ML20137D279

For Comment Issue of Draft Reg Guide,Task SI 502-4, Format & Content of Plant-Specific Pressurized Thermal Shock Safety Analysis Repts for Pwrs

ML20137D279
Person / Time
Issue date:	01/31/1986
From:	NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:
References
REF-GTECI-A-49, REF-GTECI-RV, RTR-REGGD-1.154, TASK-A-49, TASK-OR, TASK-RE, TASK-SI-502-4 REGGD-01.XXX, REGGD-1.XXX, NUDOCS 8601160594
Download: ML20137D279 (63)

Text

-

POL

/

U.S. NUCLEAR REGULATORY COMMISSION y

't, OFFICE OF NUCLEAR REGULATORY RESEARCH Jrnuary 1986 y

Division 1 o

DRAFT REGULATORY GUIDE AND VALUE/ IMPACT STATEMENT Task SI 502-4

\\..*/

D

Contact:

R.H.W. Woods (301)492-4714 O[M.~Qq (y

Y n7 k

ow r

4

~

A; #, y

,, $1 ef O

t A,Q ' Cy N 1.

% g, m a

5. y.

,%g /; w. ' ' s,- m.,7, FORMAT AND CONTENT OF PLANT-SPECIFIC PRESSURIZEDTHERMALSH5CK-S5NETYANALYSIS ^'. REPORTSFORPRESSUkIZEDWdTERREACTORS ,3 . m-e y

-(

^ \\(: f '.,4 y%

? \\g L'

( ".-;? S_.h.. [QN R il f D']f n

s. ?

K3-f/ %y, 8601160594 e60131 PDR REGGD 01.XXX R PDR This regulatory guide and the associated value/ impact statement are being issued in draf t form to involve the public in the early stages of the development of a regulatory position in this area. They have not received complete staff review and do not represent an official NRC staf f position. Public coments are being solicited on both drafts, the guide (including any implementation schedule) and the value/ impact statement. Coments on the value/ impact statement should be accompanied by supporting data. Written comients may be submitted to the Rules and Procedures granch. ORR Adit. U.S. Nuclear Regulatory Conriission. Washington. OC 20555. Coments may also be delivered to Room 4000. Nryland National Bank Building. 7735 Old Georgetown Road, Bethesda. Nryland from -x i 8:15 a.m. to 5:00 p.m. Copies of cons,ents received may be examined at the PRC Pubite Document Room.1717 H 5treet NW.. J Washington. DC. Coments will be most helpful if received by March 14, 1986. i ^' Requests for single copies of draf t guides (which may be reproduced) or for placement on an automatic distribution list for single copies of future draft guides in speelfic divisions should be nyde in w iting to the U.S. Nuclear Regulatory Cveitssion, r Washington. DC 20555. Attention: Director. Division of technical Information and Document Control. j 3

} Table of Contents Page INTRODUCTION. v CHAPTER 1 OVERALL APPROACH, SCOPE OF ANALYSIS, AND REPORT ORGANIZATION. 1 CHAPTER 2 PLANT DATA. 3 2.1 Systems Pertinent to PTS. 3 2.2 Reactor Vessel. 4

2. 3 Fluence.....

4 2.4 Inservice Inspection Results.......... 5

2. 5 Plant Operating Experience....

5 2.6 Operating Procedures.. 5 CHAPTER 3 DETERMINATION OF DETAILED PTS SEQUENCES FOR ANALYSES................. 7 3.1 Approach Used.. 7 3.2 Sequence Delineation.. 7 3.2.1 Development of Classes of Initiators. 7 3.2.2 Identification of Important Initiator Variations.... 8 3.2.3 Definition of Potential Transients Resulting from Each Initiator. 9 3.3 Operatot Effects. 10 3.4 Sequenct Quantification................. 11 3.4.1 Initiating Events................. 11 3.4.2 Equipment Failures.. 12 3.4.3 Operator Actions................. 13 3.5 Event Tree Collapse. 14 3.5.1 Specific Sequences. 14 3.5.2 Residual Groups. 15 CHAPTER 4 THERMAL-HYDRAULIC ANALYSIS. 16 4.1 Thermal-Hydraulic Analysis Plan.... 16 4.2 Thermal-Hydraulic Model.... 17 4.3 Simplified Analysis Methods. 18

4. 4 Thermal Stratification Effects.

18 4.5 Thermal-Hydraulic Analysis Results. 19 CHAPTER 5 FRACTURE MECHANICS ANALYSIS. 21 CHAPTER 6 INTEGRATION OF ANALYSES. 24 iii

