vent valves, HPI configuration) yields 2-1

significantly higher downcomer temperatures for SBLOCA than that shown in the Duke Power Company submittal on Oconee 1 (DPC-RS-1001) and assumed in BAW-1648 and BAW-1628. The higher camperatures reduce the thermal stress experienced at the inside RV wall.

Several significant conservatisms still remain in the evaluation, including but not limited to the following:

- 32 EFPY fluence values used in the analysis are based on the current 12-month fuel cycle. A reduction of approximately 30% in fluence will occur with the implementation of a low-leakage 18-month fuel scheme currently under study for TMI-1.

For example, if the 18-month cycle is implemented, the corresponding reduction in the RTndt for the WF-8 weld is from 335 to 305F at 32 EFPY.

- The transient analysis assumed minimal mitigating operator action.

- The fracture mechanics analysis used a conservative 1/4T crack ar-rest criterion. Moreover, the evaluation resulted in no defect g

growth assuming the full range of initial crack sizes from 1/40T to g

1/4T.

- No credit was taken for improvements in ISI and NDE techniques that result in increased detectability of small defects in the RV.

- No credit (where applicable) was taken for the experimentally veri-fled phenomenon of warm prestress.

- The mechanical effects of cladding were excluded from the stress analysis.

- The evaluation used the techniques of linear elastic fracture me-chanics (LEFM) instead of clastic-plastic fracture mechanics (EPFM).

- The warmest borated water storage tank (BWST) temperature used was SOF, whereas the actual historical BWST temperatures experienced at l

i l

TMI are in the range of 80-100F.

5 l

l l

- The B&W Owners Group RV Integrity Program is and will continue to g

l monitor relevant material properties presently planned out to 20 g

EFPY.

I l

l l

l 1

f 2-2

2.2.

Overcooling Transient The overcooling or secondary side irduced transient selected by GPUN for the TMI-1 overcooling transient analysin is the failure of all turbine bypass valves (TBVs) as a result of a sing: e failure in the plant integrated control system (ICS). This transient simulotes a secondary break equivalea to about 27% of total steam flow.

Operator action was assumed to isolste HPI flow and the TBVs at 12. minutes and prevent additional plant heatup at I hour into the transient.

The pres-sure and temperature versus time for this TMI-1 plant-specific overcooling I

transient are shown in Figures 3-4 and 3-5 of section 3.2.

2.3.

SBLOCA Transient The TMI-1 plant-specific SBLOCA tratsient chosen, using the selection crite-ria discussed in section 3.3.2, resalts in a realistic challenge to the resc-tor vessel from the thermal shock aapect. The resultant transient is a pressurizer safety valve being challenged due to loss of feedwater, falling B

open, and depressurizing/ overcooling the system.

Operator action to throttle the HPI flow was assumed to maintain the core out-I let temperature at less than 100F subcooled. However, in the analysis this action was not required until 180 minutes into the transient. The temperature and pressuu versus time for the TMI-1 plant-specific SBLOCA transient are shown in Figures 3-22 and 3-23.

2.4.

Mixing Analysis Evaluations of mixing of cold HPI water and hot vent valve water were per-l formed using various mixing codes and theories during the course of the B&W Owners Group and TMI-1 plant-specific analysis.

I Previous evaluations performed using the turbulent jet theory and two-di.nen-aional digital code turbulent fluid simulation (FLOW-2D) show that signifi-I cant mixing occurs. These results have been compared with test data obtained from a one-fifth scale model CREARE test facility.

The results are favorable in all cases. The jet theory prediction was found to be conservative com-pared to the CREAkE test data.

The FLOW-2D code analysis results show that mixing occurs in the cold leg and upper vessel, thereby preventing cold HPI water from entering the down<omer sf the vessel as seen in Figures 4-15 and

~

I 4-16.

Benchmarking of the FLOW-2D two-dimensional code analysis to the CREARE

)

test data has been completed with favorable results.

The mixing analysis used in this report was the three-dimensional COMMIX-1A code with input from a TMI-1 plant-specific SBLOCA analysis to determine the extent of mixing in the vessel cold leg and downcomer. The results agree with the previous evaluations that significant mixing of the cold HPI water does occur in the cold leg piping before the fluid reaches the vessel downcomer.

COMMIX-1A was benchmarked against data obtained from a test facility at B&W's Alliance Research Center (ARC) during an internal research and development program.2,3 1.5.

Vessel Material Properties and Fluence Plant-specific material properties and chemistry were determined and used in this thermal shock analysis. These properties are presented in Tables 6-1 and 6-2.

The fluence at the various weld locations and base metal was deter-mined using established procedures and chemistry as shown in the tables.

TMI-1 has completed four fuel cycles while accumulating 3.52 EFPY and a flu-18 2

ence of 1.80 x 10 n/cm at the peak location. Using the present fuel man-g agement scheme (no L3P reloads), the predicted peak fluence at 32 EFPY is W

2 1.7 x 10" n/cm,

2.6.

Fracture Mechanics Analysis LEFM analytical techniques based on Appendix A of the ASME Code,Section XI, were used to evaluate RV integrity for both SBLOCA and overcooling transients.

WPS techniques were not used in this evaluation of the TMI-1 reactor vessel integrity. As shown in Table 2-1, all have an acceptable lifetime of more than 32 EFPY with Section XI techniques.

In addition, no crack initiation was experienced during this analysis for either the SBLOCA or overcooling transients analyzed in this report.

The LEFM analysis used an acceptance criterion to judge the acceptability of the results whereby if a postulated flaw exists, it will be arrested within 1/4T. No vessel failures occurred for the postulated transients at the ves-sel lifetimes shown in Table 2-1.

I I

2-4 t

I

Continued surveillance of the reactor vessel materials, as required by 10 CFR 50, Appendixes G and H, will ensure that the assumptions used in this analysts remain valid. This need is fulfilled by the B&W Owners Group Inte-I grated Reactor Vessel Surveillance Program.

I Table 2-1.

TMI-1 Vessel Integrity Results Due to Thermal Shock (Ef fective Full-Power Years)(a)

Transient Overcooling (C}

Weld ident or heat ( }

SBLOCA I}

ARY-059 40 32 WF-70 40 32 WF-8 40(C) 32 C2789-1 40(c) 32 WF-25 40 32 C3251-1 40(")

32 I

SA-1526 40(')

32 Atypical 40 32 I

40(C) 3'2 Vessel inlet nozzle I

("} Criterion for acceptability of results:

LEFM analysis crack arrest within 1/4T.

Results also show no crack initiation for either I

transient.

(

Refer to Tables 6-1 and 6-2 for locations.

('} Predicted based on conservative mixing anal-ysis.

I I

I I

I 2-5 I

I I

I 3.

TRANSIENT ANALYSIS I

3.1.

Overcooling Transient Expe rience A review of the operating history of Three Mile Island Units 1 and 2 identified several events that exceeded the cooldown rate limit of 100*F/ hour or that had the potential to exceed the cooldown rate limit if not mitigated by automatic plant response or operator action. Each of these events is summarized below.

3.1.1.

TMI June 20, 1974 During the startup test program the turbine overspeed trip test was in progress.

I A malfunction of the "B" turbine bypass valves caused the valves to open fully and, when resetting the turbine, only the "B" stop valves opened. This caused the "B" turbine header steam pressure to drop and actuated the "B" steam line rupture detection system. This isolated feed to the "B" steam generator, which boiled dry in 3 minutes. An operator bypassed the rupture system, which re-stored feedwater to the "B" steam generator. The sudden restoration of feed-water caused a reactor trip on low pressure. The reactor coolant temperature dropped to 525F within approximately 15 minutes and stabilized. Reactor pres-sure reached a minimum value of 1650 psig and recovered within 30 minutes.

3.1.2.

TMI-l - March 30, 1975 During full-power operation a turbine trip relay failed, causing the turbine to trip. Within 4 seconds the reactor tripped on high pressure, and the plant cooled down to a low temperature of 532F on the "B" inlet loop within 1.5 min-I utes after the trip due to several factors. The "B" main feed pump was in manual and caused an overfeed for several minutes until operators intervened.

Also, two main steam safety valves f ailed to reseat properly af ter lif ting.

The temperature was restored to normal within 2 minutes af ter the trip: however, the rapid cooling caused a low pressurizer level and the operators started a second makeup pump and opened an additional high-pressure injection valve to restore level. Pressurizer level and pressure were restored to normal post-trip values within 5 minutes af ter the trip.

3-1 I

I 3.1.3.

TMI April 19, 1978 The plant was in a startup mode at 15% power, when a condensate system upset caused a reduction in steam generator levels and a subsequent reactor trip on high pressure. Af ter the reactor trip, one main steam safety valve blew down to 600 psig before resetting, decreasing the RCS temperature to 525F 12 minutes after the trip and pressure to 1730 psig. Pressure and temperature were re-stored within 15 minutes af ter the trip.

3.1.4.

TMI April 22, 1978 During a loss-of-offsite power test from 15% power, the emergency feedwater system was actuated within one minute of the reactor trip and fed the steam generators at full flow for 12 minutes in an attempt to reach the design 50%

level setpoint.

Flow was terminated at that time at 40% level due to excessive cooldown. The temperature reached 500F 14 minutes af ter trip and 495F 35 min-utes after the trip.

Pressure reached a minimum of 1970 psig. The overcooling was attributed to low decay heat levels and an excessive feed rate.

I 3.1.5.

TMI April 23, 1978 During the startup test program, the reactor tripped from 30% power due to a spurious noise spike on one nuclear instrument channel in the reactor protec-tion system. During the transient several steam safety valves lif ted and failed to rescat until steam generator pressure was below 600 psig. This, plus overfeeding conditions in both steam generators, caused the RCS to cool down to 450F within 4 minutes after the reactor trip, the RCS pressure to reach a minimum of 760 psig, and pressurizer level to go off-scale low for 1.5 minutes. The transient was terminated by operator action to reduce feed, isolation of the steam generators by the rupture detection system, and auto-matic actuation of the high-pressure injection system. RCS pressure recovered within 12 minutes of reactor trip.

This cooldown exceeded the 100 F/ hour limit. The plant was shut down to repair the main steam safety valves.

3.1.6.

TMI November 7, 1978 During a test to determine the reactor moderator temperature coef ficient, a

feed pump trip occurred and initiated a runback f rom full power. Due to an elevated average temperature because of the test, the reactor tripped on vari-able pressure-temperature limits. The initial response to the trip was normal; 3-2

I however, one turbine bypass valve remained open and reduced the pressure in the "A" steam generator to 550 psig in 12 minutes. At this time, operators isolated the leaking valve and terminated the cooldown. RCS temperature was I

reduced to 525F 10 minutes after the trip and, due to this cooldown, RCS pres-sure dropped to 1595 psig, which caused high-pressure injection initiation.

Pressure was returned to normal within 20 minutes after the trip.

3.1.7.

TM1 December 2, 1978 The plant was in a startup mode at 22% power, shif ting f rom the startup feed valves to the main feed regulating valves. The startup valves reached 80%

open, and the main feedwater block valves opened. The operator noticed de-creasing dif ferential pressure across the valves and increased the main feed pump speed.

It was later determined that the main feedwater regulating valves were blocked in the full-open position. The sudden overfeed caused a rapid cooldown, which resulted in a reactor trip on low pressure and a high-pressure injection system actuation. RCS temperature cooled down to 520F, and RCS pres-sure reached a low of 1670 psig approximately 3 minutes af ter the trip.

Pres-I sure was restored to normal within 20 minutes.

I 3.1.8.

TMI March 6, 1979 During full-power operation, the turbine tripped due to a spurious overspeed signal. A runback to 15% power commenced, but the reactor tripped because the core power imbalance limit was exceeded.

Reactor pressure decreased to 1720 psig, and reactor temperature decreased to 520F within 10 minutes due to mal-I functioning turbine bypass valves and to overfeed of the steam generators by t

manually controlling the feed system after the trip.

Temperature and steam generator levels stabilized, and reactor pressure recovered within 15 minutes.

1 3.2.

Overcooling Analysis i

3.2.1.

Introduction This section documents the TMI-l overcooling transient analysis performed by GPN to provide plant-specific input for the TMI-I reactor vessel pressurized thermal shock evaluation.

I I

The TMI-l plant has specific design features, particularly in the emergency feedwater (EW) system, which warrant that a plant-specific analysis be per-I l

formed in order for the transient analysis results to be truly representative I

3-3 I

I of TMI-l plant behavior. These features include the deletion of the steam line rupture detection system (SLRDS) for EW, the inclusion oc a flow-limiting device (cavitating venturis) downstream of the OTSG EW flow control valve in each line, and the specific flow associated with the spurious opening of all six turbine bypass valves at TMI-1.

Section 3.2.2 provides summary informa-g tion for each of the plant systems associated with this analysis.

E A vast number of possible overcooling scenario event sequences can be postu-lated to occur. A logical approach to determining which overcooling transient should be analyzed in detail is to choose an event sequence that provides an appropriate balance between likelihood of occurrence and severity of cooldown.

5 The overcooling transient analyzed herein was formulated on this basis. Th is is discussed in section 3.2.3.

The transient selected is the spurious opening of all six turbine bypass valves.

11 12 It was analyzed with the RETRAN code for the first 12 minutes and with a CSMP analysis out to 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />. The analytical methodology is discussed in section 3.2.4 and in Appendix A.

Specific assumptions regarding modeling, plant equipment, and operator inter-vention are provided in section 3.2.5.

The detailed transient results, in-cluding time plots of pertinent system parameters and data tables, is presented in section 3.2.6.

3.2.2.

Description of Applicable Systems 3.2.2.1.

Emergency Feedwater The EW system consists of two motor-driven pumps and one turbine-driven pump.

Each motor-driven pump has approximately half the capacity of the steam-driven 5

pump.

All three pumps are interconnected at their discharges so that flow from one pump is available to both once-through steam generators (OTSGs).

Each OTSG has a flow control valve in its associated piping. Downstream from this valve in each line is a cavitating venturi, which limits the maximum flow to the steam generator.

Total flow to both OTSGs is limited to 1150 gpm at 600 psig or less with all three pumps running.

The EW control valve is controlled by the integrated control system (ICS) to maintain level at 30 inches (36 inches from the lower tubesheet) with RC pumps running.

Level is controlled at 50% on the operate range (248 inches from the lower tubesheet) when RC pumps are of f in order to promote natural circulation.

3-4 l

I The original plant design included automatic isolation of EW when OTSG pres-sure dropped below 600 psig. This signal (steam line rupture detection system, SLRDS) closes both the control and block valves in the main feedwater system.

liowever, the SLRDS will be deleted f rom EW in the near future. These analyses presumed that the SLRDS was deleted from the EW system.

I 3.2.2.2.

High-Pressure Injection The high-pressure injection system (HPI) serves a combined function, providing both normal RCS makeup and emergency core cooling. The system includes three pumps, one of which is aligned in the HPI mode (i.e., suction from the BWST)

I and one in the makeup mode (suction from the makeup tanks). The third pump serves as a backup makeup pump, but is aligned as an HPI pump upon initiation of the emergency safeguards actuation system (ESAS).

Normal makeup is provided through a flow control valve, which maintains pres-surizer level at a setpoint determined by the operator (normally 220 inches).

This path is isolated upon ESAS initiation.

ESAS initiation occurs upon 1600 psig pressure. Two HPI pumps are init ia ted ;

both inject into all four cold leg injection lines through cross-connected p ip ing. Each line contains a cavitating venturi to limit and balance flow under both normal and accident conditions without operator intervention.

3.2.2.3.

Pressurizer Heaters Five banks of heaters are provided in the pressurizer. Three banks have ON/0FF operation, and two operate proportionally within their control bands. The ca-pacity of all five banks of heaters is 1638 kW.

3.2.2.4.

Main Feedwater System i

The main feedwater (MFW) system consists of two trains, which are connected W

upstream of the feedwater control station for each steam generator. The feed-l water contains two turbine-driven feedwater pumps and three condensate and con-densate booster pumps.

Each feedwater station consists of an MFW control valve, a startup (bypass) control valve for low flow operation, and main and startup block valves. All four of these valves receive a closure signal on low steam generator pressure of 600 psig.

These valves also receive control signals from the ICS whenever MW pumps are running. Upon loss of feedwater, ICS control is switched to the EW system control valves.

E 3-5 I

I 3.2.2.5.

Turbine Bypass System / Atmosphere Dump Turbine bypass is provided by a total of six relief valves which relieve g

steam to the condenser. Their total capacity is 22.5% of normal steam flow.

5 The bypass valve setpoint varies with plant condition (power operation, tur-bine trip without reactor trip, and turbine trip with reactor trip).

Two air-operated, atmospheric dump valves relieve steam to the atmosphere im-mediately af ter a trip and/or during the unavailability of the turbine bypass system. Their total capacity is about 6.5% of total steam flow.

3.2.2.6.

Core Flood Tanks The core flood tanks (CFTs) are two passive injection tanks filled with bo-rated water and a nitrogen cover gas at 600 psig. When RCS pressure goes below 600 psig, the CFT check valves open and inject water into the RCS.

The capacity of the tanks is large enough to keep them from emptying during a non-LOCA overcooling event. Thus, the nitrogen cover gas keeps the RCS pres-sure from going below 600 psig.

3.2.2.7.

Reactor Protection System The reactor protection system (RPS) has two-out-of-four initiation logic which trips the reactor to protect the core from DNB violations and the RCS from over-pressurization.

A region of acceptable operation is defined by high RCS pres-sure, high core outlet temperature, high neutron flux, low RCS pressure, vari-able low pressure (dependent on hot leg temperature), and the RC pump / power ra t io.

3.2.2.8.

Reactor Coolant Pumps Each of the four cold legs of the RCS contains an RC pump; the pumps operate at a constant volumetric flow. All four pumps are powered f rom the of fsite grid, and loss of of f site power results in loss of the pumps. Auxiliary ser-vices required for pump operation are motor cooling and seal cooling; the lat-ter is supplied either from the makeup /HPI system or by thermal barrier cooling.

3.2.3.

Basis for Transient Selection Overcooling of the plant occurs during events that result in increased steam flow with subsequent secondary side depressurization, reactor trip, and con-tinued feedwater injection.

I 3-6

i Events of this type that result from pipe breaks are less likely to occur (and are more quickly diagnosed) than those that are caused by equipment malfunc-tions.

In this regard, a single failure in the integrated pressure control I

portion of the ICS results in a full-open signal to the turbine bypass valves.

The failure is in the turbine header pressure setpoint generator, which causes a setpoint signal of 600 psig; the normal setpoint is 885 psig. Thus, all six turbine bypass valves will open. The severity of a cooldown transient is judged on its initial cooldown rate, minimum temperature, and the presence of eventual repressurization of the RCS if a crack is propagated by the earlier cooling.

Events with a fast cooldown rate, low minimum temperature, and sub-sequent repressurization are considered most challenging to the integrity of the reactor vessel. Table 3-1 compares the turbine bypass failure event with I

other potential cooldown events. This table, based on previous analyses, en-gineering judgment, and experience, shows that failures that have a more severe overcooling potential are less likely to occur. Similarly, the more likely failures have a smaller potential for overcooling.

Hence, the turbine bypass valve failure was chosen as the overcooling transient to be evaluated for re-actor vessel thermal shock.

3.2.4.

Analytical Model Description The turbine bypass valve failure overcooling transient was analyzed using the ll 12 system analysis code RETRAN and the dynamic simulation code CSMP (Continuous System Modeling Program).

For the first 12 minutes, during which a best-esti-mate of primary and secondary parameters is necessary, the one-dimensional multi-node RETRAN program provides the level of detail required to model the event. The overcooling is essentially terminated by isolation of the steam I

leak and termination of the HPI; therefore, the long-term system response is an energy balance calculation, which can be performed efficiently, by the dynamic simulation language CSMP.

Beth models are discussed below.

3.2.4.1.

The TMI-1 RETRAN Model l

RETRAN-01 is the first released version of the computer code package developed by EPRI for analysis of light water reactor operational transients.

It is

!I based on the RELAP series of codes using homogeneous equilibrium flow equations

(

and has been through extensive verification and benchmark against separate ef-fects tests and plant transient data.

References 13 to 15 present some of the l

analyses conducted by GPUNC on events which have occurred at both TMI-1 and I

3-7 lI

I Dil-2 for which actual recorded data was availabic.

Detailed modeling assump-tions for the overcooling transient will be presented in section 6.

3.2.4.2.

The CSMP Model for Long-Term Cooling In order to " terminate" the overcooling event within the first few hours af ter initiation, a stabilized PC temperature must be maintained. The plant is con-trolled by using either the steam generators as a heat sink or by the primary side feed and bleed mode of operation. Tha transient is slow-varying with time and there are no sudden variations expected between the various locations in both the primary and the secondary systems.

Therefore, the nodal diagram can be simplified to two volumes - one for the RCS and one for the secondary side.

Instead of using RETRAN, the mass and energy balance equations can be solved very ef ficiently using the dynamic simulation language CSMP for an extended transient t ime.

Figure A-1 shows the schematic nodal diagram of the CSMP model.