Table of Contents (Continued) Page CHAPTER 7 SENSITIVITY AND UNCERTAINTY ANALYSES OF THROUGH-WALL CRACK FREQUENCY.. 26 7.1 Sensitivity Analysis. 26 27 7.2 Uncertainty Analysis... 7.2.1 Parameter Uncertainties.. 27 7.2.2 Modeling Uncertainties (Biases). 28 CHAPTER 8 EFFECT OF CORRECTIVE ACTIONS ON VESSEL THROUGH-WALL CRACK FREQUENCY,....... 29 8.1 Flux Reduction Program. 29 8.2 Operating Procedures and Training Program Improvements....... 29 8.3 Inservice Inspection and Nondestructive 31 Examination Program. 8.4 Plant Modifications.... 31 8.5 In Situ Annealing.................... 32 CHAPTER 9 PREDICTION OF VESSEL FAILURE MODE............ 33 38 CHAPTER 10 LIKELIHOOD OF CORE MELT AND PREDICTION OF RISK. 10.1 Risk Analysis...................... 38 38 10.2 General Guidance............ 10.3 Core Melt Frequency.... 39 10.4 Containment Failure. 40 10.5 Source Term. 41 10.6 Site Consequence Analysis......... 41 10.7 Risk. 42 10.8 Uncertainties.. 43 CHAPTER 11 RESULTS AND CONCLUSIONS REGARDING PTS RISK. 44 11.1 Summary of Analysis... 44 45 11.2 Basis for Continued Operation........ 46 REFERENCES. DRAFT REGULATORY ANALYSIS........................ 49 e iv

I INTRODUCTION Background and Purpose of This Guide l The pressurized thermal shock (PTS) rule, S 50.61 of 10 CFR Part 50 issued on July 23, 1985 (50 FR 29937), establishes a screening criterion based on reactor vessel nil-ductility-transition temperature (RTNDT). The screening criterion was established after extensive industry and NRC analyses regarding the likelihood of vessel, failure due to PTS events in pressurized water reactors (PWRs). The analyses were applied generically and contained conservative assumptions to make the results bounding for any PWR. Based on the results, the NRC concluded that the risk due to PTS events is acceptable at any plant so long as the RT f the reactor pressure vessel remains PTS below the screening criterion. Extensive safety analyses are required by the rule for any plant that wishes to operate with RT values above the screening criterion. The recom-PTS mended methods to be used in performing the analyses are outlined in this guide. The purpose of the analyses is to assess the risk due to PTS events for proposed operation of the plant with reactor vessel RT above the screen-PTS ing criterion. Section 50.61 requires that these analyses be completed 3 years before the screening criterion would be exceeded to allow adequate time for implementation on the plant of any corrective actions assumed in the analyses before the plant operates above the screening criterion. This regulatory guide describes a format and content acceptable to the NRC staff for these plant-specific PTS safety analyses and describes acceptance criteria that the NRC staff will use in evaluating licensee analyses and proposed corrective measures.

• To avoid confusion among several (preexisting) slightly different definitions of RTNDT, S 50.61 contains its own definition of an RTNDT (called RTPTS) to be used when comparing plant-specific vessel material properties with the PTS screening criterion.

v m

The references listed in this guide include a set of analyses sponsored by the NRC that, taken together, constitute an example of the analyses described by this guide. The staff recommends that these references be extensively used, along with this guide, by those performing the plant-specific PTS analyses re-quired by the PTS rtle, S 50.61. References 1, 2, and 3, for example, each represent an analysis by the Oak Ridge National Laboratory (ORNL) predicting through-wall crack frequency for one PWR. These references will provide guid-ance through a majority of the analyses. Reference 3 (analysis of H. B. Robinson) should be most helpful because it was the last one performed and includes the experience gained in performing the two earlier analyses. Objectives of Plant-Specific PTS Safety Analysis Reports Paragraph 50.61(b)(4) requires that a licensee whose plant will exceed the screening criterion before expiration of the operating license submit safety analyses to determine what, if any, modifications to equipment, systems, and operation are necessary to prevent potential failure of the reactor vessel as a result of postulated PTS events if continued operation beyond the screening criterion is allowed. These analyses must include the effects of all corrective i actions the licensee believes necessary to achieve an acceptable PTS-related risk for continued operation of the plant. The final objective of the plant-specific PTS study, therefore, is to justify continued operation of the plant by demonstrating that the risk from such operation is acceptable. The study must include calculation, as a function of remaining effective full power years of plant life, of the expected frequency of through-wall cracks due to PTS. In addition, for plants in which the mean predicted frequency of PTS-related through-wall cracks is greater than 5 x 10 8 per reactor year, the study must also include calculations of: Expected frequency of core melt due to PTS. Expected frequency of large release of radioactivity from containment due to PTS. i Resulting person-rem exposure and early and late fatalities that may result from PTS. e vi

l In calculating these results, it will be necessary to: Identify the dominant accident sequences. Identify operator actions, control actions, and plant features impor-tant to PTS. Estimate the effectiveness of potential corrective actions in reduc-ing the expected frequency of through-wall crack and, in cases with through-wall crack frequency greater than 5 x 10 6 per reactor year, their effectiveness in reducing the frequency of core melt and large release of radioactivity. Identify the sources and approximate magnitude of the major uncertain-ties and their effects on the conclusions. Present and justify the licensee's proposed program for corrective measures. Present and justify the licensee's proposed basis for continued opera-tion at embrittlement levels above the screening criterion. This must include comparison with the acceptance criteria described below of the PTS-related through-wall crack frequency (and risk, if required) with corrective actions implemented as necessary. Staff Review of Plant-Specific PTS Safety Analysis Reports and Acceptance Criteria for Continued Operation The PTS rule specifies a screening criterion based on RTNDT (called RTPTS for use as defined within the rule) of 270 F for axial weld and plate materials and 300 F for circumferential weld materials. As detailed in SECY-82-465 (Ref. 4), these values were selected based on generic studies of the expected frequency and character of a wide spectrum of transients and accidents that could cause pressurized overcooling of the reactor vessel (PTS events) and on operating experience data. The risk due to PTS events was assessed in terms of probabilistic fracture mechanics calculations of the expected frequency of through-wall crack penetration of the pressure vessel due to the PTS events. In selecting the screening criterion based on those calculations, the conservative assumption was made that any through-wall crack could result in severe core degradation or melt. Core melt itself was viewed as an event to be vii