A detailed description of equations and the methodology are provided in the Appendix to this report.

3.2.5.

Analysis Assumptions 3.2.5.1.

Plant Equ ipmen t Performance For the selected transient, the plant systems described in section 2 were as-sumed to perform as they were designed. The SLRDS was assumed to have been deleted from the EW system. Table 3-2 provides an equipment performance sum-mary by listing the system, its automatic setpoint, performance assumptions,

=

applicable fluid temperatures and specific failure assumptions.

Section 3.5.3 discusses the operator actions assumed during the event.

3.2.5.2.

Modeling Assumptions 3.2.5.2.1.

Core and Reactor Coolant System For the postulated turbine bypass control system failure, it was assumed that E

all six valves had stuck open simultaneously. The resultant overcooling occurs W

symmetrically in both reactor coolant loops; thus, a single-loop model could be used.

Figure 3-1 is a nodal diagram of the RETRAN model. The core com-prises a vessel downcomer, core inlet plenum, bypass region, and three axial core volumes with three heat conductors.

There is also a vessel upper plenum, outlet plenum, and a two-volume upper head.

Connected to the hot leg is the pressurizer, which is modeled as a non-equilibrium volume. This provides more accurate simulation of the repressurization expected during the transient. The pressurizer heaters are included although their contribution to this rapid 3-8

E overcooling event is minimal. The pilot-operated relief valve (PORV) and the safety valves are included in the model. The steam generator is modeled with 12 primary volumes, 12 secondary volumes and 12 heat conductors. The steam 5

generator inlet plenum and outlet plenum are represented, as are the downcomer and steam lines.

3.2.5.2.2.

Safety-Related Systems The safety systems modeled include high-pressure injection (HPI), the core I

flood tanks (CFTs), and the emergency feedwater system (EW).

HPI is initiated automatically when the RCS pressure decreases to 1600 psig.

Flow is injected I

into the cold leg as a function of RCS pressure as shown in Figure 3-2.

The temperature of the HPI fluid is assumed to be 40F (in lieu of 50F used in the SBLOCA analysis).

The CFTs are located within the containment, and the fluid temperature was assumed to be 100F. The CFTs are pressurized with nitrogen, and flood tank water is introduced directly into the vessel downcomer as the reactor pressure drops below 600 psig.

The EFW system was assumed to initiate following a loss of both main feedwater pumps or all four reactor coolant (RC) pumps. The fluid temperature was assumed to be 90F, and its flow (following an initial ramp-up to full capacity) was lim-ited by cavitating venturis to 1150 gpm.

Control of the EW is determined by the steam generator level. With all RC pumps running, the OTSG level setpoint I

is 30 inches.

Unlike the main feedwater system, which is isolated as steam generator pressure drops below 600 psig, the EW at TMI-1 will not be automat-ically isolated on low steam pressure.

3.2.5.2.3.

Main Feedwater and Steam Lines I

The TMI-1 main feedwater (MFW) system is controlled by the ICS.

It takes a number of inputs, such as the unit load demand (ULD) and turbine header pres-I sure. Once an error is established between desired and actual feedwater flow, the signal is sent to speed up or slow down the feedwater pump and to change the feedwater valve position. The detailed ICS control was not part of the RETRAN model. However, it was modeled as part of a separate study using the simulation language CSMP (see section 3.2.4.2).

This model was used parametri-cally to generate the feedwater flow during this overcooling event.

At the time of the turbine trip, the ULD is run back at 20%/ minute. Almost immediately, however, the reactor cross-limits and Btu limits force the feed-water demand to zero. The high turbine header pressure exceeds the pump 3-9 I

shutoff head, and the feedwater flow tecperarily ceases.

For this particular case, the decrease in header pressure due to continuous blowdown through the turbine bypass valves (TBVs) allows the pump to start flow again. A rigorous approach in generating the desired MW flowrate is to assume a reasonable tur-bine header pressure history, using it in the CSMP program to produce the feed-water flow rate, and then using that as input boundary condition in the RETRAN run to reproduce the turbine header pressure. The iteration process is termi-nated when certain convergence criterion is met.

In reality, af ter several iterations we found that the turbine header pressure transient is determined mainly by the blowdown process and is hardly affected by the feedwater input.

The result of feedwater flowrate is shown in Figure 3-3, and was used as a boundary condition for the RETRAN analysis.

Main steam was modeled with the assumption that flow through the turbine con-trol valve remained at 100% steam flow until the turbine tripped.

In reality, there is a regulating valve that reacts to the pressure drop and steam flow re-duction caused by the steam leakage through the TBV. This is an ICS control function that matches steam flow to the ULD. The modeling of the steam flow as a constant is conservative as f ar as overcooling is concerned since it re-sults in a larger steam blowdown than would be expected.

The steam line relief valves and TBVs are modeled as junctions, as are the feedwater input and steam outlet to the turbine. All six TBVs are lumped into E

2 one in the model with a combined area of 0.523 f t and a discharge coefficient of 0.7.

The RETRAN calculation, using the Moody choked flow model, resulted in a maximum of 27% of full power steam flow, which exceeds the design value of 22.5% for the bypass capacity. The assumption is thus considered to be con-servative and the calculated cooldown rate is more severe than actually would occur.

3.2.5.2.4.

Other Assumption The core power was assumed to be at full rated power (2568 MWt) at initiation of the transient. The reactor protection system (RPS) features that were modeled include the high neutron power trip (105% of rated), the low pressure trip (1900 psig) and the variable temperature-pressure t r ip. After reactor trip the decay heat was assumed to be 1.0 times the ANS value.

I 3-10

I 3.2.5.3.

Operator Actions Operator actions have an important effect on the outcome of an overcooling transient.

If (as expected) the operator follows procedures, his actions would result in an event that is not a serious thermal shock to the reactor vessel.

Table 3-3 summarizes the actions expected of an operator during this event, as well as an estimate of when they should occur. Not all these actions are re-quired to prevent a thermal shock transient.

In fact, only RCS pressure con-I trol is required to keep plant conditions within the guideline of interin brittle fracture curve (refer to Figure 3-12).

On the other hand, no operator action at all would result in a water-solid RCS and an event similar to the worst-case small break LOCA event that has been analyzed. Therefore, analyzing the event assuming no operator action was dis-carded, and operator actions for this event were minimized to cause a thermal shock transient without the event becoming a small-break LOCA.

Indication of an overcooling event of this magnitude is clear to the operator and can be dis-tinguished f rom a LOCA within several minutes (refer to Figure 3-12, which shows the P-T plots of this event). Once an event has been diagnosed as overcooling, isolation of the TBVs is one of the first actions taken by the operator.

Ob-I viously, rapid termination of the event would result in a less severe cooldown t ran s ie n t.

Nevertheless, this analysis assumes that no actions are taken to mitigate the overcooling event for 720 seconds - 12 minutes (refer to Table 3-4).

At 720 seconds, the subcooling margin reaches 120F and the operator is assumed to throttle HPI and isolate the TBVs. No operator action was assumed for throttling of EW flow, which is an extremely conservative assumption.

Proper action would be to throttle EFW when steam generator pressure decreased 100 psi below its expected value (to 910 psig for a post-trip situation).

Using this criterion, EW flow could have been throttled about one minute after event init ia t ion.

It is important to note that this criterion is not dependent on diagnosing the type of event in progress.

Without operator action, EW flow continues at full flow until the OTSG 1evel setpoint of 50% is reached at 1200 seconds. At this time, the EW control sys-tem throttles EFW flow to maintain the setpoint level. One hour after the cooldown begins, the operator is assumed to stabilize the plant. He adjusts the OTSG pressure setpoint to prevent RCS heatup. The pressure setpoint is chosen so that the OTSG saturation temperature corresponds to the RCS average I

3-11

temperature.

If earlier action la taken by the operator, RCS heatup would be minimized. Since the minimum RCS temperature is reached after HPI has been throttled, RCS pressurization would be stopped with the stabilization of RCS temperature. However, even if the OTSG setpoint is never changed, the opera-tor can control RCS pressure via letdown (i.e., control pressurizer insurges) or using the pressurizer spray, if available.

5 The operator is also assumed to take other actions not related to overcooling events.

First, he trips the RC pumps within 30 seconds of HPI initiation on a 1600 psig condition in the RCS.

The effect of the pump trip is to aggravate the cooldown because EW is initiated automatically upon pump trip, and the steam generator level setpoint is changed from 30 inches to 50% (212 inches higher). The unheated (assumed to be 90F) EW flow and higher level setpoint would accelerate the heat removal rate.

Second, the lack of pump heat (about 16 MW) to the RCS becomes a significant reduction of the heat source as the event progresses.

Finally, tripping the RC pumps causes a more rapid loss of subcooling margin. When RCS flow decreases, the hot leg temperature increases g

and subcooling margin is lost, thus delaying the time when HPI can be throttled.

M The operator is assumed to throttle HPI when the RCS is highly subcooled (120F),

this occurs 12 minutes into the event. The action is not dependent on diagnos-ing the event in progress.

In summary -

1.

Operator actions in conformance with procedures result in no serious thermal shock transient.

2.

HPI throttling is an action that would occur, regardless of the event, upon indication of 100F subcooling. This is the most significant operator action in mitigating this event.

For analytical convenience, analysis was performed at 120F subcooling.

3.

The assumed mitigating actions related to RCS overcooling were limited to recognizing the f ailure of the TBVs at 720 seconds and adjusting the OTSG pressure setpoint at I hour.

The latter action is not necessary in controlling RCS pressure.

I 3-12

3.2.6.

Presentation of Transient Results The results of the RETRAN calculation are shown in Figures 3-3 through 3-12, and the corresponding sequence of events is shown in Table 3-4.

Eight seconds af ter the leak is initiated, the reactor scrams on high neutron flux.

Imme-diately, the turbine trips and the main feedwater (MW) runs back because of ICS action and high pressure in the steam line. At about 45 seconds, MFW re-turns as the blowdown continues and the MW pumps ere able to provide flow into the steam generators. The RCS is depressurized at 84 seconds to 1600 psig, and the HPI is automatically initiated.

It is assumed that the operators trip the RC pumps 30 seconds after HPI initiation, according to procedures.

Subsequently, EFW initiates upon loss of all four RC pumps. At 302 seconds the CFTs inject into the RCS, momentarily interrupting the natural circulation flow (Figure Since there is no leakage out of the primary side, the volume of the additional liquid exceeds net shrinkage due to overcooling, resulting in repressurization and level indication in the pressurizer.

As discussed earlier, the TMI-1 emergency procedures require the operator to isolate the leak in addition to throttling or terminating HPI flow during an overcooling event.

He may also restart the RC pumps to prevent further over-pressurization, although this was not assumed in our analysis. The RETRAN run shows that at about 480 seconds into the t ransie n t, the RCS repressuriza-I tion begins and, subsequently (720 seconds), pressurizer level indication is regained.

It was assumed that the TBV leakage was isolated, and the HPI was terminated at this time.

Procedures also require the operator to control OTSG pressure to prevent the RCS from heating up and repressurizing. This operator action was conservatively assumed to be accomplished one hour into the tran-sient with the atmospheric dump valve setpoint controlled at 450 psia.

For the long-term transient following isolation of the leak and HPI termina-tion, a simple quasi-steady program is sufficient (appendix).

Figures 3-13 and 3-14 are the extended results of this analysis of RCS pressure and cold leg temperature. RCS pressure and temperature are stabilized at 1415 psia and 456F, respectively. The analysis assumed that the operator manually con-trols the OTSG saturation terperature to a value equal to the RCS temperature.

The downcomer temperature along with temperatures in the cold leg upstream and I

downstream of the HPI injection point are presented in Table 3-5.

Also in-cluded in the table are the RCS pressure and flow and HPI flow.

3-13 I

I 3.2.7.

Summary and Conclusions For the selected overcooling transient with a maximum of 27% steam flow fol-lowed by reactor and turbine trip, the minimum downcomer temperature was cal-culated to be 330F at 720 seconds into the transient. At this time the pri-mary side is being repressurized and pressurizer level indication is returning.

According to the present plant operating procedures and the instrumentation available in the control room, the operator will isolate HPI and start second-ary side cooling by using EFW and atmospheric dump valves.

Conservative estimates were used for operator action, so this analysis repre-sents an extreme response to an initiating event that could be expected to occur during the life of the plant.

The bulk fluid temperature experiences a cooldown rate of %20"F/ min. for the first 15 minutes of the transient.

I I

I I

I I

I 3-14 I

Table 3-1.

Comparison of Turbine Bypass Valve Failure With Other Overcooling Events I

Descr ipt ion Relat ive Relative likelihood of failure severity of occurrence Comments EFW overfill Less Comparable This type of event is analyzed in section 8.0 of the TMI-1 Restart Report (Figure 8A-1)te, Stuck-open secondary Less Comparable TMI-1 Restart Report, safety valve or section 8.0 (Figure atmospheric dump 8-A-16)18 valves Less(")

More severe evercooling Stuck-open turbine More I

stop valve initially, but more likely to be terminated quickly by operator ac-I tion; i.e., the f ir st two post-trip actions are to verify reactor and turbine trip.

Therefore, overall se-verity judged to be less.

Small steam line Comparable Less breaks Large steam line More Much less Analyzed in TMI-2 FSAR, i

breaks Appendix 15B (Analysis of Large Steam Line Breaks)l', which bounds TMI-1.

Total LOFW with More Less Analyzed in this report.

l lIPI cooling See section for small break LOCA.

Rancho Seco event Comparable Much less NNI/ICS power supply of 3/20/78 modifications reduce likelihood.

(a)The turbine control and intercept valves would also have to fail to open, which is considered less likely than a failure of the turbine bypass valves.

3-15 I

l l

\\

Table 3-2.

Plant Equipment Performance Systee Automat ic ser point s Equipment pe rf o rmanc e Fluid toep, F Aseused failures High-pressure RCS pressure: 1600 Two pumps delivering 40F None injection psig flow as per Fsgure 5-1 Normal makeup Far level: 220 inches one pump with flow con-85F None Y# "*

Isolation at RCS pressure 1600 pois Emergency RC pump trip Three pumps 90F None eedwater Level control 501 Flow limited by cavita-operating range ting venturis to 1150 spa SLRDS assumed de-lated from EFW No automatic isolation below 600 psis SG pres-sure Pressurizer Normal pressure set-All five heater banks None

%sters points for all five functional heater banks Heaters interlocked off when par level 1 5 inches and reset 8

when level is re-stored Heaters disconnected f rom bus on ESAS.

Must be manually re-loaded onto bus by j

operator.

"atn feedwater

• 7V isolation below Normal ICS control of two Feedwater enthalpy None (wFW) 600 pois SC pressure feedwater pumps of 397 Stu/lba II Instant cloeure of MFW Enthalpy maintained regulating valve constant - decrease in enthalpy due to loss of feedwater heaters considered minor Integrated 385 psig at 100%

Fails at time sero, in-NA Worst-case single pressure con-power itiating blowdown of failure in subsys-I ' # I '" '"~

• • • • ' 8'"'**

910 peig after tur-

""*"*I "8

tem of ICS bine trip open 1010 peig after re-actor trip Turbine bypass NA - This was in-All six valves f ail open.

Single failure of system (TBS)

Ltiating event.

Flows 27% of full-power turbine header EEE steam flow (a) pressure setpoint I'"******

Turbine control valve remains full-open until Control valve f ails turbine trip open Core flood Flow initiates at 600 Normal operation with 100F None psig 600 psig nitrogen over-pressure Reactor pro-Trip on 1051 rated Normal trip with 0.5 s NA None 3

tection sys-neutron power time delay Trip on 1900 psig

• urbine trip initiated RCS pressure with reactor trip Trip on variable low pressure Reactor cool-None (assume manually Full flow with four-NA None t

anc pumps tripped 30 e after pump operation HP1 initiation)

Normal coastdown rate following trip Reactor in-Flow initiates at low Not modeled. No flow NA All eight valves ternal vent differential pressure conseunication from core stay closed valves between core and downcomer Note t hat actual bypass capacity is 22.5% flow although the analysis was perf ormed at 27* flow.

3-16 I

C

. es 4 l I

O 9

I G

== 1 4

em 4 3 es 4

g &

== tee 9

3 b C 1

=

C== C 3

4 4==

es

• • A 9# 3 3

C s

1 C me

==

b e we 3 3e 9

=

C 4 A C

> 1 3*

a$

g

== 3 3

gD D 3

4 to C ee me =

no ee

=*

C es I

e.

• 3 3 9 N

J 3

ee 3

v 4 ' as A 3 C C v

9 1

& A==

b :e a

==

se====

es 4

s 34 "9

h.

a es

a. W 3 C Q

C.

=e ee Y

g 4

2 to b

le "3

he 3 he 3 28 en M e&C 1 N -e es 3 2 C M

C C

2 8== A a e aC he me 3

3

== eP es e h.

e he C

8E j 4 *e 1 e is.

es g we a

e a

e *3 M e.s es me C 4 en W

3 et 3 C o C6 $

.h g

y e

me =*

es Ch==

3 a3 I

=

h.

g 3== &

3 w to n.

et 3

to he

$ 9 U b b 1 2 7

4

• Qo}

3 =

3 N e.

O O

4. =A M A se fed 4 Z

• e Q= ne Z ee h.

w.

w

~

al

-. g A C C

. A C

3 % 3 =e

• 3 4 9 =e i he== 3 u 6

a t SAe D

• W G

e U 8

C $ 1 W g C

==e es C

3 =* *o*

y 3 he es "C

h. e aC Q

to 3 3

3 he w es h=== a 2==

ame es ne ^

==

=

es %

Q Q C 3J I=*=*

C "&

3 3 4

3 Ch * ]

• =a se C 1 dg W

W a= *e t u ee W

"3 le 3 1 "3","

3 C

3

's es he es es 3

.e 7>

C 3

3J

• 7 he S b Qe W 3 he 4 Q ee we D

=2 4.e he W en A 3 g =e A

me C

3 pe 3

h 3

2 3 3

.a. =g= A 3

& 1 3 ww a

C C

> 3 he w 3

A a==

g a we A me he me a

es em Q 3 3 em C

y M A

y g y ae

(

em,,

g e3 1

== C 1

a3 3 es. b 9

to a *=

0 D W ** i=

3 S w >

3 3 1

3

==

es Cb d ao h

he 2 4 e 3

-e Q

3.}

J C

C A=9 as==

es **

4 3., C 1 3 V

W 9 8 g C h

3 3 C

3 1 me

%e W C be O "3

Y W W

• • es *3 1

< es W

2. =8 ee. *J eMQ 3 <C =e.h C 0

W s==

en C he a.

3

=De M 3 1C w me we 2

=e S C u 3 0

43==**

3

=e hg &

C J **

e 5

e y C g

ee e g es

&g ne g a.

C C to A

3 C W

=

es aC 2 =e a y C

enge 3 28

=e 1 S

= =e

=e we g w A C S

a hs=

"y o.b U

d! ne se ew w

C a se se

=#

es W ad g 20 m C >

-o $3e g

g g

• 3 es Q y A C e==

sa 0 3

y==

me a he C 1

g 3 W C e ** W = *3 *3

• • Q 3 &W 3 as

• = 9 3=

3 es h.

A las t*

hs C W

=* 9 "C

.=.eJJJ 3

C 84 ee as v aS en 1 3 I

3 leJ V==

5 Se as & se C es we w g g

& >= g be a

1. 18 M3 3 Cb w

== 9 C 3 e

==

hs =e 3

y as O O ese C. ' o Ils lis 4.0 O

. as 3

C in C &3 A

C =e e W W se 4 O

Q Q e me Q E

he S

Q Q

es 3N

== $ $ $e =e C

• e==

==

4ee he m==

C

• • Q O

es a 1 B 4 1 E G am& b *e 3

n. Q== b== ese see "3 SM U leie C he C Q

Y 4'Oh3 1M W $

G C C D C 3 *3 4

es as aE Q em 3

1 1 as hs 3 3 ime se v es e w en 2 4ga hs 3 4 &@

43 ee J C ee 3 ge qmg q.s 43 C

$ C $ es C

$ $ esg he ) go s

se e==

g =g 6

ga g ee he

E C

>* h -.$

a

=e es *3 ee y u u

te k me Clm Q e W W O he J

== a 1C uev v=e>

a>

a==

w he v 3-C C C o

= U -e 3

. 8 i

I G$ M as Q w ind dl6 as a3 em < w e a Q. =e t

be 3 he y as fel *3 a Q ene a es g me g 3 aie C

O w

e.

C U

U N,

w w

m

==.

e.

==

g,)

me ao a

a w

a o

C C

4 5

==

4*

4 6 8

3 b

4 y

m.

N.

o.

e.

o w

n w

ed I 4 o

i e

M 88t f**

=*

b Ft i**

^4 N

ec 2

o w

no ;l}

h.

A o

o o o

n.

g g-

- n n

n n

- 4:

c c

.*.e 2

g.o n

y g

a

==

c C.

w

na

<v

.ea

<C o a ac

< :ma O

3 9

w f S es 2 e4

==

C E 20 t i M

Co

-e g a e

-== he g o e e I

3 h.

h.

-3 i en

- 3, g

r

= = a A

v a

i O

J w o 3

l

==.&h.

a 2 C a

h m.

=

C w US 3

A 3 w k =.

he ee *3 M h C es y

& ** 3 4,

l 3 3

• 3 m

C e 3 G &

3 W== CU t

G ej es awC e n =

= *e 4M w

w 5 t

3 a

1 C

>u i

m 3 g a

=e D no N he== ^ C

.O 3 ** C =

c

==w me 5 g 3

3 4

-4 he h 4***

3 C=

a 3.$

&ag to to to 3 3 1 3

3a I

q g

C-i a

• e U =

.A 3 v m

a 1

M 3 3

=v

-w

=3 2

b w

he h

U 8 A B C ee C he w U AA 1

7, e.

A 3 3

&.3 1 --

3 i 3 u =*

6 Q

==

3 -

M A A D U w we =e N M D

y

• .e ne y

3 w >

U S

$ he he

=e e.

3 to =e e

g 3 g 4 tem he me & Q, 3

he 3 es 8-8 g,,

a g C D a 3 en

==C

==

y Q U C

as he S 3 y

2 hs

== b es 4

1 es es e 00 he 0 h A a C 3

==

==

3 9

==

$ '3

=* C Cw

> C

• • 4 3 3 3

m h.

n. e a a u o

.a 3 3 i.

=

C v&=

, I sa 6,e.

= - e.e 3

h.

&p i

he C and 3

.J ee9 h.

A h.

1

.e es aa ne we g M A e3

• e be Q

.J a

1 0m C

C

- = =

o1 M

3 a

-o C

a e 1

• C 7

C A

C te W 3 l

==

A y o e,

== he n.

3 3

U C

9 1 3 **

al as.o

• 4 g A A w

y C

3 4

3 53 a

4==w&

Q 10 3

'e=

1 3

3 C l

ee a

w

• e n=

3 g

C 9

Je Cb l

L 1

"I es a se "C

3 I

ee

.=

te M

U 9

C 5AE 3 *3 E

l'a M

3 D

=*

v "In*e

.&.e 3

4

=#

Lal M e se 3 a

==DA Q es a

. i==

g W he

-e a

U w es a 3

to ee A g y

as A me g

& =#

Q A7 2

ll la.

=e C

ee

-e he 3 3

h.

>*e a

.B to=

>= C6 Q 8'e S=

l C

$ he leie U )

U I

Q be as &

as C.

8..

8..

i g

4 w

N.

w b

85 N

M ef..

~

, I a

o.

6 8..

8.*

8.*

f 0

l 3..

3 l

J

==

N

&g es

==

==

m

• =e*e N

en b

s g

==

1 3-17 l

l

\\

i Table 3-4.

Thermal Shock Overcooling Transient Analysis Sequence of Events Time, s Event Comment 0

100% rated power (2568 MWt).

0.01 Steam pressure regulator failure.

TBVs start to open and stay at full open position (total area = 0.523 ft2),

Main steam flow through turbine stop valve maintained at 100%.

8 Reactor trip on high flux (105%).

Turbine stop valve closes.

8+

MFW flow to steam generator reduced to zero because of high OTSG pres-sure after turbine stop valve clo-sure.

45 MFW flow back up as MFP speeds up (ICS response).

84 RCS pressure = 1600 psig.

HPI RCS is about 70F subcooled at initiated.

this point.

114 RCP tripped by operator.

EFW auto-Operator action to crip pumps g

matically started.

(EFW flow lim-30 seconds after HPI initia-E ited by venturis at TMI-1.)

tion.

141 MW isolated OTSG pressure dropped below 600 psig.

EFW continues feeding both steam generators (TMI-l specific).

g l

302 Core flood tank flow initiates.

E 380 RCS repressurization starts.

720 Pressurizer regains level.

Steam leak through TBV isolated; HPI ter-minated. Start using quasi steady-j state method for long-term trans-1 ient.

1260 OTSG filled to 50%.

EFW flow con-i trol valves maintain level.

l l

3600 Operator sets atmospheric dump val-Operator sets OTSG pressure ves at 450 psia and controls at control corresponding to RCS g

that point. RCS pressure stable at saturation temperature. This E

1425 psia.

action prevents heatup and repressurization.

I I

i 3-18 l

[

Table 3-5.

Overcooling Transient Results e8 temperature RCS I

pressure.

RCS flow, Downcomer Upstream Downstream

HPI, Time, s psia Ib/s temp., F HPI F HPI, F m

0 2165 3.88 x 10" 554 544 554 0

30 1928 3.92 x 10" 559 559 559 0

60 1768 3.97 x 10" 546 545 545 0

90 1577 4.04 x 10" 526 527 526 790 120 1424 2.08 x 10" 511 511 510 800 3

150 1244 3.39 x 10 493 496 487 800 I

3 180 1179 2.48 x 10 477 485 470 800 3

210 1020 2.68 x 10 461 464 453 800 3

240 826 2.46 x 10 446 450 438 810 3

270 742 2.05 x 10 428 437 421 830 3

300 683 1.54 x 10 417 428 408 850 3

330 639 1.61 x 10 377 418 380 850 I

3 360 634 1.73 x 10 380 413 391 850 3

390 590 1.74 x 10 375 402 382 850 3

420 553 1.70 x 10 365 394 375 850 8

450 517 1.54 x 10 349 389 367 850 3

480 516 1.48 x 10 352 386 362 850 3

510 518 1.56 x 10 357 383 359 850 5

540 524 1.60 x 10 358 379 357 850 I

3 570 530 1.65 x 10 356 373 353 850 3

600 538 1.72 x 10 352 366 348 850 3

630 548 1.76 x 10 347 360 343 850 3

660 558 1.78 x 10 342 354 337 850 3

690 569 1.77 x 10 336 349 332 850 3

720 579 1.73 x 10 331 344 327 0

3 900 603 1.60 x 10 331 331 331 0

3 1,200 607 1.60 x 10 332 332 332 0

3 1,500 645 1.60 x 10 342 342 342 0

8 1,800 704 1.60 x 10 355 355 355 0

3 2,100 777 1.60 x 10 369 369 369 0

I 3

2,400 867 1.60 x 10 382 382 382 0

8 2,700 979 1.60 x 10 395 395 395 0

I 3

3,000 1119 1.60 x 10 407 407 407 0

3 3,300 1295 1.60 x 10 419 419 419 0

3 3,600 1425 1.60 x 10 426 426 426 0

3 3,900 1425 1.60 x 10 426 426 426 0

3 10,800 1425 1.60 x 10 426 426 426 0

(3 hrs) lE 3-19

'1 I

1 5

gi 7

I I

{*

g L

}

i T

(! =i.i

.I w

.s m

t g -L m

lb

. af I @

s$ !

g i

z a > =; i g:

a ig u

i v

o.

~3 g

td w

i (1

3

I

2. @

$\\

h if f

3 8

A l I I:

t OtG l

4 m

3.k E ! I I I 9 It 9 I< t x

g g

A-3 I

I I I

I I

I I I

I I

O

.~-

92RM&5M-S5 5

m

,- @ MY26M gw a

2

=

=

=

=

Z t

c W

(

O 3 : = L 3 a it u J

a

c I

d h

h 1

73 3

e i

=

l l

-+

3 l!

=1 h

~~

A is a! e.

-lil e

ris

..s

=

i 7

s v

m 1

i l

o 3

T3 l

l I

"WI W

I

- --- (YY4U' 1

s l

a

=- - - - "

g

I

-S n

1 r

./II 5

i n,@

e l

g g

3 e

. : s e-ee.i ei

, C~.,

=- ---

. og a.

! bW l

lNhN NTNWhNN 3-G) 5W W

T c

()

y 5

=

I e.

ge c.

EE 8:

3-20

.I Figure 3-2.

HPI Flow Vs RC Pressure 8 80 - --=

I

=

TWO HPI TRAINS -

I 800 -

TWO HPI PUMPS I

720 -

640 -

G2 a

560 -

I Iz ONE HPl TRAIN -

ONE HPl PUMP 480 -

i I

,a 400 -

I 320 -

240 g

g g

0 500 1000 1500 2000 2500 I

RC System Pressure (psig)

I I

I 3-21

ZZ-C I

FEEDWATER 2 STEAM FLOW. LB/SEC 1s.

i.

2,.

m.

31.

w i

i i

I g

[

M.

I l.

,l e

I E

i E

I 1

w a

g 1

I b

l-(

h, I

2A u

3 m

O.

M I

ek u9 3

I O

C f

_i.i i

~

h I

s i

s 9

\\

g g

9

_l1 I

Q a

{

t t

t 1

I I

I I

I

I I

e g

4 g

l 4

I I

I 1

_ 9 I

I

- a I

i:

I s

E 8

I E

- $E

~

E W.

w I

I H;-

s I

T m

e o

'I

~ $

N Ca.

g S

~

~

!E l3 I

i t

t i

i e

0052 OSZZ 0003 OSLI 0051 0521 0001 054 OS blSd 'S3Md 03'l 10H I

I l

3-23

l I

I 8

6 4

i i

e i

a r

l 9

l I

g

.t.

I m

t I

- 2 1

I E

r u

2 o

1 gg g

oe mo w

x I

R~

o e r n

A o

Y t

3 3

o

~

~

p 3

3

~

~

r

_ g l

i i

i C39 095 0*5 005 09b OZb OEIG Ott Olf

.:i *dW31 d001 I

3-24

I I

E i

i i

i i

e i

i i

r

?

e I

_ g I

o

-l 0

~ 4 3

I e

m>

w I

- U-c c

k a

I L

I r.n

_ S. *-

.e e

A ow 9

i l

E a

l C

l I

_ g g

~

I l'!

0*l 6'

g*

l' g*

s.

6 g-p P

E350d W315AS OH176ECN I

3-25

I I

I e

- g R

I I

e IG l

g 7

C M

E

/

w l

N B;

w t

r I

?

I l

- s I

- g g

E s

I I

t i

1 l

-e o

ass /87 *nola saa I

3-26

I I-E

~

8 g

~

I

- 2 s

T

' I g-U b

!_H o

E Ho

a. ~

6 I

m w

=

- ?

3 I

8 e

I I

I

'a sa o s, s.,,

S t.

os s.,

og (131 73437 asio l

I 3-27 I

l l

I I

I a

4 4

i i

i i

i i

h

,?

,g

-o I

i 9

,g m

2 en>

>0 C

'@ E E

o

~

?

n u

a.

1 j

o I

i

,?

E o

E a

a

,3 I

1

,,g g

t 5

E2 al I

I l

I f

f f

I as ocs act ocs oas can aos aos 00:

9 WdO*n01. IdH QNU dflDOW 1

I 3-28

I 1

I I

=

i i

i e

i I

g I

G ws

_ 2*

m>

I w

@>c" gO I

4 N

N 3

5 gw 2

g I

O l

7 l

og u

3 BC w

ele s

I g

I I

ou czz mi at eat as a

a-of N1 '73A31 HZ' d d

I 3-29 I

I I

=

I

_ g I

- s I

I

- s E

8 s5 l

l 8

W a.

l 8

wz a

$?

M k

n 8

- 3 l

h u.

1

_ g I

_ g 1

i t

i i

e t

i aalt acci aos a

a aos aos e

a aat aR UISd '3805538d WU31S 3-30 i

l

!I Figure 3-12.

P-T Plot TMI-l TBV Failure Overcooling g

r i g

g

,-,,t.,

I To 100 O

I g

g

.::t;;'i!.__.

5.d t

I

' * -h

= rsu=ona

.z a.aJ I

$, g

-T

~

u _.

ses esor g

f..

tr f* Ice f*300 7 = 738 SEC I

f=790 aos 7 544 sa cots f*2**.ac b

f*3ee.

I v..

T*798 ***

Oe 300 NO 300 920 960 500 540 500 620 I

TEMPERATURE (F)

I I

I

M M

M 0

0 0

l I

W 1

1 0

0 l

I 0

W 0

1 0

W i

I 0

t 0

n 9

e isn 0

W 0

a I

r 0

T 8

gn W

i 0

l I

0 i

oo 0

c 7

r

)

e W

v C

0 O

I 0

E S

e 0

r 6(

u W

l E

ia 0 M F

i I

0 I 0 T VB 5

W T

l 0

0 I

i I

M 0

T 4

W 3

0 1

I 0

0 3

W 3

eru 0

g i

I 0

F 0

W 2

0 0

I m

0 1

m 0

0 0

0 0

0 0

0 0

0 0

0 0

0 0

0 0

0 0

0 0

0 0

0 2

0 8

6 4

2 0

8 6

4 2

2 2

1 1

1 1

2 m

RCS PRESSU

/

PSj m

Ye 3

l l

ll:

E E

E E

E E

E E

E E

E E

E E

E E

E E

E Figure 3-14.

TMI-l TBV Failure Overcooling Transient 800 l

I 1

1 I

I I

I I

l l

l 550 D

0 u

N500 c

0 M

E R

y 450 d

T E

M P

/ 400 F

1 I

350 l

I I

I I

I I

I I

I I

I 300 O

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 TIME (SEC) i

I M

02 7

s 1

O MS M

I 06 s

5 e

M m

i T

s 0

V i

I 8

4 M

y t

i l

au M

Q 00) d

=

t a

4C i

e w

E H

S

(

d n

E a

M 0I g

.=

i 3 T 2

e i

L k

l M

to I

y M

t 0

5 i

i.

i 9

1 2

3 s

e e

l M

ru t

g e

h i

06 F

i 1

M M

0 I

i 0

M

~

1 oo D.

5-a 3

D*

9 M

6ms_Goa aEiJI a2G ouJ-8x k

t M

M utw#'

I I

I 3.3.

Small Break LOCA Analysis 3.3.1.

Introduction to SB LOCA Transient During normal operation, the reactor coolant flows within the RCS as shown in Figure 3-16.

However, during a small-break LOCA (SB LOCA) transient the flow pattern is altered.

Figure 3-17 illustrates the flow for a stuck-open power-operated relief valve (PORV) or failed-open pressurizer safety valve af ter the flow coastdown period of the transient. During this phase, the RCS pressure and temperature are dropping and flow is decreasing as a result of the opera-tor tripping the RC pumps on ESFAS actuation as required by the procedures.

The temperature in the RCS is higher than in the secondary side of the steam generator; thus, the generators are available to assist in removing heat and I

circulating the RCS coolant. Assuming that no feedwater is available, the re-actor core must be cooled by injecting coolant from the HPI system through the core and out the valve. The warm RCS loop water is well mixed with the cold HPI flow as it passes the HPI nozzle; thus, relatively warm water enters the vessel downcomer.

If the transient proceeds unhindered, RCS temperature and pressure continue to fall and steam voids will form in the system. The rate of temperature and pressure decreases, and the void volume is primarily a function of break size.

At this point, natural circulation loop flow can cease with or without feed-water. Assuming an extended loss of feedwater and no forced RC flow, loss of I

natural circulation could occur in 8 to 15 minutes as a result of voids at the top of the hot leg.

When primary loop flow ceases, the cold HPI water injected into the cold leg will begin to cool the cold leg and upper vessel if no hot vent valve fluid is available for mixing. However, the cold HPI fluid mixes with the warm vent I

valve fluid in the cold leg piping before it flows into the vessel downcomer (as shown in Figure 3-18).

The system will remain in this condition, with the downcomer fluid temperature gradually decreasing as decay heat decreases, until RC loop flow is initiated. The RC pumps can be restarted once 50F subcooled conditions have been re-established around the entire loop. No credit is taken for this in the analysis; instead, it is assumed that the RC pumps are not started. Operator action was assumed (1) to throttle HPI flow when the core outlet temperature reached 100F subcooled and then (2) to maintain approximately 100F subcooling at the core outlet.

3-35 I

3.3.2.

Definition and Selection of SB LOCA Transient The following criteria were used in selecting the SB LOCA and overcooling t rans ient s:

1.

The transient must be realistic.

It may be a result of a normal operating transient and will evolve in a mechanistic manner.

2.

The transient must have a reasonable probability of occur-ring.

Systems must be challenged to fail and may contain single or multiple failures, i.e., an initiating event plus one or more failures.

3.

The transient provides a realistic challenge to the reactor vessel; along with the resulting failures, it provides a thermal shock to the vessel.

The initiating event (i.e., loss of all feedwater) for the SB LOCA transient was chosen to provide a mechanistic manner in which to challenge the system and provide the necessary criteria for obtaining a thermal shock of the vessel, that is, by providing a once-through mode of cooling. The mode of cooling is the cold HPI flow, which is injected into the cold leg of the plant, passes l

through the core, and exits the system through a break.

Therefore, the initiating event that will challenge the code safety in a mech-anistic manner is a loss of all feedwater or any event that delays auxiliary feedwater (AFU) for more than 8 minutes after the loss of main feedwater (MFW).

l To allow the code safety valve to be challenged during the transient, the PORV is assumed to be isolated.

W The mechanistic break size chosen using these criteria was a single pressur-g izer code safety valve (0.018 ft2).

The PORV break size was considered in W

1 BAW-1648 since it led to repressurization of the RCS. However, the PORV u an-sient provides a less severe thermal shock than the large safety valve break 2

because the RCS temperature remains considerably warmer than with the 0.018-f t break, and the HP1 flow is less. Also, mechanistic break sizes larger than a single code safety valve (0.018 ft2) located in the pressurizer are expected to be less probable when challenged. That is, during the challenge of the hardware, it is expected that a single safety valve failure is more likely than either both safety valves or a combination of PORV and safety valves.

3-36 I

i 2

Break sizes larger than 0.018 ft result in more rapid depressurization to pressures at which the low-pressure injection (LPI) system provides makeup I

(with little or no repressurization).

The transient response of these larger breaks is similar to that of the large-break LOCA being considered under NRC Task Action Plan A-II.

An examination of the system response for various break locations shows that the limiting condition will occur for a break in the pressurizer or hot leg independent of break size. This results from the following considerations:

For cold leg breaks, the hot water leaving the core flows (1) through the hot I

leg, steam generator, and broken cold leg to the break; and (2) through the vent valve, downcomer, and broken cold leg to the break. The latter path has the least flow resistance and thus allows a large portion of the hot water to enter the downcomer for mixing.

Furthermore, the diversion of HPI water to the break reduces the total amount of HPI water entering the downcomer. For hot leg or pressurizer breaks, more HPI water is available to enter the down-comer.

In addition, less vent valve flow occurs in a hot leg or pressurizer break, thus decreasing the amount of hot water available for downcomer mixing.

The analyses that follow used a pressurizer break to evaluate system condi-I t ions.

It is further concluded that breaks within the hot leg will respond in the I

long term in a fashion similar to those in the pressurizer. Thus, the com-bination of system response, break location, and break size determined that the evaluated accident would be the failure of a pressurizer code safety valve.

The SB LOCA transient thus analyzed in this report is a failure of the TMI-I pressurizer code safety valve (0.018 f t2) to reseat after being challenged by high pressure because of a loss of all feedwater. The valve cannot be iso-lated and thus continues the blowdown until the RCS can be placed on the decay heat removal (DHR) system. Since feedwater is not restored during the event, both forced and natural circulation RCS flow are assumed to be lost for the duration of the event.

3.3.3.

Major Inputs for Selected Transient It is known from previous B&W work on thermal shockl that conservative, bound-ing assumptions that bound all B&W operating plants will result in vessel crack I

3-37 I

instability within a service lif e of 3 to 6 EFPY. Therefore, the najor plant-specific inputs were chosen to provide a realistic value for each specific pa-rameter. The break size and initiating event discussed in section 3.3.2 were chosen to produce a transient that unfolds in a mechanistic manner.

l The major parameters used in this plant-specific SB LOCA analysis are provided in Table 3-6.

The final inputs provide a plant-specific, realistic SB LOCA transient analysis for TMI-1.

3.3.4.

Analytical Method 3.3.4.1.

Assumptions and Conservatisms l

Input assumptions and conservatisms used in the BAW-1648 SB LOCA analysis have been evaluated in detail to allow a more realistic and plant-specific analysis. Thus, the bounding assumptions used in the previous analysis have been eliminated in this analysis.

In general, the previous analysis was a worst-case analysis using inputs bounding all currently operating B&W plants.

The following assumptions are used in the CRAFT analysis":

1.

Loss of main and auxiliary feedwater occurs simultaneously at time zero.

2.

The reactor tripped at RPS high pressure trip setpoint.

3.

The break initiated at the code safety valve pressure setpoint.

4.

HPI is initiated at the ESFAS setpoint.

1 5.

The RC pumps tripped at the ESFAS setpoint.

6.

The Moody correlation, with a discharge coef ficient of 1.0, was used to calculate saturated flow through the open pres-surizer code safety valve.

7.

Bernoulli's correlation, with a discharge coef ficient of 0. 7, was used to calculate subcooled liquid flow through the code safety valve.

Two assumptions were used in the steady-state calculations:

1.

HPI flow is throttled if the core outlet temperature reaches l

100F subcooled and is throttled to maintain 100F subcooling.

l 2.