avoided even though risk to the public due to such an event in terms of person-ems and early and late fatalities was not calculated with any certainty. The estimated through-wall crack frequency developed as a function of RT fr NDT axial welds (Fig. 8.3 of Ref. 4) is shown in Figure 1. The RT screening criterion selected by the staff corresponds to a mean PTS (or average) "best estimate" surface RT f 210 F. The staff used a "2-sigma" PTS value (spread between "best estimate" and " upper limit") of 60*F;* thus the screening criterion expressed in terms of RTPTS, which, by definition, is this upper limit value, was selected at 210 + 60 = 270 F. For axial weld and plate materials, Figure 1 gives a through-wall crack frequency of about 5 x 10 6 per reactor year at 210 F, which corresponds with an RT f 270% For ckcum-PTS ferential welds, the same frequency occurs at approximately 300 F (Ref. 4). The Commission concluded that the PTS-related risk at any PWR is acceptable so long as the RT values remain below the specified screening criterion. PTS It was realized that there are many unknowns and uncertainties inherent in the probabilistic calculations; thus it was with deliberate intent that conser-vative assumptions such as those stated above were made. The expectation was that the true risk at any plant due to PTS events would in all likelihood be considerably below that derived from Figure 1 and would therefore be acceptable. Also contributing to the belief that the real PTS risk at any given plant was lower than that resulting from the analysis in Reference 4 was the belief that many of the generic plant assumptions made in Reference 4 (e.g., material properties, system performance, crack distribution) would prove to be overcon-servative for analysis of a specific plant and that the resulting plant-specific analysis, when performed, is likely to result in a reduced prediction of PTS risk. If tne plant-specific PTS analyses submitted by licensees in accordance with S 50.61 using the methodology described in this guide (or acceptable equi-valent methodology) predict that the PTS-related, through-wall crack penetration n Based on preliminary RT data from many plants (see Table P.1 of Enclosure A q NDT to Ref. 4). viii

I LONGITUDINAL CRACK EXTENSION NO ARREST SECY-82 465 PRA RESULTS ] 10'2 1 LEGEND: I O PRA TOTAL O STEAM LINE BREAKS ~ A S.G. TUBE RUPTURES ~ V SBLOCA W/WPS / 10'3 / 2 O EXTENDED HPl T / 8 $i / 10 r O Oe s i i a y ,e 0 s$ ' t u. f 10'5 r l / 7 / f l / j V / p // j ris / $lI l l i l i l i i di i i [ l i iiliiiiliiiilI I I I 10 6 ii 175 200 225 250 275 300 325 350 MEAN SURFACE RTNOT( F) Figure 1 D

frequency will remain less than 5 x 10 0 per reactor year for the requested period of continued operation, such operation would be acceptable to the staff. The acceptability of continued operation with predicted through-wall crack frequency greater than 5 x 10 6 per reactor year must be justified by analyses of predicted core-melt frequency, predicted frequency of large release of radio-activity from containment due tu PTS, and resulting person-rem exposure and early and late fatalities that may result from PTS. Thus, the risk analyses described in Chapters 9, " Prediction of Vessel Failure Mode," and 10, " Likelihood of Core Melt and Prediction of Risk," of this guide will be necessary as part of the plant-specific PTS analyses only for plants in which the PTS-related through-wall crack frequency is greater than 5 x 10.G per reactor year. In all the analyses performed, the licensee must justify that the impor-tant input values used are valid for the remaining life of the plant. It is anticipated that the first PT.3 analyses required by S 50.61 will not be started until several years after publication of this guide. In view of the present extensive ongoing efforts by the Commission in the areas of quantitative assessment of risk and definition of acceptable risk, it is not possible to publish in this guide the quantitative risk criteria that may be in effect when the first PTS analyses are reviewed. In evaluating the acceptability of the licensee's justification of continued operation, the staff will consider the core melt frequency and early and late fatality results presented by the licensee and, in addition, will use any quan-titative criteria that have been established by the NRC at that time. Recommended Format The recommended content of plant-specific PTS safety analyses is presented in Chapters 1 through 11 of this guide. Use of this format by licensees will help ensure the completeness of the information provided, will assist the NRC staff in locating the inforlation, and will aid in shortening the time needed for the review process. If the licensee chooses to adopt this format, the x

numbering system of this guide should be followed at least down to the section level. Certain sections may be omitted if they are clearly unnecessary to pro-vide for comprehension of the analysis or if they are repetitive. Additional guidance on style, composition, and specifications of safety analysis reports is provided in the Introduction of Revision 3 to Regulatory Guide 1.70, " Standard Format and Content of Safety Analysis Reports for Nuclear Power Plants (LWR Edition)." Any information collection activities mentioned in this regulatory guide are contained as requirements in 10 CFR Part 50, which provides the regulatory basis for this guide. The information collection requirements in 10 CFR Part 50 have been cleared under OMB Clearance No. 3150-0011. D xi