The pressurizer water and core outlet temperatures are the same.

3-38

I 3.3.4.2.

Analytical Model l

An eight-node CRAFT model ' was developed to determine the thermal-hydraulic l7 conditions existing in the RV for the NRC-requested analysis completed in BAW-1648.1 The present analysis used the same eight-node CRAFT code with plant-specific inputs. Tnis model was compared (in BAW-1648) with the 22-node CRAFT model developed for the SB LOCA analysis. A comparison of the results from both models (shown in Figures 3-19 and 3-20) shows that the sim-plified model can adequately predi-t the thermal-hydraulic conditions in the vessel. The sudden drop in the downcomer temperature shown in Figure 3-19 caused by the initiation of HPI and the delay incurred in the eight-node model is not considered significant. The system noding for the eight-region model is described below and shown in Figure 3-21.

Node 1: Reactor vessel downcomer and lower plenum Node 2: Reactor core and upper plenum Node 3: Cold legs between RC pumps and reactor vessel Node 4: Hot legs Node 5: Primary side of steam generators and cold legs between steam generators and RC pumps Node 6: Secondary side of steam generators Node 7: Pressurizer Node 8: Containment Node CFT: Core flood tank The CRAFT code assumes homogeneous mixing of the liquids in a node and deter-mines its thermodynamic conditions based on the thermal equilibrium between the steam and liquid phases. This assumption will result in complete mixing of the cold and hot fluids entering the downcomer region from the cold legs and vent valves, so the downcomer node temperatures calculated by CRAFT are mixed mean temperatures.

However, this temperature is not used in the mixing or fracture mechanic analysis.

The primary objective of the LOCA analyses I

was to determine the HPI flow rate, vent valve flow rate and temperature, and RCS pressure for use in the fluid mixing analysis.

I 3-39

I The SB LOCA analyses used the CRAFT code for the f irst 143 minutes of the transient. A steady-state analysis was performed to determine RV conditions once the core exit became 100F subcooled.

The analyses assumed that 100F subcooled was reached 180 minutes into the transient, the steady-state anal-ysis continued for some 40 minutes (or until 240 minutes) to provide steady-state data for RCS pressure, etc.

Once the 180-minute RCS parameters were E

obtained, the transient was extrapolated from the 143-minute (end of CRAFT analysis) to the 180-minute data calculated by the steady-state code.

This 1

steady-state analytical method was benchmarked in BAW-1648 against the CRAFT 2

analysis" for the 0.007-ft pressurizer break with no operator action. The results indicate that the steady-state code predicts the downcomer temperature approximately 9% above the CRAFT prediction. The 9% deviation in downcomer temperature has no impact on the core outlet temperature calculated by the steady-state code.

Under the steady-state assumption, therateofchangeofmass(gCS}""

• "*#EY (E

RCS RCS, dE dM RCS

= 0.

(3-1)

E dt dt g

Then the core outlet enthalpy is calculated:

I 9

+h h

~

h W

HPI HP I I

where h = core outlet enthalpy, Btu /lbm, h

= enthalpy of HPI water, Btu /lbm, HPI W

= HPI flow, Ibm /s, gp7 q = decay heat, Btu /s.

The vent valve flow is determined by performing an energy balance in the downcomer region:

~ bPI' (3-3) c y

,y vv HPI

-h where h = core inlet enthalpy, Btu /lbm, W

= vent valve flow, Ibm /s.

vv I

3-40

I The vent valve flow, which can also be calculated from the elevation pressure drop across the vent valve, is given by 2

288 x gpA 3p g x 3p 2

c h vv

= 96.26 A (3-4)

W vv K

vv K

I p

re utlet density, 1bm/ft3,

=

h A

= vent valve flow areas, ft2, loss coef ficient, K

=

I AP = elevation pressure drop, psi H(o

-p h)/144

=

= core inlet density, 1bm/ft3, o

I H = elevation head, f t.

Assuming a system pressure and a downcomer water temperature (T ), the vent c

valve flow W can be calculated using equations 3-3 and 3-4.

The downcomer yy water temperature is determined by iterating on the assumed Tc until equations I

3-3 and 3-4 predict the same vent valve flow.

In the steady-state condition, the relationship of HPI flow to leak flow is I

defined as:

P 4

HPI W

=W (3-5)

L HPI v

W = leak flow, lbm/s, W

= HPI flow, lbm/s, HPI 3

v

= specific volume of HPI, f t /lbm, HPI 3

v = specific volume of core outlet water, ft /lbm.

Assuming a system pressure and 100F subcooling, the leak flow can be calculated by Bernoulli's equation with a discharge coefficient of 0. 7.

Equation 3-6 is used to determine the HPI flow (WHPI) required to maintain 100F subcooling.

l Equation 3-5 is used as a convergence criterion for determining the system pressure.

If W and W satisfy the equation, the system pressure is deter-g mined, and then the equations (3-3 and 3-4) are used in the same manner to de-termine the vent valve flow.

I f

3-41 lI

I 9"W

~ HPI)

(3-6)

HPI h q = decay heat rate, Btu /s,

%' = enthalpy of core outlet water based on the 100F subcooled state, Btu /lbm, h

= enthalpy of HPI water, Btu /lbm, MPI W

= HPI flow rate, lbm/s.

HPI I

An evaluation was performed to determine whether choked flow ever occurs downstream of the pressurizer code relief valve. Typical calculations were provided in BAW-1648.1 It is suf ficient to say that choked flow will occur W

only through the code relief valve during the transient.

Modeling of the piping and quench tank downstream of the pressurizer code re-lief valve was treated as part of the containment since flow choking always occurs at that valve.

For purposes of this analysis, the HPI flow has been injected directly into the downcomer at the inlet e leva t ion.

HPI flow into the cold leg pipe volume W

(node 3) between the reactor vessel and the RC pumps will reduce the node temperature below the lower bound of the steam table used in CRAFT".

In order to avoid this dif ficulty, the HPI flow was injected directly into the down-comer. The downcomer temperature (Figure 3-22) calculated by CRAFT is not used in the fracture mechanics analysis.

3.3.5.

SB LOCA Results 3.3.5.1.

Transient Results The sequence of events for a pressurizer code safety valve failure transient is summarized in Table 3-7.

This SB LOCA was allowed to evolve in a mechanistic manner using TMI-I plant-specific setpoints and component characteristics.

Following initiation of the loss of all feedwater, which is assumed to initate the SB LOCA transient, the loss of cooling caused the RCS temperature and pressure to rise until the re-actor tripped on high RCS pressure (Figure 3-23) at 18.2 seconds into the transient. The pressure continued to rise since the PORV is assumed to be isolated and the pressurizer code safety valve is challenged, opens, and then is assumed to remain open for the remainder of the transient. Once the code safety valve opens (%21 seconds) the system pressure decreases rapidly and 3-42

I IIPI is initiated at the ESFAS low-pressure setpoint at 54 seconds. The op-erator is assumed to trip the RC pumps at this time.

From this point the transient proceeds unhindered until the operator begins throttling HPI flow when the core outlet temperature reaches 100F subcooled, which occurs 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> after the loss of feedwater incident. The loop flow ceased early in the tran-I sient (%I1 minutes), as shown in Figure 3-24, because of the loss of steam generator cooling and voiding in the RCS.

The total loss of feedwater assump-tion was used only to allow a mechanistic way of initiating a realistic tran-sient that would challenge the pressurizer relief valves. Since the break size assumed in this transient will cause a loss of natural circulation even with feedwater available, the feedwater assumption contributes little to the I

overall impact of the transient. Steam generator heat removal was essentially lost at 5 minutes. The core outlet temperature (Figure 3-25) reached satura-tion by 2.3 minutes into the transient and returned to subcooled at 17 minutes.

I After the core outlet temperature returned to subcooled at 17 minutes, the sub-cooling margin (Figure 3-25) continued to increase until about 95 minutes into the transient. The CFTs began discharging into the RCS shortly before 95 min-utes, causing the pressure to decrease at an increasing rate (therefore, TSAT decreases). This results in a slight decrease in the subcooled margin. At s100 minutes, the core outlet temperature begins decreasing its rate of cool-down due to the leak rate decreasing (Figure 3-26) as a result of the quality increasing in the pressurizer leak flow path.

This results in the subcooling margin actually decreasing until about 125 minutes, when the pressurizer final-I ly becomes subcooled and the leak flow rate quickly increases. The increase in leak flow results in an increase in core cooling and thus the core outlet subcooling margin reaching 100F at 180 minutes. At this point, it is assumed that operator action to throttle HP1 flow to maintain core outlet at 100F sub-cooled would take place (Figure 3-27).

At the beginning of the transient the pressurizer level (Figure 3-28) quickly rises to an almost full condition, followed by a quick decrease to the normal operating level af ter reactor trip.

As soon as the ECCS is activated, the pressurizer level again increases to the full condition for the remaining portion of the transient. The LPI system, although actuated, will deliver minimal (if any) flow to the RCS until RCS pressure drops below 200 psig. This will have no effect on the transient since the operator will be managing the core injection flow from all available sources at this time to maintain core outlet at less than 100F subcooled. The I

3-43 I

vent valve flow (Figure 3-29) continues throughout the transient. The vent valve temperature is the same as the core outlet temperature (see Figure 3-25).

The qualities for vent valve and pressurizer relief valve leak flow are pro-vided in Figure 3-30.

The TMI-I plant-specific HPI system was used in this analysis; it is compared with the previous bounding analysis (BAW-1648) in Figure 3-31.

The TMI-l plant also contains venturis in the HPI lines, which limit HPI flow once the system pressure decreases below about 1100 psi.

The smaller pressurizer code safety valve area (0.018 vs 0.023 ft2) also decreases

=

the rate of energy removal and thus leads to the core outlet temperature and downcomer temperature remaining to less than 100F subcooled for the entire 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> of the analysis.

Thus, the combined ef fect produces a slower cooling transient.

The SB LOCA transient analyzed is shown in a pressure-temperature diagram in Figure 3-32.

This is the presentation an operator would see using the new displays developed and included in the new approach for handling abnormal transients, i.e.,

the Abnormal Transient Operating Guidelines (ATOG) approach.

The operator would immediately be aware of the loss (140 seconds) of subcool-ing as the transient goes f rom point A to saturation at point B.

With the UPI system at maximum flow, it is not until point C is reached (180 minutes) that any operator action need be taken to throttle HPI flow to meet the re-quirement to limit core outlet subcooling margin to 100F.

If one is looking W

at this display, it will become obvious that even shorter times are long enough for the operator to take action to prevent exceeding the 100F subcool-l ing margin limit.

3.3.5.2.

Effect of Decay Heat Assumption on Results l

The decay heat power level assumption was changed from infinite operation and 1

1.2 ANSI (BAW-1648 ) to 400 EFPD operation and 1.0 ANSI to make the assumption W

l l

more real is t ic. S ince the transient pressure and temperature are atfected by the assumption of operating history for decay heat level, the effect of assum-ing decay heat levels of less than 400 EFPD was evaluated. A CRAFT analysis was performed with only a change in the input decay heat level. This run used I

a 1.2 ANSI versus the 1.0 ANSI analysis reported here. With this 20% differ-ence in decay heat and considering the period of interest for thermal shock as the first 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> of the transient, the reduction in decay heat causes the RC pressure to decrease about 20% faster and core outlet temperature to decrease 3-44

I about 30% faster, while HPI and vent valve flow remain escentially unchanged.

If the reference 400-EFPD operation curve for the first 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> after a re-actor trip is decreased by about 20%, the results bound operation as low as about 20 EFPD.

It is considered realistic for the operating history to be longer than 20 EFPD during the time an operating reactor is pressurized or isolated f rom the decay heat system. Therefore, if a period of 20 EFPD were analyzed, it is expected that the pressure and temperature would decrease by the 20 to 30% stated above with 100F subcooling being reached slightly sooner than the 180 minutes obtained in this analysis. This should have only minimal impact on the fracture mechanics analysis.

3.3.5.3.

Ef fect of Planned Modifications to CRAFT 2 on SB LOCA Results The effects of planned upgrades to the CRAFT 2 code, as required by NUREG-0737 (II.K.3.30), were evaluated to determine whether the results obtained with the present CRAFT 2 code would be significantly altered when these changes are im-plemented. Therefore, an evaluation of each planned modification was per-I formed to determine its impact on this SB LOCA thermal shock transient anal-ysis.

These code modifications are as follows:

1.

New steam generator model.

2.

Non-equilibrium pressurizer model.

3.

Two-phase pump model.

4.

Two-phase flow model.

Upgrades 1 and 2 could have an impact on the analysis results. The steam gen-erator model upgrade could influence the transient prior to the loss of the steam generator heat sink.

However, since auxiliary feedwater is assumed not to function in the analysis, only the first few minutes of the transient would be affected. The non-equilibrium pressurizer model could af fect the transient since the break analyzed was the failure of the pressurizer code safety valves.

During the first few minutes of the transient a steam space is maintained in I

the pressurizer, and non-equilibrium pressurizer conditions might occur.

How-ever, once the pressurizer fills and a two-phase mixture exits the valve, the non-equilibrium model would not affect the transient.

The conclusion is that since these changes will affect the transient for only I

the first 5-10 minutes and the thermal shock concern exists for longer than the first 5-10 minutes of the transient, then these changes will have an in-significant effect on the present results.

3-45 I

I Item 3 relates to improved prediction of RC pump performance while the system i

is voided and the pumps are running.

Item 4 relates to improved prediction of the two-phase flows that can occur while the system is voided and the RC pumps are running. Since the thermal shock analysis assumes that the RC pumps are tripped, these code modifications will not impact the present results.

The changes planne ; in response to II.K.3.30 could have an impact on the SB LOCA transient analyzed for thermal shock.

However, this impact is limited to the first few minutes of the transient and will not significantly af fect the present thermal shock SB LOCA analysis.

Table 3-6.

SBLOCA Analysis Assumptions Parameter Value 2

Break size, ft 0.018, single code safety valve Location Top of pressurizer Initiating event Loss of all feedwater BWST temperature, F 50 HPI system TMI-1, two HPI pumps 1600 psi, 784 gpm 1200 psi, 835 gpm l

600 psi, 841 gpm

=

300 psi, 844 gpm No. of internal vent valves 8

Core power level, MWt 2568 Decay heat 1.0 ANSI Cycle length, EFPD 400 Structural metal heat used Yes E

RC pumps tripped Yes, at ESFAS actuation W

Operator action Yes, at 100F subcooling at core outlet, 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> I

I I

3-46

I Table 3-7.

SBLOCA Transient Sequence of Events Time,

Sequence of events minutes Loss of all feedwater 0

Reactor and turbine trip 0.3 SB LOCA initiated (pressurizer code safety 0.34 valve failed open)

I HPI initiated, RC pumps tripped 0.89 Reach saturation at core outlet 2.3 Loss of steam generator heat removal capa-5 bility Loss of natural circulation 12 I

Regain subcooled state at core outlet 17 Achieve 50F subcooled at core outlet 60 Core flood tanks began dumping into RCS 80 Achieve 100F subcooled at core outlet 180 Began throttling of HPI flow 180

'I I

I I

I I

I I

3-47

'I

I Figure 3-16. RCS Flow During Normal System Operation HOT LEG PIPE B

h-I COLO LEG PlPE h

'f PRESSURIZER I

ln I

b Lk

'b d I

-.p.

h F

CORE suggE LINE a

N REACTOR VESSEL DOWNCONER

_ HPI INJECTION N0ZZLE STEAM GENERATOR I

I I

I l

3 3

3-e

I I

Figure 3-17.

RCS Flow During Total Loss of Feedwater Event With a Small Break in Pressurizer VALVE HOT LEG PIP

!l TO QUENCH TANK 6

s

I r-COLO LEG PIPE

',n.'

e,.

I PRESSURIZER VENT VALVE U

/

l M

h I

!i

\\

'l Y

.**L Ad

',y

=s n

l h

f CORE SURGE LINE

~~

. V

• EACTOR R

u s"

D DOWNCOMER HPl INJECTION FROM BWST STEAN i

GENERATOR

.I I

15

I I

3-49 I

I I.

m. 3 18.

s._

v... 1 mm, e.m....

s

+

I-1 1

I

.,xx_v @ trg QM%Q l

y V

l I

~

L fz ei e

3 M l0l$~!lUldl*,'4 g

c v k

k;!l j e em 27 =-

v +

'kI

!ji '

- (l ourur norru, re "***La g-

.E f t-F E

^

s

,,,,,u,.

~

Q, g

,,, m y ; g

, y

.=7 (T

s

/

s t I

s y'

=

g stacrom vtsstL

~

l a

s d

/

N i !

I s m s

4 s

d ((.".., -,,_....si '{

k k

k a

g y

aisaaU 4 2 8 aM g

f %.k8; al2.. _

'1 I

---.d' I

I 3

se

I Figure 3-19.

Downcomer Temperature Vs Time, Comparison of Multinode and Eight-Node CRAFT Models, Stuck-Open PORV, Two HPI Pumps, AW at 40 s 575 570 565 l

I i

l' il 580

'I 3

SHARP'OROP CAUSED BY HPl INITIATl0N e

j 555 S

E o

,I IlulTI.N00E 550 lI 545 3.N00E I

540 I

I I

I I

I 800 700 O

100 200 300 400 500 Time, Sac I

I I

lI 3-51 I

m

___,_y

I Figure 3-20.

Downcomer Pressure Vs Time, Comparison of Multinode and Eight-Node CRAFT Models, Stuck-Open PORV, Two HPI Pumps. AFJ st 40 s I

2400 I

3 2200 2000 I

E 1800 i

I a.

O E

1600 3

a Ea 8-N00E 1400 I

1200 NULil-N00E I

1000 I

g I

100 200 300 400 500 l

Time, sac l

I I

3-52 I

O E

E E

E E

E E

E E

E E

E E

E E

E M

E t

Figure 3-21.

CRAFT Noding Scheme. Eight-Node Model of RCS

.i g

Pzr y

Hot y

lee u-O G 23 I

4

.t

n.

Core 6

$p a

u" J

Flood ag Tank v

w g

5 core c,1, 8

'8 Q t

Q 2

a G

l NODE 8 IS CONTAIMENT N00E

Figure 3-22.

RV Downcomer Temperature Vs Time, SBLOCA Transient With Operator Throttling of HPI at 3 Hours 61 0 580 550 520 490 o'

460 430 400 v

0 2

3 370 340 310 280 250 220 190 1

A I

R a

i I

I I

O 20 40 60 80 100 120 140 160 180 200 Time, minutes l

W W

W W

W W

W m

M M

M W

W W

W W

W W

W

1 l

l Figure 3-23.

RV Pressure Vs Time, SBLOCA Transient With Operator Throttling of HPI at 3 Hours 2560 2400 '

2240 1

l 2080 '

]

1920,

1 1

1760 m

)

M i

1600 a

5 1440 -

0 u

h

[

1280 1120 960 800 640 480 320 160 1

I I

i i

R R

t t

3 i

0 20 40 60 80 100 120 140 160 180 200 Time, minutes

I l

Figure 3-24.

RCS Loop Flow. Path 3 SBLOCA Transient 40,000 37,500 Q 35,000 i

^

32,500 i

30,000 a

j g

27,500

.m 3

25,000 l

22,500 r

l 4.

l o

l Y

Z 6,000 l

5,000 1

4,000 3,000 2,000 l

1,000 i

0 a

i e

0 1

2 3

4 5

6 7

8 9

10 11 Time, minutes 1

l t

m M

e m

m m

m W

M M

M M. _ _

-.M._M.

. _ _.e..

M m

e e

i i

Figure 3-25.

Core Outlet Temperature and Subcooling Margin Vs Time SBLOCA Transient With Operator Action at 3 Hours 640 i

610 I

i 580 1

550 l

u.

0 o

520 -{

g N#

d f

N 3

l 490 o

N

=

460

[

50 =

u.

N 7"'%

2 N

g y

g 430 i

U b

1003 400 E

l j

370 p

340 310

{

280 j

250 220 190 8

l 0

20 40 60 80 100 120 140 160 180 200 Time, minutes i

i i

i I

I Figure 3-26.

Leak Flow Rate Vs Time. (Path 7) SBLOCA Transient With Operator Action at 3 llours 220 200 180 O

~

i 160 5

SYSTEM RETURNS SOLID C

140 m

lf i

120 100 u

I

~

I 80 60 40

/

20 0

20 40 60 80 100 120 140 160 180 200 Time, mi nu t e s W

M W

W W

W W

W W

W W

W W

W m

W W

W W

A M

M M

M M

M M

M M

M M

M M

M M

Figure 3-27.

HPI Flow Vs Time, SBLOCA Transient With Operator Action at 3 liours 120 N

l OPERATOR ACTION TO l

100 THROTTLE HPl FLOW TO MAINTAIN CORE OUTLET TEMPERATURE 100F SUBC00 LED i

i 80 i

2 a

1 1

1 0

l g

60

=

i o

m l

5 40 i

i l

l 20 j

I I

I I

I I

I I

I l

0 i

0 20 40 60 80 100 120 140 160 180 200 Time, minutes i

I

l I

Figure 3-28.

Pressurizer Level Vs Time, SBLOCA With Break on Top of Pressurizer 45 I

40 i

PRESSURIZER FULL t

35 4

m

~

i 30 u

t j

25 i

20 l i

i 0

20 40 60 80 100 120 140 ISO 180 Time, minutes W

M W

W W

W W

W W

m W

W W

m m

a m

e

Figure 3-29.

Vent Valve Flow Vs Time, SBLOCA Transient With Operator Action at 3 Hours 1500 -

1400 1300 1200 ay 1100 E

1000 d

o 900 j

800 u

i u.

E 700 L

g 600

___ CHANGE IN VENT VALVE RESISTANCES c

s 500 INPUT AT CRAFT RESTARTS 400 300 l

l 200 100 t

I i

i i

I i

I I

I O

20 40 60 80 100 120 140 160 180 200 Time, minutes i

m e

0 8

1 e_

0 i

6 1

e_

em i

0 e_

T e

4 1

sV y

t e

i la 0

\\

i 2

u 1

Q e

e v

la V

s 0

e t

i 0

t n

V i

e 1

u e

n d

m n

a e

i 8

i m

0 m

ka T

e L

E V

r L

m e

A z

V ir T

0 u

Nn 6

s E

s K

V m

er A

E P

L R

0 E

0 S

Z i

4 3

I A

3 R

E U

e S

r S

u E

m g

R i

P F

0 e

2 m

0 W

0 0

0 0

0 0

0 0

0 0

0 0

0 9

8 7

6 5

4 3

2 i

1 1

• a3a W

m w

W il

!4 jjIi i

lI!l

lII, I,

I

0 0

0 2

0 0

1 8

1 N

0 I

0 0

8 E

1 S

U

)

8 G 4 N 6 I

1 D

0 N W 0

U A 8

4 O B 1

B

(

pI er u

s s

0 e

8 0

m r

2 p

1 g

P S

C e

t R

a s

0 R

V 0

i 0

w 1

o wo l

F l

F I

P H

0 a

0 8

1 3

3 N

0 er E

0 t

u T

6 g

S P Y i

M S F

l

- U I

P I

M P

T 2 H 0

t 0

4 0

0 t

2 0

0 0

0 0

0 O

0 0

0 0

0 5

0 5

0 5

2 2

1 1

af E paf[E

["

i

Figure 3-32.

RCS Pressure Vs Temperature for SBLOCA Transient 1

2600

/,n 2400 f /

2200 A

/

2000

/

1800 100 F SUBC00 LING f

1600

/

/

ai 1400

,/

e m

4.

U 1200 2

1000 M

OPERATOR ACTION TO

/

2.3 MIN.

/

800 THROTTLE HPl FLOW

/

y rr 3 HOURS

/

15.5 MIN.

600

,'/

l 400

/

p' SATURATION 200

~~

0 8

8 200 300 400 500 600 620 RCS Temperature, F

i um amm sua amm som uma mas uma sus em num mim imm uma amm amm a

num

=

I E

4.

FLUID MIXING IN RCS COLD LEG AND VESSEL DOWNCOMER The results of the vessel stress analysis depend on the vessel wall thermal grad ie n t. The thermal gradient is governed largely by the extent to which the cold HPI fluid and hot vent valve fluid are mixed. This is particularly important when there is no cold leg flow, as in a SB LOCA with loss of all feedwater.

In that case, the fluid entering the downcomer could be cold HPI fluid if it were not mixed with the hot vent valve fluid. As part of the B&W Owners Group and the TMI-1 thermal shock analysis program, various studies were undertaken to evaluate this mixing process. The ultimate output from these evaluations was the fluid temperature adjacent to the reactor vessel (RV) wall, which determined the vessel azimuthal temperature profile. The geometry of the vessel cold leg and vent valves required multi-dimensional analysis. Since natural circulation exists af ter RC pump coastdown for a period of time, gravity effects and turbulent exchange models must be included to ef fectively model any mixing that might occur.

During these programs the evolution of mixing models progressed from a simple approach using a turbulent jet analytical model that assumed no mixing in the I

vessel cold leg to the use of multi-dimensional digital computer codes.

FLOW-2D, a two-dimensional code, was developed and benchmarked to the CREARE 1/5 test facility data.

FLOW-2D modeled the vessel cold leg and upper down-comer, including internal vent valves.

The FLOW-2D benchmarked results indi-cate that the code will provide conservative mixing results for use in any future thermal shock analysis.

In addition, a three-dimensional COMMIX code mixing analysis was performed to provide realistic results to the obvious three-dimensional flow geometry of the vessel cold leg and downcomer. Both codes predict the vigorous mixing in the ccid leg and ensure that the cold HPI water does not enter the vessel downcomer.

Forced and free convection heat transfer models were applied to I

calculate the maximum wall film heat transfer coefficient during natural or forced circulation.

4-1 I

I 4.1.

COMMIX Mixing Analysis 2,3 Early in the SB LOCA transient, the loop and vent valve flow rates were suf-f ic ient to promote near perfect mixing in the cold leg and reactor vessel downcomer. During this phase of the transient, fluid temperatures and veloc-it les obtained from the CRAFT 2 analysis were used to calculate boundary con-ditions for the fracture mechanics calculations. As the transient progressed, vent valve flow decreased, loop flow was interrupted, and the cold leg and downcomer mixing processes began to depend largely on bouyancy-driven phenom-Multi-dimensional analyses were required to determine the cold leg and ena.

downcomer temperatures and velocity field during this phase of the transient.

The B&W version of COMMIX-1A was used to perform quasi-steady-state mixing g

analyses during this latter phase of the SB LOCA transient.3 E

4.1.1.

The COMMIX Computer Program The COMM1X-1A computer program developed at Argonne National Laboratory for analysis of LMFBR components, was adapted for use in calculating fluid tem-perature and velocity fields in the cold leg and reactor vessel downcomer.2 COMM1X, a three-dimensional computer code, solves the transient continuity, momentum, and energy equations as a boundary-value problem in space and an m

initial value problem in time.

Steady-state solutions, such as those required for the thermal mixing calculations, are obtained by imposing constant bound-ary conditions for vent valve and HPI flows and integrating the transient governing equations to a converged solution.

The governing equations that are solved by the COMMIX computer code were de-rived under quasi-continuum assumptions. That is, the equations account for the presence of fixed structures that do not conform to a global (cartesian or cylindrical) grid system. These structures are represented by specifying cell volume porosity and permeability for each of the six cell faces.

The cartesian forms of the governing equations are as follows:

v 3t + 3x (Y pu) + 3y (y ov) + az (y ow) = 0; (4-1)

E Y

x y

z g

Y

"+

(Y ou ) +

(y puv) +

(y puw) = -Y

+ Y og 2

v 3t x

y y x 2

3u 2

3 u' 2

l

+u y

+Y 3u+Y g

2 (4-2) z I

4-2

2 Y

+

(y ovu) +

(y ov ) +

(y ovW) " -y

+ y Cg v 3t x

g y

y y 3 y' j

2 2

2

+u y

+Y 3,+y 3y (4-3) 32 32 z n2 2

+ y og yv

+

(y pwu) +

(y owv) +

(y ow ) " -y y

g x

y g

y

+u y

+y

[+y (4-4) y

+

(y ouh) +

(y ovh) +

(y owh) y k

=

3 g

3 -

T' 3'

(4-5)

_ 3y,k 3y.

3z,k a z, y" = volume porosity, the ratio of volume occupied by fluid to the total computational cell volume, y = surface permeability, the ratio of flow area to total cell face area in the 1 (x, y, or z) direction, p = fluid density, h = fluid enthalpy.

T = fluid temperature, u = fit.id viscosity, k = fluid conductivity, u,

v, w = fluid velocity components in the x, y, and z directions, respectively.

Equations 4-1 through 4-5 are discretized with respect to an Eulerian grid of fixed cells.

Pressure, energy, and density are computed at cell-centered lo-cations, while the velocities are centered on the cell faces. The cylindrical I

forms of equations 4-1 through 4-5 are obtained by:

1.

Replacing x, y, z with r, 0, z, respectively.

2.

Using surface permeabilities to modify the r-component of divergence terms.

3.

Scaling derivatives with respect to e by the radius.

4.

Adding the centrifugal acceleration terms to the r and momentum equations.

The resultant set of equations is solved as follows:

I 4-3

I 1.

The explicit finite dif ference approximations of equations 4-2 through 4-4 are used to advance velocities.

2.

The continuity equation is iterated to convergence by adjusting the cell-centered pressures, recalculating the three velocity components and solving the finite difference form of equation 4-1 for the mass residual.

3.

The finite dif ference form of the energy equation (4-5) is ad-vanced explicitly.

In this step, the updated value of the cell-centered enthalpy is calculated.

This sequence of calculations is repeated at each time step until converged steady-state desnity, energy, and velocity fields are obtained.

4.1.2.

Modeling Strategy The approach used in developing the COMMIX input model was selectecd based on B&W experience in modeling thermal mixing in Oconee 1 and the ARC Mixing Test Facility, and observations reported by CREARE during the course of EPRI-funded experimental work.

The input model was designed to be a realistic three-dimensional representation of the TMI-1 vessel cold leg and downcomer. Where compt omises were necessary because of geometric considerations, a conservative approach has been employed.

The model used in the COMMIX calculations was a representation of the discharge run of one cold leg and the associated quarter-symmetric section of the reactor vessel downcomer. The cold leg represented in the TMI-1 mixing calculations was selected on the basis of its long horizo. ital run.

This strategy was used to conservatively predict the vent valve fluid penetration to the point of HPI injection.

The coordinate system used in the COMMIX model was a vertically oriented right circular cylinder (Figure 4-1).

This geometry was selected to enchance the physical modeling of the reactor vessel downcomer.

The cold leg geometry was a

preserved using input volume porosities and surface permeabilities.

The HPI nozzle and vent valves were modeled using specified inflow boundary conditions. Flow rates and fluid temperatures for these COMMIX model inputs were obtained from the CRAFT 2 analysis.

A continuous momentum outlet boundary condition was used to model flow at the lower extremity of the downcomer madel.

4-4 Y

a

The COMMIX model for the TMI-1 mixing analysis is illustrated in Figures 4-1 and 4-2.

4.1. 3.

COMMIX Model Results The COMMIX program was executed using vent valve and HP1 flow boundary condi-tions obtained from the CRAFT 2 system simulation. The vessel downcomer azimuthal bulk fluid temperature profile was then obtained from COMMIX.

The fluid state was initialized to the temperature of the HPI fluid, and veloci-t ies were initialized to zero. The field equations were integrated in time until a converged steady state was obtained.

Mixing calculations were performed using CRAFT 2 boundary conditions at 60, 120, and 180 minutes af ter transient initiation.

The boundary conditions used in these cases are presented in Table 4-1.

In all cases, hot vent valve fluid penetrated the cold leg and mixed with the stream of cold HPI fluid in the cold leg piping.

The mixed stream entering the reactor vessel downcomer was significantly warmer than the injected HPI fluid.

The r-z velocity field I

at an azimuthal location close to the cold leg centerline is presented in Figure 4-3 for the 60-minute calculation.

At 60 minutes, the 18% density ratio resulted in penetration of the vent valve fluid to the point of in-jection.

At 120 and 180 minutes, the density ratios were 12 and 7% respec-tively.

For these cases, the penetration of vent valve fluid in the cold leg was less pronounced since the core outlet and vent valve temperature decrease with time (Figure 3-25).

This is reflected in the gradual decrease in the temperature of fluid entering the reactor vessel downcomer from the cold leg. The bulk fluid temperatures versus times below the cold leg nozzle for the location of the WF-70 and WF-8 welds are shown in Figure 4-4.

These temperature profiles were used in the thermal model for the SB LOCA transient to obtain the vessel wall temperature and thermal gradients through the vessel wall (section 5).

4.1.4.

Qualification of COMMIX Mixing Code To evaluate the accuracy of the COMMIX computer program, it was benchmarked with data obtained from the B&W Alliance Research Center (ARC) Thermal Shock Mixing Test Fac ility.3 The ARC facility was a 1/10-scale plastic test section with a rectangular tee configuration. The vertical and horizontal legs of the facili-ty (illustrated in Figure 4-5) represented the reactor vessel downcomer and the 4-5

cold leg, respectively.

The test conditions for the three experimental runs are presented in Table 4-2.

The COMMIX model of the ARC test facility is illustrated in Figure 4-6.

The solution was performed over a 1x 12 x 22 grid system. The horizontal and vertical legs had 70 and 110 cells, respectively.

The concepts of volume porosity and surf ace permeability were employed to account for the presence of the perforated plates at the entrance of the cold section. Comparisons of the COMMIX results and test data, illustrated in Figures 4-7 and 4-8 (vertical and horizontal legs), demonstrate the ability of the COMMIX computer program to provide a satisfactory representation of the mixing processes that occur in the cold leg and the downcomer.

In other related work, the COMMIX program was checked by comparison with exact solutions for thermally stratified developing duct flows.

The velocity profiles calculated by COMMIX were in excellent agreement with the exact solution. The calculated temperature profiles agreed well with the analytical solution at Reynolds numbers on the order of those encountered in thermal mixing analyses.

These duct flow calculations were performed using Cartesian and cylindrical coordinates to validate the cold leg modeling scheme used in thermal mixing E

analyses.

4.2.

FLOW-2D, Two-Dimensional Code Simulation A second independent analytical model was also used to evaluate the mixing processes in the upper vessel and vessel inlet piping. The FLOW-2D program (derived from SOLA-V0F') was applied by Flow Science, Inc., to perform studies of the mixing behavior using two-dimensional computing regions. These sutdies were concentrated on analyzing the fluid mixing expected to occur in the vessel inlet piping between the HPI point and the inlet nozzle.

The FLOW-2D analyses also confirmed that a significant flow of vent value fluid would penetrate the cold leg and would mix with the high-energy HPI stream near the injec tion site.

The calculations showed that the cooler mixed fluid would flow to the vessel, countercurrent to the vent valve fluid at the top of the pipe, with additional temperature dif fusion between the two streams.

Scoping studies were performed with this two-dimensional code to isolate the effects of gravi-l tational and turbulent exchange terms. The most significant conclusion drawn from the FLOW-2D studies was that the fluid mixing processes in the vessel 4-6

inlet piping will be suf ficient to raise the temperatures of flows into the vessel substantially above the HPI temperature.

4.2.1.

FLOW-2D Mixing Model FLOW-2D is an extension of the SOIA-V0F solution algorithm, which has performed well for a wide class of fluf > dynamics problems.

FLOW-2D was developed to solve the time-dependent, incompressible flow of a Navier-Stokes fluid contain-ing f ree surfaces or involving flow of two immiscible fluids.

For this appli-cation the two fluids were cold and hot water. The method used for interface treatment is referred to as the volume of fluid method; it utilizes a function whose value is unity in a cell filled with fluid No. 1 (cold water) and zero in a cell filled with fluid No. 2 (hot water). The average value of this function (F) in a cell then represents the fractional volume of the cell occupied by cold water. To avoid numerical smoothing of F, a special advection algorithm is used.

The governing differential equations for the fluid dynamics are the Navier-Stokes momentum equations:

=-f

+g

$+3

+C (4-6)

+u

+v

+v x

3 and

=-f+g 3}+y}+h (4-7)

+u

+v

+v y

Fluid pressure is denoted by p.

Velocity components (u,v) are in the Cartesian coordinate directions (x,y) or axisymmetric coordinate directions (r,z).

The l

choice of coordinate systems is controlled by the value of (, where ( = 0 cor-i responds to Cartesian and ( = 1 to axisymmetric geometry.

Body accelerations l

l

% denoted by (g,g ), v is the coefficient of kinematic viscosity, and o is the fluid density.

The fluid parcel condition of constant density is expressed by the mass con-l tinuity equation b + b + " = 0.

(4-8) i i

3x By x

l ll B

Equations 4-6, 4-7, and 4-8 are discretized with respect to an Eulerian grid I

of fixed rectangular cells.

Grid cells may have variable sizes, say 6x for 1

i 4-7

the ith column and $y for the j th row, as shown schematically in Figure 4-9.

Dependent variables are located at the staggered grid locations indicated for typical cell in Figure 4-10.

The basic procedure for advancing a solution a

through one increment in t ime (5t) consists of three steps:

1.

Explicit finite-dif ference approximations of equations 4-6 and 4-7 are used to compute first guesses for the new time level velocities.

In this step the initial dependent variable values, or the values from the previous time level, are used to evaluate all advective, pressure, and viscous accel-erations.

2.

Pressures are iteratively adjusted in each cell to setisfy the continuity equa t ion (equation 4-8).

As each pressure value is changed, the veloci-g ties dependent on this pressure are also changed. This pressure iteration W

is continued until equation 4-8 is satisfied to a prespecified level of accuracy.

3.

The F function (see equation 4-9) defining fluid regions is updated to give the new fluid configuration.

Af ter all necessary bookkeeping adjust-ments are completed (including data output) this three-step process can be restarted for the next time level calculation.

At each step, of course, suitable boundary conditions must be imposed.

The actual finite-difference approximations used in FLOW-2D for equations 4-6/7 and 4-8 are not a crucial part of the algorithm. That is, various ap-proximations could be used without af fecting the basic solution procedure.

The particular approximations used are essentially those contained in refer-ence 23.

This flexibility does not apply, however, to the way the F distri-bution is advanced in time.

Because F is a scalar quantity fixed in the fluid, its evolution is governed by pure advection in the absence of turbu-g lence.

E E + 1_ 3rFu, 3Fv =0 (4-9) 3t r 3x 3y where r = x when E = 1 and r = 1 when E = 0.

I FLOW-2D employs a type of donor-acceptor fluxing to advance the values of F in each cell. The amount of F crossing a cell boundary depends on the dis-tribution of F in the donor cell. Generally, the donor cell distribution is established from a surface orientation algorithm.

However, if the acceptor 4-8

cell is empty of i or if the cell upstream of the donor cell is empty, then the acceptor cell determines the donor cell distribution regardless of the surfnce orientation. This means that a donor cell must fill before any F fluid can enter a downstream empty cell.

To calculate the effects of turbulent mass dif fusion, equation 4-9 becomes

= 1 ---

E+1

"+

3t r ax By r ax 3 x,

+ 3y--

v (4-10) rv By, where v is the turbulent kinematic viscosity, which is assumed to have the same value for mass and momentum diffusion.

4.2.2.

FLOW-2D Mixing Model Results Using the FLOW-2D code with the vessel inlet cold leg geometry and with the flow conditions existing in the SB LOCA transient, results with and without gravitational effects were obtained. These conditions consisted of HPI flow only (since natural circulation had ceased) and were modeled by the computa-tional mesh shown in Figure 4-11, giving the following results for the generic SBLOCA transient analyzed and reported in the Duke Power Co. Oconee 1 anal-ysis (dated 1/15/82). Without gravitational ef fects there are no forces acting I

to move the lower density hot water up the cold leg piping over the higher density cold HPI water. The velocity field (Figure 4-12) shows no penetration of the hot water into the cold leg piping and no mixing of the vent valve ar.d HP1 flow streams in the case of no gravitational af fects.

However, the flow is very dif ferent when buoyant forces are allowed to act, and both experimental and theoretical investigations have demonstrated that the hot water will penetrate the inlet pipe.

Using the sarre computational mesh, with both gravitational effects and turbulent mixing allowed to occur in the cold leg piping, the velocity field of Figure 4-13 and temperature contours of Figure 4-14 result. The velocity field clearly shows that the vent valve hot water flows up the cold leg piping; indeed, it flows to the point of injection early in the transient. At this point, 20 minutes into the SB LOCA transient, due to the high turbulence at the point of injection the flow is no longer a stratified flow, but one that is well mixed. The temperature contours, Figures 4-14 and 4-15, show the mixing effectiveness obtained when gravitational effects are considered and turbulent mixing as a result of the HPI flow injection is modeled. The HPI fluid is injected into 4-9

Ii the cold leg piping 150 in. upstream of the vessel downcomer.

This substan-tiates that the cold ilPI fluid does not flow into and down the reactor vessel wall.

The temperature results (Figure 4-15) at 20 minutes into the SB LOCA transient show the fluid entering the downcomer of the reactor vessel to be greater than 286F. This analysis was repeated at 90 and 180 minutes into the i

SB LOCA transient. While density dif ferences between the hot vent valve fluid and the cold IIPI fluid decrease from 20% at 20 minutes to 5.2% at 180 minutes, and the hot vent valve fluid is not driven as far into the cold leg piping at 180 minutes into the transient, the turbulent mixing that occurs ensures that the cold IIPI does not directly enter the downcomer of the vessel. The bulk fluid temperatures at the inlet to the vessel downcomer (point A in Figure 4-15) and 8 inches down into the downcomer (point B in Figure 4-15) are shown in Figure 4-16 for 90 and 180 minutes of the SB LOCA transient. The mitigation of vessel downcomer temperature versus time is evident (from Figure 4-16) when compared to the conservat ive assumpt ion, 50F, used in past thermal shock anal-

=

yses.