1. OVERALL APPROACH, SCOPE OF ANALYSIS, AND REPORT ORGANIZATION } This chapter should describe the overall approach to the analysis and should outline the individual tasks in terms of the nature and source of input, the methods used for analysis, and the nature and subsequent use of the output. The interrelationship of the tasks should be described and should be illustrated by a flow chart. Describe how the analysis tasks are integrated to achieve the results and conclusions. Major emphasis should be placed on analyzing event sequences leading to vessel through-wall cracking. Analysis of subsequent core melt and release of radioactivity is required only in cases in which the predicted through-wall crack frequency is greater than 5 x 10 '3 per reactor year. The report should include both deterministic and probabilistic fracture mechanics analyses. The deterministic analyses should be used to predict mate-rial condition limits beyond which a given event would cause vessel through-wall D pressed in terms of a " critical RT crack penetration for the most limiting initial crack size. This can be ex-The probabilistic analyses should be NDT. used co determine the statistical likelihood of vessel through-wall crack pene-tration, assuming a crack size distribution appropriately justified for the vessel being analyzed, and appropriate uncertainties and distribution of the significant input parameters such as material properties. Both methods should be used to compare the predicted frequency of vessel through-wall cracking before and after implementing the proposed corrective measures. The input to the probabilistic analysis should be best-estimate based on appropriate assumptions. Uncertainties and conservatisms should be explicitly presented in the decision rationale for the licensee's proposed corrective mea-sures and basis for continued operation. The analysis should include effects of operator actions, control system interactions, and support cystems such as electric power, instrument air, and service water cooling. )D 1

\\ The report should be uPganized by starting with a description in Chapter 1 of how the report chapters and supporting appendices are interrelated and what material is in the appendices. The main report should describe the objectives and overall approach used in the study, outline the plant systems analyzed, describe T.he engineering anal-yses performed, present the results obtained and conclusions drawn, and present and justif*/ the licensee's proposed program of corrective measures. Appendices should contain data, detailed models, sample calculations, and detailed results needed to support the various chapters of the report. Appen-dices should contain little supporting text. Instead, the nature and relevance of material in the appendices should be described in the pertinent chapters of the main report. Certain details (some are noted later in this guide) may be available for NRC inspection in audit notebooks and need not be directly submitted with the analyses. \\ e 2

I 2. PLANT DATA This chapter should briefly describe plant systems and operations pertinent to PTS. Chapter 2 of Reference 3 (the H. B. Robinson analysis by ORNL) provides a good example. Supporting appendices or refer,ences should present the design and operating data used in the analysis or needed to understand the analysis. References to other documents (e.g., the final safety analysis report (FSAR)) should indicate specific sections. (Reliability data, however, should go in Section 3.4, " Sequence Quantification," or its supporting appendices and references.) 2.1 Systems Pertinent to PTS Summarize design and operating features of systems pertinent to PTS. Illustrate each system with a simplified process and instrumentation diagram or a single line diagram. Identify on each illustration any interfaces with other systems. For each system, include a table summarizing key design and D operating data. Give the maximum, minimum, and nominal values for those cases in which design data may vary with time (for example, high pressure injection (HPI) water temperature may vary with season). Such values used in the analysis should be identified and justified. Refer to appendices or other documents (e.g., specific sections of the FSAR) as necessary for more details. Systems to be considered should include pertinent portions of: Reactor cooling system Condensate and main feedwater systems Steam system Auxiliary feedwater system Reactor protection system Chemical and volume control system Emergency core cooling systems Instrumentation and control systems Support systems - Electric power D - Instrument air - Service cooling water 3

{ 2.2 Reactor Vessel Summarize the reactor vessel construction and its material properties. Use tables, drawings, or graphs to show: Vessel design (including weld locations and hot leg and cold leg penetrations). Vessel materials and chemical composition in the beltline region (including both base and weld material properties). Vessel fabrication procedures particularly welding and cladding. Vessel properties (e.g., RTNDT, initial RTNDT, appropriate fracture toughness data, including the upper-shelf regime, residual stresses, flaw density distribution, etc.). Describe and justify methods used to calculate or otherwise determine properties. Available information on the vessel properties should be reexamined in detail to fill any gaps in the supporting data for making an estimate of RT NDT and to support resolution of any disagreements about the validity of values used. Few data are currently available and validated to support the selection of a value for the initial RT The confidence that can be placed in estimates NDT. of the initial RT depends not only on material tests, but also on the accurate NDT documentation of welding technique, v L sire used, and weld flux used. The credibility of such estimates cea W er lanced by performing more tests on archival material, by discoverin; pre.. Aly unreported test results on weld specimens from the particular plant, or by evaluating properties of welds con-sidered typical of the plant-specific weld. 2.3 Fluence Present the current and projected fluence on the vessel using benchmarked computer programs and methodology and information from neutron flux surveillance dosimetry. Use the weld locations and fluence values to identify the critical welds. Show how the fluence varies along the length and depth of the critical welds. Describe the basis for these estimates and their uncertainty. 4

2.4 Inservice Inspection Results To the extent pertinent to the probabilistic analysis and proposed correc-tive actions, summarize: Results - The number, size, depth, and location of any flaws found should be well defined and described. Methods used - The method used to perform the inspection should be well described with documentation of any validation information. I Note: Only those inservice inspections (ISIS) that have actually been per-1 i formed should be discussed in this section. Improved ISI programs as proposed by the licensee should be described under corrective measures in Chapter 8, "Effect of Corrective Actions on Vessel Through-Wall Crack Frequency." l l 2.5 Plant Operating Experience Summarize overcooling transients that have occurred at this station or at D a sister station. Also, summarize lessons learned from these and other tran-sients, and indicate actions taken to prevent recurrence or minimize severity of overcooling transients.