These FLOW-2D results have been used in the evaluation of the vessel inlet nozzle in this report. The FLOW-2D model has been benchmarked to the test g

data obtained from the one-fif th scale model CREARE test f ac il ity. The data E

were obtained as part of an EPRI-sponsored research program in the summer of 1981.2' 4.2.3.

Overcooling Transient M ix ing, I

During the overcooling transient analysis, various plant condit ions were simu-la ted.

This simulat ion considered that RC pumps were tripped by the operator during the t rans ien t after the initiation of the ESFAS. When the pumps were l

tripped, natural circulation continued. Therefore, the thermal analysis as-sumed complete mixing of the IIPI flow and natural circulation flow.

I l

4-10 I

Table 4-1.

COMMIX Input Data for TMI-1 SB LOCA Analysis Time into transient, minutes Parameter 60 120 180 Vent valve flow, Ib/s 956 736 482 Vent valve temp, F 461 363 276 HPI flow, Ib/s 117 117 110 HPI temp, F 50 50 50 RCS loop flow 0

0 0

RCS pressure, psia 762 309 188 I

Table 4-2.

Test Conditions for ARC Mixing and Visualization Studies Test Test Test I

Parameter No. 1 No. 2 No. 3 Qhpi, 8Pm 5.29 13.23 13.23 V

ft/s 0.04 0.1 0.1

hpi, Q

, gpm 13.73 13.73 59.51 V

ft/s 0.3 0.3 1.3 Qyy/Qhpi 2.6 1.04 4.5 FR 0.07 0.17 0.17 cold l

Fgg 1.1 1.1 4.8 Ao/o 0.028 0.028 0.028

(#

l Fr = (Q

/A )/[gD (Ap/p )).

4-11 lI

l Figure 4-1.

COMMIX-1A Computational Mesh for INI-l Cold Leg and Vessel Downcomer l

VENT VALVES RC PUNP ZI OlSCHARGE j

/

/

p' ll..

j U-BAFFLE e'

/

, f COLD LEG PIPING VESSEL 00nNCOME HIGH PRESSURE INJECTION 1

I i

l VENT l VALVES I

18 K

h r1 m

m m

m m

m m

m m

m m

m m

m m

m M

M M

Figure 4-2.

Location of TMI-1 Vessel Welds in 0, Z

=

Coordinates for SBLOCA Transient I

r VENT VALVES s F

IN

=

PLANE l=2

' COLD 0

LN BLOCKAGE

-g I

I I

WF-70 207.5" WF-8 I

WF-25 l

ll SA-1526 h

9

-30 0

60 I

-44. 7" 0"

89.4" Distance From Q of Cold Leg 4-13

I Figure 4-3.

Velocity Profile Along J = 6 f2 4E rs ',-

l6 y y W

f * 'l

,\\\\

=3

)

~

r *#,/

=

_~

a.

y

=

\\,\\

-y y,g s -

ff U

COLO LEG PIPING y

)\\

VESSEL DOWNCOMER l

I I

1r

)

I u

J=6 I

I 1

l I

l I

l l

k-[k

I Figure 4-4.

Bulk Fluid Temperature From COMMIX Mixing I

Model for TMI-1 SBLOCA Transient I

i

)

i I

DOWNCOMER BULK TEMPERATURE 500 N\\

(FROM CRAFT)

I O

a$

E I

s WF-8

'sh 400 I

e N

I WF-70 300 x

I

\\ N N N 200 I

0 20 40 60 80 100 120 140 160 180 Time, Min I

lI 1

I I

'I 4-15 l

- - - ~ ~ -

w

I Figure 4-5.

ARC Mixing Test Section

{

0.98*

E T

g I

REFERENCE WALL g

g STATION 2 STATION 3 STATION 4 V. D STATION I t

r 2.8' 3.85=

T y

v. D T ___

l n

4 e

IW STAil0h 5 i.84*

I v.T 3.78' j

l0.406" REFERENCE WALL

~

6.81' n

STATION e

• i'g.

l

y. T W

4*

STATION 7 V. T e

V - VELOCITY ME ASUREMENT T - TEMPERA %RE HEASUREMEhT D - DYE INJECTION POINT I

I I

I 4-16

5 1

Figure 4-6.

COMMIX-1A Nodalization Scheme for the 1l B&W Test Facility Mixing Tests 1

t 3

)

!I l

'l

)

!I

.i I

l I

l.

l l

l i

I

(

,.N.

l i

j I

I I

l.

l l

u....

.i 1

}

....I J

s l

..MMW 4

4

..N 4

1 1

s

..N.

I i l lI n

i OJ 4-17 l

J

I Figure 4-7.

COMMIX-1A Temperature Profile - ARC Test Run 2 l

_ _ _ _ DATA u

CCMulI 190 STATION 7 TEMPERATURES i

l g

.;y u

T I

180 s.s 1.70

~

37,7,,,

160 iI 150

.NT 140

,,/N s..

-s 130

's '

m s

s J

120

\\,

\\

l 2

110

\\~

g N

e s

0 100

"\\

g.,

l

'\\

90 Ng u

B0 70 60 50 40 l

O. 2 0.4 0.6 0.8 1.0 l

Distance From Reference Wall (Vertical Leg), in.

4-18 i

N

a -..

e'a

I Figure 4-8.

COMMIX-1A Temperature Profile - ARC Test Run 2 i

0ATA g

COMMIX STATION :

,Z

'f 180

L o. uos-

'I g

170

-p.,

I

• N ll 160 l

'q.---}--.4 150

/

!l l

140 p

/

.. 1 130 I

STATION 2 TEMPERATURES m

I u

lE 3

120

/

/

.g m

k 110 I

I l

~

100

/

/

I j

90 l

80 I

tl "I

~

10

(

/

60

[}

/

50

/

ll

--r' 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.3 Distance tram Reference Wall (Horizontal Leg). in tI.

I 4-19 I

Figure 4-9.

Finite Difference Mesh With Variable Rectangular Cells l

s, 1==-

l h

t 6

11 i

r l

I Figure 4-10.

Location of Variables in Typical Mesh Cell V

l i.J+

=

,r P.

I I.

3 U;+j.j U i"~*J o

F;,;

6y; O

l I

=

i.j-6i H l

I I

4-20

I Figure 4-11. FL0k'-2D Computational Mesh for Cold Leg Inlet Piping (150 in.) Region I

I

?

I I

-7..

,,em I

I I

I t

1I

I I

I 4-21 I

I i

Figure 4-12.

Velocity Field and Interface Between Vent Valve and HPI Streams in Absence of Gravitational (Buoyancy) Effects and Natural Circulation VENT VALVE FLOW or

=

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

HPI FLOW be.........

j n

,n a w W

n*

N'use su s.

Inulet D

teW Wr l

>,w

,e w gi$*

Ia Wei' 5

M se" 38 I

I I

4-22

I Figure 4-13.

Velocity Fields With Gravitational Effects and Turbulence VENT VALVE FLOW I

U I

3"" ".......................

HPI Ft.0W I

INJECT 10N g

I n N eeeee e.

e a

D D.88g M

I I

\\\\\\\\\\1 I

" \\\\\\1 I

..,,gg.....

.., i um - - -

I

..s,s_______

I g

I

• • • 1. //f gIC./ f.* f l*

0//////// /#

1/ //r ' ' *

• I 4-23 I

4

I I

Figure 4-14.

Temperature Contours Associated tJith Velocity Field in Figure 4-13 I

H=544 F HIGH j

W I

x, f

I t = n.,

I I

I I

I I

I I

I I

I I

4-24

!I

,g Figure 4-15.

Temperature Contours Acros,a Downcomer 3

at Lower Lip of Inlet Pipe

/

g 286 F

[ 297 f Core 351 F

• -- Ves sel Barrel Wall

/

g 315'F 354*F 333*F i

n i

g I

/

,I j

( 20 minutes into the SBLOCA transient )

Note: Temperature is 286F at vessel wall and 354F at core barrel.

I I

I I

I 4-25 I

I Figure 4-16.

Bulk Fluid Temperature From Flow-2D Mixing Model for SBLOCA Transient 600 I

550 DATA FROM CRAFT ANALYSIS 500 Point A on Figure 4-15 at

'e<,sel Inlet l

Point B on Figure 4-15 at

.a "assel Downcomer m

400 g

5 3

3 S

E i

a0 300 POINT "B" I

N I

N 200

\\

POINT " A" N

100 50 0

20 40 60 80 100 120 140 160 180 Time, minutes I

I l

I I

4-26 l

I I

I 5.

VESSEL WALL THERMAL ANALYSIS Once the transients for thermal shock analysis were selected and the RCS pa-rameters were determined using the methods described in section 3, the results were used in section 4 to determine the vessel downcomer fluid temperature I

versus time.

The next step is to calculate both the radial and azimuthal wall temperature profiles as described in this section.

This requires the determination of the fluid-to-vessel wall heat transfer co-ef ficient and the thermal gradient through the vessel wall. The vessel ther-mal conditions for normal and abnormal operation are illustrated in Figure 5-1.

5.1.

Film Heat Transfer Coefficient and Azimuthal Temperature Distribution The heat transfer coefficient (HTC) consists of the wall film coefficient de-termined as the larger of (turbulent) the forced or free convection coeffi-cient. The HTC for both transients consists of only the film coefficient since the model used to calculate the vessel gradient for these transients included the cladding.

5.1.1.

Forced and Free Convection HTC In selecting ar appropriate convective HTC, the surface heat transfer via forced convection, free convection, or some combination of the modes must be considered.

ine dominant mode of convective heat transfer at the vessel wall will depend on local conditions, which may vary considerably.

Because of the uncertainty associated with accurate determination of the local heat transfer regime, a conservative approach was taken. Heat transfer coefficients appli-cable to both forced and free convection for turbulent flows were calculated at each location on the vessel wall. The higher of the two coefficients was used to calculate the surface heat t' lux.

I I

5-1

I The HTC for forced convection was determined by the Dittus-Boelter equation *S:

h

~

" ~

~

forced where 2

K = fluid thermal conductivity, Btu /h-ft _.F, D = hydraulic diameter, ft (D % 2W, where W = downcomer width),

Re = Reynolds number, the ratio of inertial to viscous forces, Re = VD/v, V = fluid velocity, fps, 2

v = kinematic viscosity, ft /s, Pr = Prandtl number, the ratio of storage to conduction energy E5 transfer, Pr = C u/K, p

specific heat, Btu /lbm-F, C

=

p u = dynamic viscosity, 1bm/h-ft.

2s The free (or natural) convective HTC is given by h

= 0.094 K/L (Gr Pr)l/ 3 (5-2) free where L = heated length, ft, Gr = Grashof number, the ratio of buoyant to inertial force, l

3 2

Gr = g8ATL /v,

g = gravitational acceleration, 32 ft/s2, B = fluid thermal expansivity, S = 1/p 3p/3T (1/F),

AT = governing temperature difference, wall to fluid, F.

Note that the HTC for turbulent free convection is independent of heated length.

The film HTC is calculated for the region below the vessel inlet nozzle at the locations of tFe various welds. The HTC at each mesh point may switch from forced to free convection at any time in the transient. However, the variation in the magnitudes of the coef ficients among the different locations is not great at any given time.

The SB LOCA HTC data shown in Figure 5-2 are typical of all mesh points in the l

TMI-1 vessel.

It is evident from the figure that free convection governs the heat transfer over all but the first 20 minutes of the SB LOCA transient.

l The resultant film HTC for the overcooling transient, which consists of all l

turbine bypass valves failing open on reactor trip and the main feedwater sys-tem continuing to feed the steam generators until their pressure falls below 5-2

I 600 psig, is shown in Figure 5-3.

This transient has the RC pumps tripped but maintains natural circulation in the primary loops.

Since loop flow is main-tained at all times, the heat transfer rate to the vessel is controlled by the I

forced heat convection coefficient.

The film HTC for the vessel inlet nozzle was determined as the larger of the I

two coefficients obtained from a flat plate or circular pipe consideration for the SBLOCA transient. The coefficient varied from N1500 Btu /h-ft2_oF at 2

at 2 minutes to %50 Btu /h-ft _ F at 180 minutes in the transients.

For the overcooling transient, the HTC used for the vessel nozzle analysis I

varied from 1000 Btu /h-ft

• F at the beginning of the transient to 235 Btu /h-2 ft

• F after 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />.

5.1.2.

Base Metal Azimuthal Temperature Distribution The base metal azimuthal temperature profile at the location of the WF-70 weld (65 inches below the nozzle outlet) was assumed to be constant since only a small bulk temperature profile was calculated for the SBLOCA transient. The base metal temperature is for the vessel metal immediately beneath the cladding.

The cooldown rate at the base metal-to-cladding interface for the WF-70 weld location is 1.l*F/m,inute (Figure 5-4) compared to the core outlet temperature cooldown rate of 1.7*F/ minute. Thus, the cooldown rate is such that the tem-perature at the vessel base metal is always higher than the bulk fluid tem-perature at the same point in time.

The bulk fluid cooldown rate for this SBLOCA is about the same as the normal designed heatup and cooldown rate of 100*F/ hour.

Since WF-70 is located directly below the cold leg, it experiences I

the coldest temperature of all welds.

For the overcooling transient, homogeneous mixing in the vessel downcomer was I

used with the coldest inlet temperature determined from the overcooling tran-sient analysis. The homogeneous flow results in all locations below the ves-sel inlet nozzle being at the same temperature; therefore, no azimuthal tem-perature profile exists as in the SBLOCA transient.

The vessel base metal wall temperature distribution versus time for the over-cooling transient is shown in Figure 5-5.

The cooldown rate for the vessel base metal surface is 10.8*F/ minute as compared to the vessel inlet tempera-ture cooldown rate of 20.2*F/ minute for the first 15 minutes of the transient.

After 20 minutes, the vessel inner surface begins to heat up for the next 40 5-3 I

I minutes of the transient. After 60 minutes, the inner surface remains con-stant at 426F for the duration of the transient.

5.1.3.

Vessel Inlet Nozzle Inner Surface Profile The FLOW-2D results, described in section 4.2.2, are used to evaluate the thermal conditions existing in the vessel inlet nozzle during the SBLOCA tran-sient. The resultant nozzle base metal temperature versus time for the " knee" of the nozzle is shown in Figure 5-6.

The FLOW-2D bulk fluid temperature at the nozzle wall, used as input to the finite element program to generate the thermal profiles, is shown for comparison.

For the overcooling transient, homogeneous mixing, using the coldest inlet temperature determined from the overcooling transient analysis, was used for the nozzle thermal calculations. These are the same data used in evaluating the vessel wall welds during this overcooling transient.

5.2.

Wall Thermal Gradient Determination 5.2.1.

SBLOCA Transient When no cold leg flow exists, normal mixing of the cold HPI flow with the usual hot fluid flowing in the cold leg of the RCS occurs by a different pro-cess.

Where this situation exists for extended periods of time, it has been assumed in previous studies that the cold HP1 fluid would begin to flow down the vessel wall, thereby causing a drastic reduction in the vessel wall tem-perature below the cold leg nozzle.

The results from sections 4 and 5.1 were used to evaluate the actual mixing under these conditions and to determine the base metal wall azimuthal temperature distribution versus time for the spe-cific SBLOCA transient.

Since large gradients had been calculated to exist in the surface of the base metal, and vessel welds are located at varying distances below and to the side of the cold leg nozzle, a three-dimensional finite element model had been used in past analyses to accurately determine B

the temperature gradient through the vessel wall as a function of time at the W

various weld locations. However, the COMMIX results showed very little tem-perature gradient in the axial direction; therefore, an r-0, two-dimensional heat transfer model was used in this analysis. The two-dimensional heat trans-5 fer analysis (using the ANSYS code ) determines nodal temperatures through the vessel wall with additional nodes that coincide with the critical pressure I

5-4 I

I evaluation nodes of 1/40T to 30/40T.

In the radial and azimuthal directions, the nodes coincide with the locations of all longitudinal and circumferential welds within the beltline region of the TMI-1 vessel, as shown in Figure 5-7.

The wall temperature profile for the WF-70 circumferential weld is shown in Figure 5-8.

As a result of the significant mixing occurring in the cold leg I

and upper downcomer region, the various weld locations do not experience the vastly different thermal gradients shown by past conservative analyses. Thus, any weld not directly under the cold leg will experience approximately the same thermal gradient, but it will be smaller in magnitude than for the WF-70 weld.

5.2.2.

Overcooling Transient The wall thermal gradient of the overcooling transient was calculated in a manner similar to the SBLOCA analysis discussed in section 5.2.1, except that a two-dimensional axisymmetric heat transfer model and appropriate film HTC were used as explained in section 5.1.1.

The point of injection is highly turbulent, and the flow enters the downcomer of the vessel as equilibrium or homogeneous flow.

Since the flow is not stratified but equal to the bulk fluid temperature at all azimuthal weld locations, a two-dimensional axisym-5 metric heat transfer analysis, using the ANSYS code, was used to determine the thermal gradient through the vessel wdl.

Node points through the wall I

were selected to correspond to the critical pressure evaluation nodes.

Figure 5-9 shows the wall temperature profile for the overcooling transient.

It is obvious that large wall thermal gradients are induced in the vessel that are larger than the SBLOCA transient profile shown in Figure 5-8.

Thus, I

the SBLOCA transient becomes less severe than this overcooling transient.

An additional difference in the transient is that the vessel inner and outer I

walls for the SBLOCA transient experience only cooling, whereas the overcool-ing transient cools and then begins a heatup and repressurization as the tran-sient continues.

5.3.

Inlet Nozzle Temperature Gradient Determination Since the cooldown rates are higher than design basis, a single check was made on the inlet nozzle to determine the thermal gradient through the " knee" of the nozzle. The procedure for determining the temperature gradient through I

5-5 I

I the thickness of the inlet nozzle, involved the use of a two-dimensional axi-symmetric finite element model (shown in Figure 5-10) with the ANSYS code.

The output consists of detailed stress and thermal profiles.

The stress pro-file, due to pressure through the " knee" of the nozzle, is used in conjunction 27 with the BIGIF code to generate a stress intensity factor (SIF) due to pres-sure.

The vessel inlet nozzle temperature profiles at the nozzle knee are shown in Figure 5-11 for various times during the SBLOCA transient. Large ".hermal gradients through the nozzle wall are experienced during the SBLOCA transient with the maximum AT existing at 25 minutes into the transient. These large 5

thermal gradients exist for about an hour, beginning around 20 minutes and decreasing rapidly at about 90 minutes into the transient.

The same homogeneous mixing data obtained from the overcooling transient anal-ysis for the vessel wall are used in evaluating the nozzle analysis. The noz-zie temperature profiles at the knee are shown in Figure 5-12 for various times during the overcooling transient. The nozzle temperature profiles, when com-pared to the vessel wall temperature profiles, exhibit similar trends but with larger temperature gradients. The inner surface in both cases reaches a mini-mum temperature in approximately 20 minutes and then begins to increase because of the characteristics of this overcooling transient. The transient, as simu-lated, begins repressurization at approximately 6 minutes and a heatup approx-imately 15 minutes into the transient. Therefore, the inner surface tempera-ture began to increase while the outer surface was still cooling. This caused l

a rapidly decreasing temperature gradient through the nozzle wall at this time.

l I

l I

I 5-6 I

I I

Figure 5-1.

Thermal Gradient in Reactor Vessel During Normal and Abnormal Conditions STAINLESS REACTOR STEEL I

COOLANT CLADDING FLOW I

555 F

/

~555'F I

(

/

'" 555'F 5 5 5 *."

/

/

I y

REACTOR INSIDE

/

VESSEL WALL OUTSIDE I

/

/

8.4"

/

(NOT TO SCALE)

/

I NORMAL TEMPERATURE DISTRIBUTION, STEADY STATE STAINLESS COOLANT CL 0 ING REACTOR VESSEL WALL I

r

/

P 550'F

/

/

INSIDE

/

/

420*F OUTSIDE 350*F y

/

g,4n

.19

1. :(NOT TO SCALE)

I ABNORMAL TEMPERATURE DISTRIBUTION EXISTING DURING OVERC00 LING TRANSIENT I

5-7

r l

I Figure 5-2.

Wall Convective Heat Transfer Coefficient (h)

Vs ~_.ansient Time for SBLOCA Transient l

P 5000 I

~

t T

( 2000 FORCEO 3

~

1000 c

3 500 C

]

FREE

1. ~~~

200

/

-~

s 100 hforced=0.0231Re0 0. 8 0

Pr.4 0

5 50 hfree = 0.094 1 (Grt Pr)1/3 L

2 20

.018 FT PZR BREAK 0

1 1

.5 1

2 5

10 20 50 100 Transient Time, minutes I

I l

l I

I I

I I

5-8 I

'I f

I Figure 5-3.

Wall Convective Heat Transfer Coefficient Vs i

Time for Overcooling Transient RC PUMPS TRIPPED I

d 2000 E

'= 1000 j 500 I

G_

$ 200

=

100 I

3 50 I

z t,

I 5

u I

i 1

l g

.5 1

2 5

10 20 50 100 l

Time. Minutes I

I I

I I

I I

5-9

,I

I Figure 5-4.

Base Metal Temperature Vs Time for SBLOCA Transient at Location of WF-70 Weld l

600 o'

\\

WF-70

- 500 5

\\

r N

I

\\

400 N

00WNCOMER BULK FLUID 300 TEMPERATURE FROM COMMIX

\\ \\

I 250 0

20 40 60 80 100 120 140 160 180 Time, Min I

I I

l I

I I

I 5-10

lI I

Figure 5-5.

Base Metal Temperature Profile Vs Time

)

for Overcooling Transient 600 RC PUMPS TRIPPE0 i

VESSEL BASE METAL TEMPERATURE VESSEL INLET TEMPERATURE

.I soo i

i

~

\\

a

\\

l 3

400 1 l

A

\\

s 1

/

l 3

\\</

l

~

l 300

'E 200 E

20 40 60 80 100 170 140 160 180 i

Time, min.

,i l

l

!I

'I i

i!I I

5-11 il

^

I 1

Figure 5-6.

Vessel Inlet Nozzle Base Metal Temperature Vs Time for SBLOCA Transient at Knee of Nozzle I

l 550 I

500 I

400 g

.i u

a l

300 BASE METAL 200 I

FLOW 2 0 FLUl0 TEMPERATURE 100

+

a 20 40 60 80 100 120 140 160 180 Time, Minutes I

1 I

I I

I I

5-12

m M

M M

M M

M W

W W

W W

W W

.i

.l Figure 5-7.

Inside Surface of Reactor Vessel Weld Locations, TMI-l W

x y

7 w

I I

I N.V.

f LANGE MATING SURF ACE d

iL 223.5" Rff.

28" 1.D. REF.

"o

_q'4. 7 " REF.

1 E 'o c

C.F.

_' b __'

,_ >_ r -

6' 4

p s

0 4

h b

\\. p_l

[1%

(}N Ogi (y'

4 Wito g

O 1

(,1 1

0 1

~

T h

'y 342.9" REF.

N0ZZLE BELT

( WELD U

N 117.44"

<I 7

LW F -7 0 72.74",

_WF-8 M

U w

,8 SHELL w

4 WELD U

134.10" i_WF-25 l

89.4" SA-1526 g LOWER l

- 4 SHELL l

l I

536.4" l

?

I Figure 5-8, SBLOCA Transient - Wall Temperature Profiles for Location of WF-70 Weld Using C05D1IX I

600 CLAD L0N,N I

2 500 p

=

N 72 MlN E

E t

E 400 180 MIN T

300 E

250

/6

1. 0 2.0 3.0 8.4 li Distance thru vessel wall, incnes OUTSIDE SURFACE SURFACE E

I I

I, I,

I 5-14 I

M M

W W

W W

M 4

W W

M m

m Figure 5-9.

Wall Temperature Profile for overcooling Transient 600 6 MIN.

24 MIN.

500 45 MIN.

~

~

400 120 MIN.

g a

~

w

,L 5

TBVs FAILED OPEN E

u E0 300 l

200

.21" 1/4T 1/2T NI 100 e i i i iii I Iii l

l l

h 2.1 4.2 6.3 8.4 o

INNER Distance Thru Vessel Wall, inches OUTER SURFACE SURFACE

L I

I Figure 5-10.

Vessel Inlet Nozzle Finite Element Model I

_=Zs I

e Ca I

m=

I KNEE OF N0ZZLE vvv vVV I

I I

I PLANE OF N0ZZLE TEMPERATURE PROFILE I

I I

I 5-16

. = - -

1 i

i Figure 5-11.

SBLOCA Transient - Inlet Nozzle Temperature Profiles at Plane of Nozzle Knee 550 l25 MIN 500 CLAD l

[60 MIN l

400 O

.i 90 MIN b

5 u

300 y

g 180 NIN 200 I/27 I/4T e

u I

100 O

l.2 2.0g 3.166 6.33 g,gg 13.16" Distance inru Nozzle WalI. Incnes a

o Outer inner Surface Surface

Figure 5-12.

Overcooling Transient - Vessel Inlet Nozzle Base Metal Temperature Profile Vs Time at Knee of Nozzle 600 20 MIN CLAD 500 50 MIN 400 180 MIN l

E i

300 2

I 200 i

(

i 1/4T 1/2T 100 i

I e

i i

i i

I e

i n

1.2 2.06 3.766 6.33 8.89 13.16" f

Osstance intu Nozzle Wall, Incnes INNER OUTER SURFACE SURFACE I

l l

l l

I I

l 5-18

I 6.

VESSEL MATERIAL PROPERTIES 6.1.

Determination of Vessel-Specific Material Properties for Input to Linear Elastic Fracture Mechanics Analysis The material properties used as input to the linear elastic fracture mechanics (LEEM) analysis are based on the material certification and weld metal quali-fication records where available. These properties and data are available in BAW-1439, which rt.::ct-the results of the last capsule removed and tested at TMI-1,6 For those c c s where the properties were not available because they were not required at the time the TMI-1 reactor vessel was fabricated, the properties were established in accordance with the requirements of BAW-10046A, Rev. 1.7 There are two exceptions to these procedures, and because of their overall importance to this analysis, they are discussed separately.

6.1.1.

Copper and Phosphorus Contents of Welds The true copper and phosphorus contents of the weld metals were not determined as part of the weld metal qualification procedure; subsequently, a detailed study was conducted to ascertain the representative chemical composition of all core region weld metals of 177-FA vessels. This study established the chemical composition as well as the copper and phosphorus contents of the 9

deposited weld metals. These values, given in BAW-1511P, were used to estab-lish the copper and phosphorus values used in the LEFM analysis.

6.1.2.

Fluence at Weld Locations The fluence at the weld location determines the radiation damage exhibited by the specific weld metal. Thus, it is important that each weld metal be iden-tified as to the specific fluence it will experience for a given service peri-od.

This fluence is related to the location of the weld metal relative to the core region and must take into account the azimuthal and axial relation-ships of the weld metal.

Since the longitudinal weld exists in the axial flux plane, the highest flux over the weld is used in the analysis. The I

6-1

values used to determine the relative fluence for each material used in this analysis were obtained from BAW-1485." These geometry f actors are applieti to the peak fluence calculation for the specific vessel as deseirbed in :iection 7.

6.2.

Adjustment of RTg F

The RT values for the reactor vessel materials were adjusted to account for g

the ef fect of irradiation damage to both the base and weld metals in accordance with Regulatory Guide 1.99, Revision 1, using the fluence values developed as described in section 6.1.2 and the chemistry as described in section'6.1.1.

No adjustment was made to the RT of the velsel inlet nozzle materials be-g cause the predicted end-of-life fluence (s le3s thar the threshold for radia-tion damage as defined in 10 CFR 50, Appe9di,t A, Paragraph II. A.

6.3.

Upper Shelf Requirement for LEFM And va I

A definition of the upper shelf fracture toughness value2 for the LEFM anal-ysis is required because the ASME Code does not define an upper shelf value for the K urve.

For consistency with previous analyses, an upper shelf R

value of 200 ksi /E was established for the JW analysis.1 6.4.

Assumptions and/or Conservatisms in Analysis -

The assumptions used to develop the data base are included in the reference documents 6*710 and no additional assumptions were made during' development E

of the data for these analyses.

Many conservatisms were used in the input for this analysis. The copper and phosphorus contents are believed to be realistic, ar.d the variations in chem-istry that may exist are believed to be minor. The variations in chemical content will have only a negligible impact on the predicted value of the shif t in RT Likewise, the fluence value is believed to be conservative and any g.

additional refinement will have a positive impact on the overall cons 9rvstism of the analysis. Furthermore, the use of Regulatory Guide 1.99 te.prei.ict the change of RT as a function of chemistry and fluence has proven to be con-g servative as demonstrated by comparing measured irradiated properties data from 12 surveillance capsules with the predf etions based on Regulatory Guide 1.99.

Although a detailed qualitative evaluation of conservatism for these data has not been made because of insufficient data on a luantitative basis, I

6-2 1

l

I, I_

it is reasonable to assume that an overall evaluation of the input data would demonstrate that it is conservative.

6.5.

Results The results of the materials determination of vessel-specific material proper-ties are given in references 6 and 9.

The properties for all base and weld metals in the beltline region of the vessel used in the LEFM analysis are given in Table 6-1.

The properties for the vessel inlet nozzle forging used La the LEFM analysis are given in Table 6-2.

The predicted fluence for the 2

inlet nozzle at 32 EFPY is calculated to be less than 1.0 x 10 " n/cm. Since the fluence is below the threshold for analysis as defined by 10 CFR 50, Ap-pendix H. minimal irradiation damage occurs; therefore, no adjustment of the initial RT is calculated for this analysis. The chemistry of the nozzle g

forging has not been determined in the same manner as the vessel beltline ma-terials; however, since the accumulated fluence is lower than the damage thres-hold value at this position, negligible irradiation damage is expected to oc-cur during tha. life of the vessel. The results of the beltline region mate-rials evaluations are conservative, and the materials are expected to continue to behave in a predictable manner as demonstrated by the results from the RVMS capsule evaluated for TMI-1.

This behavior is further supported by the results of capsule evaluat ions from other B&W 177-FA vessels f abricated during the same period. Because the capsule is irradiated at fluence levels greater than those being experienced by the reactor vessel, the material behavior of the capsule leads that of the actual reactor vessel wall. This ensures advance knowledge of the behavior of the material properties in the reactor vessel, and in t he unlikely event that abnormal material behavior is observed, ade-quate time will exist to evaluate the impact of such property changes on re-actor vessel integrity. The materials behavior of the TMI-l reactor vessel I

will be menitored by the scheduled future evaluation of surveillance capsules.

U In addition,' the results from the integrated RVMS program for all serat-ing plants and other materials research programs will be available et evaluate and predict the future behavior of the TMI-l materials as affected by their I

service environment.

o 6-3

Table 6-1.

Material Properties Used in LEFM Analysis of TMI-l Reactor Vessel, 32 EFPY

}

Material identification ("

Chems it ry Neon fluem(c)

Initial M EFPY Weld or Cu, P.

(inside surface).

NDT' NDT' heat No.

Type Location wt I wt %

n /c:n' F

F ARY-059 SA 508, cl 2 Nozzle belt 0.08 0.006 1.30E19

(+60) 102 forging WF-70 Circumferential Nozzle belt /

1.30El9

(+20) 320 weld upper shell C2789-1 SA-302B, MOD Upper shel1 0.09 0.010 1.7E19

(+40) 118 WF-8 Longitudinal Upper shell 1.7E19

(+20) 335 weld WF-25 Circumferential Upper shell to 1.7E19

-14 301 weld lower shell SA-1526 Longitudinal Lower shell 1.4E19

(+20) 325 weld C3251-1 SA-302B, MOD Lower shell 0.11 0.012 1.7E19

(+40) 157 Atypical Circumferential Upper shell to 1.7E19

(+90)

(f) weld lower shell

(*}Per BAW-1439.6 Chemistry per BAW-1511P.'

(Weld data are proprietary.)

Predicted fluence at 32 EFPY based on existing out-in fuel management.

Estimated RT v u s per B M -100% A, Rev. 1.7 NDT Per Regulatory Guide 1.99, Rev. 1.

}Per BAW-10144A.2 s l---

Table 6-2.

Material Properties for TMI-l Reactor Vessel Inlet Nozzle g,) fluence,(b) RTET'fe)

Neutron Material identification 2

Type Location Chemist ry n/cm y

Forging, SA 508 Cold leg nozzle NA

<1.0E16

(+60) cl 2 i

(* Since end of life fluence is <10 " n/cm, no chemistry was used in 2

analyses of nozzle material.

( } Predicted end of life fluence for vessel inlet nozzle.

• Estimated RT vales per BAW-10046A, Rev. 1.7 ET I

lI I

6-5

7.

VESSEL AND WELD FLUENCE DETERMINATION 7.1.

Analytical Procedure for Fluence Determination Vessel fluence as a function of time was determined by first calculating the I

fluence for an irradiation period for which surveillance capsule dosimetry is available and then extrapolating that result to future operational conditions.

Weld fluence can then be obtained from axial and azimuthal shape factors that relate weld location to relative variations in fast flux (E > 1 Mev). Data 6

from the TMI-1E capsule fluence calculations are presented in BAW-1439, and analytical procedures for both vessel and weld fluence calculations are de-scribed in BAW-1485.10 TMI-1 vessel fluence for cycle I was calculated with an analytical model which was normalized to the capsule dosimeter measurements. The measured-to-calcu-I lated normalization factor was 1.10.6 Vessel fluence at the end of 32 EFPY was determined by extrapolating cycle 1 flux.6 Subsequently, an improved pre-l8 diction methodology, based on the proportionality of core escape flux data from fuel management PDQ calculations, has been used to predict future vessel fluence. Using this methodology, an end-of-life (32 EFPY) fluence of 1.7 x 10 " n/cm2 is predicted for the maximum fluence location on the vessel inner surface. The corresponding time-averaged fast flux is 1.68 x 10" n/cm

-s.

2 The location of maximum fluence in the vessel for an equilibrium cycle, shown in Figure 7-2, is the inside surface of the base metal at an azimuthal posi-tion approximately 11' from a core major axis and near the core midplane in elevation. Because of the power distribution in the initial cycle I core, the maximum azimuthal fluence occurred approximately 8* from a major core axis.

Vessel fluence as a function of wall penetration (Figure 7-1) is obtained from the analytical model described in BAW-1485.10 From a plot of fast flux versus wall thickness, lead factors are obtained which represent the ratios of sur-face flux-to-flux at T/4, T/2, 3T/4, and T positions, where T is the thickness of the pressure vessel (18.6 in.) The corresponding flux ratios at the above 7-1 I

wall locations are 1.8, 3.7, 7.7, and 21, respectively. These data are con-sidered to be independent of azimuthal and axial location.

Shape factors for converting maximum vessel fluence to a weld location are obtained from R9 and RZ reactor model calculations of flux distributions in the pressure vessel using the DOT 3.5 transport code. An R8 model produced relative f actors for azimuthal displacement as a function of angular deviation from 11* off a major core axis (angular position of maximum fluence), and an RZ model was used for axial displacement as a function of distance from core midplane. The relative azimuthal flux produced by the R9 model for a typical 177-FA plant is shown in Figure 7-2.

Because of symmetry in the power dis-tribution in the azimuthal direction, only one-eighth core data are shown.

The longitudinal weld in the TMI-1 vessel with the highest fluence was deter-mined to be WF-8 located at 11* off the core major axis. Vertically, the

'JF-8 weld passes through the pesk location in the axial fluence profile. Thus, shape factors for the WF-8 weld are F = 1.0 for the 11* location and F = 1.0 g

g for an axial location near the core midplane. Existing longitudinal welds in the TMI-1 vessel are located relative to the maximum fluence as shown in Fig-ures 7-2 and 7-3.

However, when material properties, fluence, and thermal environment are taken into account, the limiting weld will be determined by the fracture mechanics analysis discussed in section 8.

7.2.

Assumptions The principal assumptions used in the determination of vessel and weld flu-ences are listed below.

1.

Normalization of calculated vessel fluence is equivalent to the measured-to-calculated surveillance capsule fluence.

2.

Structural configuration, materials, and coolant conditions are sufficiently similar between 177-FA reactors that core power dis-tribution is the only input parameter that significantly affects excore fast flux distribution with respect to a specific reactor.

3.

The axial shape of fast flux for all cycles can be calculated from an " equilibrium" cycle power distribution.

4 The azimuthal shape of fast flux does not vary significantly from cycle to cycle.

5.

Excore flux (E > 1 Mev) is proportional to core leakage flux (E >

1.8 ev) as determined from cycle-dependent power distribution cal-culations.

I 7-2

6.

For cycles with equivalent designs (i.e., out-in versus in-out-in shuffle schemes), relative core escape fluxes (cycle-to-cycle) are essentially the same for all 177-FA reactors.

7.

The point location of maximum flux (azimuthal and axial) is invari-ant from cycle to cycle.

The procedure used to establish the calculational model for the analytical de-termination of f ast fluence in the pressure vessel has been benchmarked through participation in the " blind test" phase of the light water reactor pressure vessel dosimetry improvement program (LWRPVDIP).

In addition, the measured fluence at the end of cycle 2 in Oconee 1 capsule OCl-E (to which analytical flux data were normalized) was verified by HEDL under another phase of the LWRPVDIP to within 1% of the B&W value.29 HEDL is currently evaluating TMI-1 capsule TMI-1E.

Confidence in the overall fluence calculational procedure has been achieved f rom the analysis of 12 surveillance capsules from eight B&W 177-FA reactors. Comparison of calculated and measured dosimeter activities has consistently been within 10%.

Uncertainty limits have been estimated for reported fluence in the reactor ves-sel - !15% for time periods corresponding to concurrent in-reactor surveillance capsule irradiation and t18% for reactors without capsules. For extended I

time periods, this uncertainty limit is !21% for reactors without capsules.

The uncertainty limit for welds is somewhat greater and is a function of loca-tion.

However, at no time can the upper limit of a weld fluence exceed the upper limit of maximum vessel fluence.

7.3.

TMI-1 Fluence Results TMI-1 has completed four fuel cycles, which correspond to 3.52 EFPY and a neu-18 2

tron fluence of 1.80 x 10 n/cm at the peak location. This peak value also applies to circumferential weld WF-25.

Subsequent operation with same fuel 18 management will produce a maximum fluence accumulation rate of 0.055 x 10 2

n/cm per EFPY.

Based on the B&W Owners Group Integrated RV Surveillance Program and depending on whether the fuel management scheme is changed to an in-out-in low leakage fuel cycle, then a reduction of 30% in total fluence can be realized at 32 EFPY as shown in Figure 7-4.

The corresponding reduction in the RT NDT WF-8 weld is f rom 335 tc' 305'F.

7-3 I

Figure 7-1.

Fast Flux Attenuation Through Pressure Vessel Wall 1.0 1

1

-l I

g CLAD -

I:

PRESSURE VESSEL d

I i

'l l

l

.5 l

.l l

I l

l i

i g

1 I

I b

l I

l 1

5 l

i

'l l

I l

e l

=

l l

I e

i 0.1 1

_I l

l l

J l

-1 l

I I

l T/4 T/2 3T/4 I

t t

i I

3 2

4 6

8 10 12 14 16 18 20 22 inickness, em I

I 7-4

I Figure 7-2.

Typical Azimuthal Flux Profile Normalized to Peak Flux Location 1.0

.9 I

5

=

^

w

.8 E

C

~

.5

'I

~

[l d SA 1526 2

J I

.6 I

i Y

0 4

8 12 16 20 24 28 32 36 40 44 48 Azimutnal angle, degrees f rom major core axis I

I I

7-5 I

I Figure 7-3.

Longitudinal Weld Locations to Azimuthal Fluence Profile f

OUTLET N0ZZLE VESSEL X

b 7

/~ T W

Y INLET N0ZZLE

/

\\

(4)

FLUENCE NORMAllZED TO PEAK FLUENCE g

LOCATION 12' W

L 0* SA-1526 g igo WF-8 WELD WELD

.93

.95 1.0.92

.84 U

.71

.69 I

I CORE REGION l

I

/

I I

I I

7-6

I Figure 7-4.

Reductions in Peak Reactor Vessel Fluence in TMI-l 3.0 INITIAL PREDICTIONS (FSAR) 2.5 I

i*

8 5

PRE 0lCTED BASED ON

~,

2. 0 OWNERS GROUP INTEGRATE g

g t

7, ' '$

E RV SURVEILLANCE PROGRAM THESE DATA WERE USED IN

[

y

^

THE TMl-1 ANALYSIS.

'I e

i.3 G

5 a

5 8

E I

f yE s

n 0

t

[

1.0 f

I

/

/

PREDICTED BASED ON

/

LOW l.EAKAGE FUEL CYCLE

0. 5 p

I

/

assu e co m asion to 'o* u = = run cicu I

(

1 I

i l

l I

I i

0 4

8 12 16 20 24 28 32 Ef f ective Full Power Years (EFPY) 1I I

I 7-7

I 8.

VESSEL FRACTURE MECHANICS ANALYSES Over the lifetime of an operating reactor, the reactor vessel is subjected to a variety of transients that were not part of the initisi design basis; how-ever, all these transients would satisfy the stress criteria of the ASME Code,Section III, for unflawed vessels. However, if a f'.aw were postulated in a vessel undergoing a severe transient, it would become necassary to evaluate the structural integrity of the pressure vessel to determine whether the flag would grow during the transient. The transients defined in section 3 are more severe than design basis transients and were evaluzited esing Code Se<: tion XI.

Appendix A LEFM techniques to determine whether creck initiation is predicted and, if so, whether crack arrest occurs within 1/4T.

This is the criterion used for acceptability of all LEFM analyses as describec in this section.