2. 6 Operating Procedures This section provides procedural data, e.g., what the operator is supposed to do and when.

This section, for example, should present and describe the important operator actions as defined by existing procedures associated with potential overcooling transients. The conditions under which the operator takes each action, the expected time for performing the action, and how the time was derived should be identified. Some examples of these operator actions are: Trip reactor coolant pumps. Throttle / terminate emergency core coolant. Throttle / terminate main and emergency feedwater. I S 5

Restore main and emergency feedwater. Isolate break (primary or secondary). Supply a summary of training materials associated with overcooling events in general and with respect to principal initiators. In addition, a summary of. simulator exercises associated with potential overcooling events should be provided. Note: Proposed improvements in procedures, diagnostic instrumentation, display systems, and operator training should be presented in Section 8.2 under the li-censee's program of corrective measures. 6

D 3. DETERMINATION OF DESILED PTS SEQUENCES FOR ANALYSES This chapter presents the methods and analyses used to identify those transient sequences that could contribute significantly to the PTS risk. A good example is presented in Chapter 3 of Reference 3. The scope includes iden-tifying initiating events, developing event trees, modeling and quantifying the reliability of relevant systems and operator actions, and collapsing the event trees to identify specific relevant sequences. Detailed modelo, data, and sample calculations should be included in appendices or referenced. However, the logic of the analysis, criteria used, results, and insights gained should be described in the main report. 3.1 Approach Used Describe how the material presented in this chapter fits into the overall PTS study. Provide a general description of the process used to identify PTS sequences. It should be made clear how the approach used will result in com-D pleteness of identification of all c' lasses of events that could contribute sig-nificantly to PTS risk, how specific events are selected for more detailed anal-ysis to represent each class, and finally how the events so analysed are used to determine total PTS risk at the plant. 3.2 Sequence Delineation Identify potential overcooling transients in a well-defined manner, and document them in such a way that it is clear to a reviewer that all important potential overcooling conditions have been considered. Classes of initiators should be developed, important variations of initiators within each class should be identified, and potential transients resulting from these initiators should be defined. 3.2.1 Development of Classes of Initiators Any class of transients that could lead to overcooling of the reactor ves-sel should be considered in the analysis. It should, however, be appropriate to use logical arguments to eliminate classes of transients as actual PTS 7 1 \\ l

initiators whenever justifiable. Examples of initiators that should be included l are: Loss-of-coolant accidents (LOCAs), including steam generator tube rupture accidents Steam line breaks Overfeeds Combinations of these 3.2.2 Identification of Important Initiator Variations After the classes of potential initiators have been identified, it is im-portant to consider variations within any individual class. These variations should include: 1. Decay heat level - The decay heat level, determined by recent operating history of the plant, can have a major impact on the potential consequences of a given event. Thus, various decay heat conditions should be considered. l Clearly, decay heat associated with a reactor trip from full power (assuming operation at full power for some considerable time) should be examined. Zero decay heat represents the opposite extreme but for all practical purposes occurs only once at the beginning of life for the plant when PTS is not important. Therefore, the analyst may choose to use some other level of decay heat that would ccver potential decay heat conditions after the initial startup of the plant. The reasons for choosing particular decay heat levels for analysis should be documented. Each identified initiator should be examined at all decay heat levels defined whenever appropriate. 2. Power level - Power level may be important since certain equipment conditions or configurations may only exist at certain power levels, e.g., hot standby. As in the case of decay heat level identification, the reasons for the selection of specific power levels for analysis purposes should be stated. It should be noted that under certain conditions a reactor system may be at a high power level with a low decay heat condition. e 8

3. Location of event - In many instances the location of the event is defined. For example, an event consisting of a failed open turbine bypass valve has the location defined since it is a specific valve failure. However, for some events such as pipe breaks, the location is not defined and could have an l impact on the progression of the event. In the case in which location is not l defined, all locations that could be significant should be considered. Each location should then be eliminated by logical argument, bounded by consequences associated with another location, or treated as a separate event. 4. Magnitude of event - Many of the initiators can occur to various degrees. For example, a LOCA can range from a very small break to a full guil-lotine pipe break. Break sizes should be examined to identify categories of sizes that lead to similar system conditions. In the case of the LOCA event, special consideration should be given to the identification of break sizes that could lead to loop flow stagnation. The larger-sized LOCAs typically do not contribute to PTS risk since the pressure cannot be maintained because of the large flow out of the break. I 3.2.3 Definition of Potential Transients Resulting from Each Initiator After the complete set of significant initiators has been defined, event trees are required to identify potential sequences resulting from each initiator. The development of the event tree headings and branches should be done in a consistent and logical manner. This was done in the ORNL studies (Refs. 1, 2, and 3) by using what have been called system state trees. These trees define the potential states of each plant system of interest conditional on specific thermal-hydraulic conditions. Initiator-specific event trees can then be developed by examining the system state trees with respect to each initiating event. A similar or equivalent approach should be used to ensure traceability of the event trees and to ensure that important sequences are not inadvertently eliminated. Support system failures must also be presented within some type of event tree structure. If the event trees are developed as previously described, any support system failure would most likely lead to a sequence of events that is 9