The LEFM analysis uses as input various vessel material properties and neutron fluences on the vessel wall to evaluate the integrity of a specific reactor vessel. The plant-specific material properties (base metal and weld metal),

such as copper and phosphorus content and RT s W t and hacture touinness, NDT are described in section 6.

The plant-specific vessel neucron fluence is ob-tained per the procedure outlined in section 7.

The copper and phosphorus l

contents in the welds and the accumulating fast neutron fluence on the vessel i

as a result of continued operation result in lowered vessel toughness and, i

hence, lower resistance to fracture.

The SBLOCA plant transient data selected for fracture mechanics analysis were l

obtained from section 3.3.

The plant overcooling transient selected for f rac-ture mechanics analysis was obtained from section 3.2. The thermalanalysis for all transients was performed as outlined in sections 4 and 5.

The LEFM evalu-ations described here determine the maximum lifetime that a vessel can experi-

!I ence with crack arrest within 1/4T for the severe transients considering its decreasing fracture toughness.

For the thermal shock concern, various regions within the vessel were evaluated. LEFM analyses were performed on the vessel l

i 8-1

!I

I beltline region longitudinal and circumferential velds, the base metal, and a vessel inlet nozzle.

8.1.

Analytical Model 8.1.1.

Vessel Beltline Region Based on LEFM techniques from the ASME Code,Section XI, Appendix A, the PCRIT computer code determines the critical pressure as a function of time using the plant-specific vessel weld chemistry, the weld location, the flaw tip (vessel gradient) temperature as a function of time, and the material properties as a function of EFPY (fluence) and fluence attenuation through the vessel wall.30 The flaw shown in Figure 8-la is evaluated as a semi-elliptical flaw with a 6:1 aspect ratio.

Flaw sizes from 1/40T to 10/40T in 1/40T increments are evaluated at various time steps throughout the transient for an input set of operating histories (EFPY). The procedure is the same for evaluation of the vessel longitudinal weld, circumferential weld, and base metal. Warm prestress-ing (WPS) techniques were not used during analysis in this report.

8.1.2.

Vessei Inlet Nozzle The output from the two-dimensional axisymmetric finite element model (section 5.3), consisting of a detailed pressure and thermal stress profile, is usad in 27 conjunction with the BIGIF code to determine the K versus a through-the-

" knee" section of the nozzle (see Tigure 8-1).

A plot of the stress intensity factor (SIF) versus a/t for flews in the nozzle material can be made from the pressure stress results. The SlF f'r the inlet nozzle was determined hera using the BIGIF computer code in lieu of those provided in the Welding Research Council Bulletin 175?

because the Bulletin has been shown to be everly con-servative for small flaw sizes",

i.e., flaw sized lese than a/r < 0.15.

The n

flaw geometry shown in Figure 8-lb is assumed in the evaluation of thermal shock in the " knee" of the vessel inlet nozzle.

Using the flav size, fracture toughness, and flaw tip tenperature, the critical pressure can be determined as a function of time using tha SlF.

The cri?fcal pressure is then compared te che actual transient pressure.

Thtse evaluations are iterative to determine the caxir.m vessel lifeti:7.e (in EFPY) for the tran-sient being censidered-8-2 I

t

8.1.3.

Criteria for Acceptability Several criteria were considered for the LEFM analysis to determine whether a particular analysis gave an acceptable result for vessel life. Each criterion would be judged on its intent and the conservatisms it provided in the final evaluation of a particular set of fracture mechanics results. Two criteria for acceptability were considered for the LEFM analysis evaluation of the thermal shock concern:

1.

No crack initiation during the transient.

2.

Crack initiation with arrest within 1/4T.

The first criterion has the intent of preventing crack initiation at all times; therefore, maximum conservatism exists but is overly conservative in that it may restrict plant lifetime ut.necessarily.

The second criterion allows credit to be taken for possible small existing flaws that may run to slightly larger flaws and arrest; this allows additional service life for the reactor vessel with only a slight reduction in conserva-tism.

This criterion follows the philosophy of Section III of the ASME Code for plant decign anal: sis and is the one selected for use in the LEFM analysis of the chermal shock concern.

8.2.

Assumptions and Conservatisms the use of LETM techniques implies certain basic assumptions per the ASME Code; other major assumptions ce changes to the basic ASME Code,Section IX, Appendix I

A, are as follows:

1.

Flaw Description - The aspect ratio for the initial flaw (Figure 8-1) was M

assumed to be 6:1 as recommended by Section III, Appendix G, of the ASME Code.

I 2.

Material Fracture Toughness - The material reference temperature, RTg, is adjusted to account for radiation embrittlement effects. The amount of I

adjustgent (see section 6) ic compared using NRC Regulatory Guide 1.99 and is a function of fluence and chemistry (i.e., Cu and P).

3.

Upper Shelf Toughness - A limit of 200 ksi E is used as the upper limit for the materia! toughness (see section 6).

I 8-3 i

4.

Cladding - The effects of cladding have been taken into consideration in develop'ng the SIF(s) due to thermal stress, but were excluded from the SIF due to pressure. The cladding effects were also included in the de-termination of the heat transfer film coefficients for the inner surface of the three-dimensional vessel and nozzle models.

5.

Material Properties - Material properties are taken from the ASME Boiler and Pressure vessel Code appendixes, 1980 Edition. The vessel cladding is stainless steel. A cladding thickness of 0.125 in. is used for conserva-tism; the nominal thickness is 0.1875 in.

The vessel base metal is Mn-Mo, ferritic steel.

8.3.

Discussion of Fracture Mechanics Results 8.3.1.

Fracture Mechanics Results - Critical Pressure Vs Time Histories Critical pressure curves were determined for each longitudinal and circumfer-ential weld, the vessel base metal and the vessel inlet nozzle for both SBLOCA (section 3.3) and overcooling (section 3.2) plant transients. A typical crit-ical pressure curve is compared to the actual transient pressure versus tran-g sient time in Figure 8-2.

The acceptance criterion was satisfied and thus E

the welds are determined to be acceptable as long as the critical pressure curve did not intersect the actual transient pressure curve. The PCRIT program also determines the existing flaw size, what flaw size becomes critical or does not arrest, and at what time in the transient this occurs.

The allowable versus actual pressure comparison curves are similar for all welds and transients evaluated.

8.3.1.1.

SBLOCA Transient Fracture Mechanics Analysis Results Evaluations of the TMI-1 reactor vessel welds, base metal, and inlet nozzle materials were made using various conservative mixing codes and generic SBLOCA transient.

For the welds that proved limiting in the conservative analysis, a reanalysis was performed using the COMMIX code to determine the mixing in the vessel cold leg and downcomer and the results from a TMI-1 plant-specific SBLOCA, section 3.3.

Using the latest results and LEFM methodology, the eval-uations showed all welds would meet the criteria of crack initistler. with ar-5 rest within 1/4T, at 40 EFPY. Actually, the results showed no crack initiation 8-4 1

was experienced during the transient. The results indicate the base metal, longitudinal and circumferential welds, and inlet nozzle would provide integ-rity for the vessel for greater than 40 EFPY.

Based on the parallel mixing analysis program results discussed in section 4 and the comparison to the one-fifth scale CREARE Test Facility data reported in reference 24, FLOW-2A and COMMIX mixing analyses have shown good agreement in mixed temperatures below the inlet nozzle. WPS was not used in the SBLOCA fracture mechanics analysis.

The inlet nozzle was evaluated using the FLOW-2D mixing results since the data were available when the nozzle analysis was begun. The nozzle temperature pro-file is shown in Figure 5-8.

Since the nozzle was found to be acceptable for I

40 EFPY using these data, no additional analysis will be needed for the SBLOCA transient.

The LEFM results for the SBLOCA transient show the following in terms of EFPY, using acceptance criterion 2; however, no crack initiation resulted from this I

SBLOCA transient.

EFPY Vessel inlet nozzle

>40 Longitudinal weld

>40 Circumferential weld

>40 Vessel base metal

>40 All LEFM analytical results for SBLOCA transients are listed in Table 8-1.

8.3.1.2.

Plant-Specific Overcooling Transient Fracture Mechanics Analysis Results Fracture mechanics analysis was performed for the overcooling transient de-scribed in section 3.2.

WPS was judged not to be applicable on this over-cooling transient because of the repressurization that occurred during the transient. The LEFM results for the overcooling transient are as follows in terms of EFPY, using acceptance criterion 2; again no crack initiated in the vessel as a result of this overcooling transient.

I I

8-5 I

I EFPY Vesse? inlet nozzle

>32 Longitudinal weld

>32 Circumferential weld

>32 Vessel base metal

>32 All LE7M analysis results for the above overcooling transient are listed in Table 8-1.

I Table 8-1.

LEFM Results for Thermal Shock Analysis of TMI-1 Reactor Vessel (a)

Weld ide t Overcooling (c) or heat b)

SBLOCA 40( )

32 ARY-059 WF-70 40 32 WF-8 40( }

32 40(*}

32 C2789-1 WF-25 40 32 C3251-1 40("}

32 SA-1526 40( )

32 Atypical 40 32 Vessel inlet nozzle 40( }

32 I

(" Criterion for acceptability of results:

LEFM analysis crack arrest within 1/4T.

g Results also show no crack initiation for either transient.

Refer to Tables 6-1 and 6-2 for locations.

(

Predicted based on conservative mixing analysis.

I I

8-6

i I

Figure 8-1.

Flaw Geometry Used in Vessel Thermal Shock Analysis TENSION S10E g

IN BENDING

^

?

a t

t/a = 6 TO 1 ONPRESSION S10E SURFACE FLAN IN BEN 0 LNG (t = SECTION THICKNESS)

I a.

Asstaned flaw geometry in vessel I

I

'I KNEE I

\\

Y I

I b.

Assumed Flaw Geometry in Vessel Inlet Nozzle I

I I

8-7 I

ll amu me 0

8 1

am u

Y 0

m P

6 i

F 1

E un 2

3 E

t R

s a

U 0

u 4

S a

d S

1 l

E e

R W

P sn 0

T u

7 N

0 E

2 F

I 1

W S

S N

e N

A t

O R

n I

T s

e E T 0

e i

R A

0 t

I s

U D C

1 u

a S O L

i a

n S N O

ll r

E C m

m T

R B

u P F S

A O

e C

L 0

m 8

i m

O A T L

C E T

u B

S I

S T

I R

r R O o

C F m

f 0

u y

6 s

ro ts e

i l

m l

i e

0 r

4 us n

s u

e n

r P

0 2

m 2

u n

8 er g

m u

i 0

u i

n F

0 0

0 0

0 0

0 0

0 0

0 0

0 0

5 0

5 0

5 3

2 2

1 1

__a $5 S mun

=

• sus l

ll11 1

l1l llll{

I I

9.

REFERENCES I

1 Thermal-Mechanical Report - Effect of HPI on Vessel Integrity for Small Break LOCA Event With Extended Loss of Feedwater, BAW-1648, Babcock & Wilcox, Lynchburg, Virginia, November 1980.

2 W. T. Sha, et al., COMMIX-1: A Three-Dimensional Transient Single-Phase Component Computer Program for Thermal-Hydraulic Analysis, NUREG/CR-0785 (ANL-77-96), September 1978.

3 Y. A. Hassan, " COMMIX Predictions of a Thermal Mixing Test," ANS Transac-tions, Vol 40 (1982).

" B. D. Nichols, C. W. Hirt, and R. S. Hotchkiss, SOLA-V0F - A Solution Algo-rich for Transient Fluid Flow With Multiple Free Boundaries, LA-8355. Los Alamos National Laboratory (1980).

S G. J. de Salvo and J. A. Swanson, ANSYS Computer Code, Swanson Analysis Systems, Inc., Fuly 1979.

8 A. L. Lowe, et al., Analysis of Capsule TMI-1E, Metropolitan Edison Com-pany, Three Mile Island Unit 1 - Reactor Vessel Materials Surveillance Program BAW-1439 Babcock & Wilcox, Lynchburg, Virginia, January 1977.

7 H. S. Palme, H. W. Behnke, and W. J. Keyworth, Methods of Compliance With Fracture Toughness and Operational Requirements of 10 CFR 50, Appendix G.

BAW-10046A Rev. 1, Babcock & Wilcox, Lynchburg, Virginia, July 1977.

e A. L. Lowe, e_t al., Irradiation-Induced Reduction in Charpy Upper-Shelf Energy of Reactor Vessel Welds, BAW-1511P, Appendix B, Babcock & Wilcox, Lynchburg, Virginia, October 1980.

S A. L. Lowe, et al., Irradiation-Induced Reduction in Charpy Upper-Shelf Energy of Reactor Welds, BAW-1511P, Babcock & Wilcox, Lynchburg, Virginia, October 1980.

18 C. L. Whitmarsh, Pressure ilegal Fluence Analysis for 177-FA Reactors, BAW-1485, Babcock & Wilcox, Lynchburg, Virginia, June 1978.

I

11 RETRAN - A Program for One-Dimensional Transient Thermal-Hydraulic Analysis of Complex Fluid Flow Systems, EPRI CCM-5 (1978) and NP-2175 (December 1981), Electric Power Research Institute, Palo Alto, California.

12 F. H. Speckhart and W. L. Green, A Guide to Using CSMP - The Continuous System Modeling Program, Prentice-Hall (1976).

I3 T. G. Broughton and N. G. Trikouros, "RETRAN PWR Applications at GPU,"

Nuclear Technology, Vol 54 (1981), p. 342.

1" "RETRAN Simulation of the TMI-1 Turbine Test," Trans. Am. Nucl. Soc., Vol 32,(1979), p. 445.

l' L. C. Pwu and T. G. Broughton, "RETRAN Simulation of the TMI-2 Startup Overfeed Incident," Nuclear Technology, Vol 54 (1981), p. 358.

18 CRAFT - Fortran Program for Digital Simulation of a Multinode Reactor Plant During LOCA, BAW-10092, Babcock & Wilcox, Lynchburg, Virginia, April 1975.

17 D. F. Ross (NRC) to J. H. Taylo (B&W), Letter, "Information Request on Reactor Vessel Brittle Fracture," July 12, 1979.

le Report in Response to NRC Staff: Recommended Requirements for Restart, Three Mile Island Nuclear Station, Unit 1.

l' Final Cafety Analysis Report, Three Mile Island Unit 2.

I 20 " Steam Line Rupture Detection System (SLRD) Actuation," TMI-1 Abnormal Procedure 1203-24, Rev. 13.

21 Abnormal Transient Operating Guidelines (ATOC) for TMI-1, BWNP-2007, June 1976.

22 " Loss of Reactor Coolant / Reactor Coolant Pressure Causing Automatic High Pressure Injection," TMI-1 Emergency Procedure 1202-6B Rev. 7.

23 C. W. Hirt and B. D. Nichols, "A Computational Method for Free Surface Hydrodynamics," ASME Pressure Vessel and Piping Conference, San Francisco, August 12-15, 1980; Paper No. 80-C2/PVP-144, J. Pressure Vessel Technology, 103 (1981), p.

136.

2" P. H. Rothe, W. D. Marscher, and J. A. Block, Fluid and Thermal Mixing in a Model Cold Leg and Downcomer With Vent Valve Flows, Interim Report. EPRI NP-2227, Electric Power Research Institute, Palo Alto, California, March

=

1982.

I 9-2 I

I 25 F. W. Dittus and L. M. K. Boelter, University Publications in Engineering, Vol 2. University of California (1980).

26 C. C. Vliet and D. C. Ross " Turbulent Natural Convection," ASME J. of Heat Transfer (1975).

27 BICIF - Fracture Mechanics Code for Structures. Failure Analysis Associates, December 1980.

2e K. E. Moore, et al., Evaluation of the Atypical Weldment, BAW-10144A, Babcock 6 Wilcox, Lynchburg, Virginia, February 1980.

29 W. N. McElroy, et al., Surveillance Dosimetry of Operating Power Plants, llEDL-SA2546, llanford, October 16, 1981.

30 PCRIT - Critical Pressure Calculation Linear Elastic Fracture Mechanics, NPCD-TM-585. Babcock & Wilcox, Lynchburg, Virginia, November 1981.

31 PVRC Recommendations on Toughness Requirements for Ferritic Materials, PVRC ad hoc Group on Toughness Requirements, WRC-175, August 1972.

32 Improved Evaluation of Nozzle Corner Cracking, EPRI NP-339, Electric Power Research Institute, Palo Alto, California, March 1977.

I I

I I

I I

I

'~'

I

l l

t I

t

(

i lI

(

l lI l

l l

lI

I I I I

APPENDIX l

l I E CSMP Model for Long-Term Cooling Il l

l I

I l

I I

,I I

I I

A-1 I

The steam generator secondary side, modeled as a saturated two-phase volume with total volume V, consists of a vapor space (which occupies a fraction a 7

of V ) and a saturated liquid space.

The specific enthalpies and volumes of T

vapor and liquid are denoted as h.h, vg, and v, respectively. The quality g g (x) and mixture average enthalpy (h ) are related by a/vg x=

(g_g) a/(v ) + (1 - a)/v g and h = xh + (1 - x)h.

(A-2) m g

f Then the conservation of mass requires dMT dt fw - W (A-3)

=W stm where W and W are feedwater flow and steam flow, respectively, with their g

associated enthalpies h and h. The conservation of energy requires g

b=W h

-W h

+Q (A-4)

M dt fw fw stm g sg I

where Q is the heat transferred from the primary side, approximated by Sg sg)

(A-5)

Q

= uA(T

-T sg avg where u = heat transfer coefficient, A = wetted surface area tube circumference / cross sectional area,

= M (I - X)Vf x

T RCS average temperature, T

=

steam generator secondary temperature E

T

=

sg = T at steam generator secondary pressure P.

3 sat Since the total internal energy U is defined by U = M (H - Pv )

(A-6)

T where tg is the total mass and v is the mixture specific volume.

Substituting equations A-6 and A-3 into A-4, we obtain I

A-2 1

I dh d P' g

g (hg

-h)-W (h -h)+Q,

+V (A-7)

W

=

Tk.

From equation A-2, its derivative with respect to time is db

'3h '

'3h '

- dh dh ~

I

+

=

=

x

+ (1 - x)

+

^~

3p g

dP_

fg Equating equations A-7 and A-8, we obtain

~ dh dh V"

h

+

x

+ (l ~ *)

~

(Nfw(hfw - h,)

g dP dP

-Wsm(h - h,) + Q,].

(A-9)

The equation is in the form of first-order linear dif ferential equation of x and P.

In order to solve both, another equation is required; it is found in the equation of state of the system:

T " H [*"g + (l ~ *}"f l V

T

-V(x,P,g)

T

= constant.

(A-10)

Thus, dV

'3v

'3V '

"T T

T dx T

dP T

dt

,3x j dt

( 3P,

E 3M dt p

T

- dv dv ~

+M

+ (l ~ *)

= M ("g ~ "f)

T,*

dP_

T dMT

+ [xv + (1 - x)v ] dt

= 0.

By rearranging the terms and substituting equation A-3 for dM /dt, we have I

~ dh dv ~

T M "fg

+b*

+ (l ~ *)

T s m} ( *"g + (l ~ *)"f].

(A-II)

= -(W

-W g

I A-3

I Note that saturation is maintained throughout the transient, and the deriva-tives of h and v with respect to P are evaluated along the saturation lines.

The energy balance of the primary side is given by dT FL C avg = Q

-Q (A-12)

Rp dt c

sg Q = core heat, M = RCS mass, R

C = specific heat.

Equations A-9, A-11, and A-12 are existing simultaneously with unknowns d, P, and T They are readily solvable by algebraic manipulation and then inte-gration as long as the boundary conditions of the feedwater and steam flows are specified.

The RCo pressure and level during the transient are calculated by assuming that the pressurizer vapor space is adiabatic; i.e., no heat is transferred into and out of the vapor.

Then, PVf*3 = constant (A-13) where V is the pressurizer vapor space volume, given by

~

~

v res L

~ Yf avg' res res where V is the total RCS volume (a constant) and v is the specific volume g

g of liquid at T and P avg res A simple program to solve the RCS pressure and temperature and steam generator pressure and level was written using the CSMP (continuous system modeling pro-gram).

It is capable to predict the general trend for an isolated primary system with the steam generator controlling the secondary.

Boundary conditions at the secondary side are controlled by the user, such as feedwater throttling to maintain a constant level and/or cycling steam relief valves to maintain a constant pressure.

Since the program assumed a uniform RCS temperature (T

)

and loop flow was not accounted for, the required cold leg or downcomer tem-E perature is obtained by subtracting from T a constant that would produce A-4 m

I continuity in switching from the RETRAN to the CSMP calculation. These as-sumptions are believed to be conservative; thus, the simplified method is especially useful for solving long-term transients.

I Figure A-1.

Schematic Model Diagram of Reactor Coolant System and Steam Generation for Long-Term Transient Calculation PRCS PRESSURIZER E

I T g=Tsat (P)

MR 3

s

+

h y

g WSTM hg I

,A AA A A AA AA E A u

PRCS OSG hg Qew+

ws+

f ayg PCSI V (T P

Wpw hpw RCS SG E

I I

I A-5

.