already mapped out on the event trees, but in many instances with a higher probability of occurrence. In other cases, it may be necessary to define event trees resulting from a support system failure. In either case, it is important that the support systems be examined to identify their potential impact on over-cooling conditions. The results of this examination should be presented as a separate section with the identification of specific support system failure sequences of interest. The support system review should at least include: The electrical supply system The compressed air instrument system The component and service water systems 3.3 Operator Effects The operator effects are analyzed in two separate sections. In this section the potential operator actions are identified. These actions are further analyzed in Section 3.4 in which the probabilities associated with the performance of an operator action are developed. The operator can improve, aggravate, or initiate an overcooling transient. All three of these categories should be discussed in this section. 1. Procedures and/or the operators' general knowledge can lead to actions that improve the conditions associated with an overcooling event. Explana-tion should be included as to why it is perceived that this action would be taken. Where appropriate, these operator actions should be either included directly on the event trees or presented as separate operator action trees that can later be coupled with the principal event trees. 2. Although the ORNL studies (Refs. 1, 2, and 3) did not include operator-initiated events or events aggravated by operator actions contrary to proce-dures, this category of events should also be examined as part of a plant-specific analysis. 10

3. Events not normally considered to be overcooling events should be examined to identify the potential for operator actions resulting from misdiagnosis or operator error that could lead to an overcooling event. In addition, previously identified initiators should be analyzed for potential operator actions that could lead to a more severe cooldown. The confusion matrix approach (Ref. 5) used in human reliability analysis could provide a struc-ture for identifying and analyzing these potential operator actions. 3.4 Sequence Quantification Quantify the event trees by using identified initiating event frequencies, appropriate conditional probabilities associated with the success or failure of specific equipment operations, and success and failure probabilities asso-ciated with operator actions. Plant-specific data should be used whenever ap-propriate to define these probabilities, including appropriately adjusted simu-lator studies. This should be supplemented by vendor-specific or PWR generic data bases when plant-specific data do not appear to provide an adequate data base. Reference 6 includes guidance about treatment of generic and plant-spe-cific data. Its appendices include an updated generic data base that should be used. Identify by specific reference or provide in appendices all the reliability data used as input to quantify the event sequences. An explanation should be supplied as to how the data were derived for each data point. 3.4.1 Initiating Events Initiating event frequencies should be developed based on the number of observed events within selected periods of operation for the general type of plant under consideration. If no failures have been observed and no other information is available with which to estimate a probability, a standard sta-tistical method such as the Poisson distribution can be used to determine a probability, or the technique described in Appendix B to Reference 3 for esti-mating plant-specific initiating event frequencies can be used. For some initiators, it may be necessary to estimate the frequency of events in a par-I ticular operating mode, e.g., hot zero power. The data should be researched to 11

i identify trends associated with the occurrence of the event and the operating mode. Ir addition, the initiator itself should be examined to identify physical conditions that might favor failure in one mode rather than another. If this examination reveals no evidence of correlation between frequency and operating mode, the fraction of time spent in each operating mode can be used as a weighting factor. 3.4.2 Equipment Failures Following each initiating event, certain components are designed to perform in a defined manner. Failure of a component to perform its required function could lead to PTS considerations. Thus, it is necessary to assign a failure and successful operation probability for each component on a per-demand basis. These probabilities can be obtained by estimating the number of failures observed within a period of time, combined with an estimate of the number of demands expected within that same period, or by developing fault trees. If no failures have been observed and no other information is available with which to estimate a failure-on-demand probability, a standard statistical method as previously discussed can be used to develop a probability. As with all event trees, the probability associated with a particular branch is conditional on the prior branches in the sequence. Questions of con-ditional probability should be carefully considered before a failure probability is assigned. The potential for coupled or common cause failures must be examined in the analysis. Careful consideration should be given to increasing the failure potential of a component, given the failure of one or more components of the same type. As additional components of a particular type are postulated to fail, the probability for the next component of the same type to fail should increase. Based on the ORNL analysis, a simplified approach would be to assume that the failure probability of the second component, given that the first com-ponent has failed, might be as high as 0.1. The third component might be assumed to fail with a 0.3 probability, given the failure of two identical components. One could then assume that, after the failure of three components of the same 12

type, all remaining components of that type would fail wit probability of 1.0. The licensee should discuss how these types of coupled failures are handled in the analysis. Common cause failures of a different type may occur, as previously dis-cussed, through the failure of a support system or a control signal. An analysis of these poter tial failures should be made and the branch probabilities should be adjusted whenever appropriate. 3.4.3 Operator Actions Operator action probabilities are particularly difficult to determine be-cause of the lack of a data base. The problem is further complicated when time becomes an important variable. The procedure outlined below represents one approach to quantifying operator actions. This procedure should be conservative for any operator action performed as required by procedures assuming that the equipment required is operational. For operator actions that might not be associated with procedural steps, it is not clear that this simplified approach I would produce conservative frequencies. Therefore, the approach described would only be recommended for operator actions associated with procedural steps. Regardless of the method used, the human error probabilities used in these analyses should be supported by data validated for the plant being analyzed. 1. Identify operator actions - In this step the procedures associated with each initiator would be reviewed to identify those operator actions that would have an impact on downcomer temperature. 2. Identify time constraint - In the case of each operator action, the transient would be reviewed assuming no operator action to identify the time frame available for successful completion of the operator action. 3. Assign screening failure probabilities - In this step a conservative value for the failure of the operator action would be identified. For operator actions required by procedures to be performed within the first 5 minutes of the transient, the time reliability curve as presented in NUREG/CR-2815 (Ref. 6) l 13

could be used to identify a screening value. After 5 minutes, a value of 0.9 for success and 0.1 for failure would be assumed for all operator actions. The entire PTS analysis would then be completed using these screening values. 4. Identify dependency factors - In some instances, there may be coupled failures associated with operator actions just as there were coupled failures associated with equipment failures. In many instances, the potential failure of an operator action may be linked, to various degrees, to the success or fail-ure of a previous operator action. Thus, it is recommended that each operator action be reviewed with respect to dependency. This can be accomplished using the dependency tables as presented in the human reliability handbook (Ref. 7). 5. If any of the dominant sequences involve the failure of an operator action, a more comprehensive evaluation of the failure would be performed for that operator action. When necessary the comprehensive evaluation should be performed using one of the accepted human reliability methodologies. 3.5 Event Tree Collapse Collapse the event trees using a frequency screening criterion to form a list of specific sequences and a set of residual groups to be analyzed. This is important since the event trees may generate thousands of end states that cannot be individually analyzed. A screening value of 1.0E-7/ reactor year is recommended. This value should ensure that important sequences are treated individually, and it should also help to keep the size of the residual small. This is particularly important since it may be necessary to treat the residual l using a bounding consequence condition. 3.5.1 Specific Sequerces Those sequences that survive the frequency screening should be defined and their frequency noted. It is recommended that some identification be assigned to each sequence to enhance its traceability through the remainder of the anal-ysis. Grouping and identifying each sequence with respect to initiator type may also prove helpful. 14

3.5.2 Residual Groups Those sequences that do not survive the frequency screening must also be considered. They should be grouped together based on transient characteristics to form a set of residual groups. The residual groups should be reviewed to identify sequences that should be grouped with previously defined sequences be-cause of transient similarity or should be specifically evaluated because of their severe consequence. It is important to attempt to reduce the size of each residual group since it will be necessary to assign a bounding consequence that would apply within each group. Each residual group should be defined and its frequency noted. I i 15

1 4. THERMAL-HYDRAULIC ANALYSIS This chapter should present the reactor coolant pressures, temperatures, and heat transfer coefficients at the vessel's interior surface in the beltline region for the set of overcooling sequences that anvelops the plant's potential for experiencing a PTS event. A good example is presented in Chapter 4 of Reference 3. It should also present the details of the analysis methods used to obtain these fluid conditions. This chapter should include the following sections: 1. The thermal-hydraulic analysis plan and logic. 2. A description and evaluation of the thermal-hydraulic models. 3. A description of any simplified analysis methods used in the study. 4. A description of the methods used to evaluate the effects of thermal stratification and mixing. 5. Graphs of all the best-estimate thermal-hydraulic results with their associated uncertainties and a detailed explanation of the transient behavior observed. 4.1 Thermal-Hydraulic Analysis Plan This section should outline the logic and identify the subtasks in the thermal-hydraulic analysis. Subtasks include detailed thermal-hydraulic systems analysis, simplified thermal-hydraulic systems analysis, and thermal stratifica-tion analysis. The logic should describe the sampling plan used to select sequences for detailed or simplified analysis. ORNL experience favors selecting detailed thermal-hydraulic analysis sequences, including at least a few severe examples of each type of postulated overcooling transient in order to understand and benchmark the plant behavior for subsequent simplified calculations. The order in which the scenarios are evaluated can result in a considerable reduction in expenditures. By first analyzing the scenarios that are expected to be the bounding cases (i.e., the most severe), calculations for an entire class of 16

overcooling scenarios may be deemed unnecessary if the bounding case is not of PTS concern. Similarly, careful selection of the first set of scenarios to be evaluated can permit simple extrapolation or interpolation of the results to other scenarios that share common controlling thermal-hydraulic phenomena. During the analysis, the sequence identification analyst and the thermal-hydraulic analyst should coordinate activities to ensure that pertinent details of the delineated sequences are thoroughly understood. Similarly, close coor-dination must be maintained between the thermal-hydraulic analyst and the fracture mechanics analyst so that the transient fluid conditions are calculated at th2 appropriate vessel locations.

4. 7.

Thermal-liydraulic Model This section and supporting appendices should present a detailed descrip-tion of the thermal-hydraulic computer models used in this analysis. The models should include an accurate representation of the pertinent parts of the primary and secondary systems. This includes the condensate system, the main and auxiliary feedwater systems, and parts of the steam system. The model should include appropriate secondary side metal heat capacity. Particular attention should be given to the modeling of control system logic and characteristics such as valve closure times and liquid level measurements. References 8 through 11 illustrate some of the modeling details included in such a study. The thermal-hydraulic models should be capable of predicting single and two phase flow behavior and critical flow as required. The models should be capable of predicting plant behavior for LOCAs, steam line breaks, and steam generator tube ruptures. In general, a one-dimensional code is suitable for most overcooling transient calw1 stions. However, if any of the control systems are dependent solely on the fli.id conditions in a single loop (e.g., reactor coolant pump restart criteria), a method of estimating the three-dimensional effects in the downcomer may be necessary for some of the asymmetric cooldown scenarios en-countered in the PTS study. Sensitivity of calculated results to the nodaliza-tion schemes used should be discussed. This section of the report must also present the results of benchmarking I the computer models against suitable plant data or data from experimental 17

facilities. As a minimum, the plant data comparison should fully exercise the modeling features that are employed in the thermal-hydraulic computer programs such as the pressurizer sprays, heaters and liquid level controls, the steam generator liquid level controls, and the turbine bypass (i.e., steam dump) con-trols under steady-state and transient conditions. If overcooling transients have occurred at the plant or at a similar plant, they should be benchmarked against the computer models. The licensee is encouraged to use codes and methods accepted by the NRC at the time the calculation is performed. The models should be capable of accurately predicting steam condensation in the pressurizer during the repressurization phase of an overcooling transient. Effects of noncondensible gases, if present, on system pressure and temperature calculations should be addressed. All code input and modeling assumptions should be documented in an audit notebook available for NRC review. 4.3 Simplified Analysis Methods This section should present the technical bases for any simplified analysis methods that are applied in the study. This includes the grouping of similar sequences by controlling phenomena and any extrapolations used to modify existing calculations. If a simplified thermal-hydraulic plant model is used to predict portions of the plant transients, all the simplifying assumptions inherent to this model must be stated and justified. Reference 12 provides examples of how to group sequences and develop a simplified thermal-hydraulic model suitable for portions of the analysis. 4.4 Thermal Stratification Effects Transient thermal-hydraulic computer programs available to analyze LWR response to overcooling scenarios do not model fluid behavior with sufficient detail to predict the onset of HPI thermal fluid stratification in the cold leg ana the subsequent cold leg and downcomer behavior. As a result, additional analysis methods may be needed to determine which transients are affected by thermal stratification and the extent of such effects. 18

h This secti n sh uld describe and justify the thermal fluid mixing analysis F methods that have been applied in the study. References 13 through 18 describe the results of recent mixing analyses and experiments. Reference 13 identifies a useful stratification criterion to determine which overcooling transients will require the additional mixing analysis. Particular attention should be given to scenarios that involve HPI under very low flow or stagnant loop condi-tions. When stagnation is partial (i.e., not all loops stagnate), stratification is expected.)nly within the cold legs of the stagnant loops. However, scenarios invoiving complete loop stagnation will require the evaluation of a transient cooldown in the presence of stratified layers both in the cold legs and in a portion of the downcomer. The mixing model should include the effect of metal heating on the mixing behavior, particularly in a stagnant flow situation.

Also, the effect of noncondensible gases, if present, must be included.

References 13 through 17 describe tools that have been used for such an analysis. This section should also document the heat transfer correlations applied in the mixing analysis. The research efforts described in References 13 through 18 indicated that the downcomer heat transfer coefficients generally exceeded 300 Btu /hr-ft2_oF. These values of heat transfer coefficient were generally high enough to keep the vessel wall surface temperatures within a few degrees of the downcomer fluid temperature. Furthermore, because the vessel wall cool-down was controlled by conduction processes rather than convection processes, the vessel wall surface temperatures were insensitive to heat transfer coef-ficient variations due to changes in flow and heat transfer regimes. 4.5 Thermal-Hydraulic Analysis Results This section should present graphs of the best-estimate downcomer pressures, fluid temperatures, and heat transfer coefficients and their associated uncer-tainty ranges as a function of time at the critical weld areas. This includes the results of the detailed thermal-hydraulic model, the simplified model, and mixing analysis calculations. The duration assumed for each overcooling scenario must be justified. It is assumed that a scenario duration of 2 hours may be reasonable for many D cases since the overcooling transient would probably be identified and mitigated l 19

prior to that time. However, there may be scenarios requiring lengthier eval-uation periods because the controlling phenomena delay the scenario's evolution. Also include a discussion of the accuracy of the results and how the pre-dicted plant behavior compared to plant history and operator experience. Time-dependent uncertainty estimates for the downcomer pressure, fluid temperature, and heat transfer coefficients at the critical welds must be provided for each scenario. These uncertainties are often limited by physical phenomena. For example, the pressurizer power-operated relief valve (PORV) setpoints will limit the system pressure for certain high pressure scenarios. Therefore, the uncer-tainty is limited by PORV operating characteristics. Referencas 10 and 12 de-scribe some uncertainty analysis techniques. l 8 20

5. FRACTURE MECHANICS ANALYSIS For each sequence identified in Chapter 3, " Determination of Detailed PTS Sequences for Analyses," calculate (or for unimportant segments estimate, using bounding conditions) the conditional probability of through-wall crack penetra-tion given the occurrence of the event vs. effect ve full power reactor years. A good example is provided in Chapter 5 of Referent 3 3. Input for these calcu-lations includes the primary system pressure, the te perature of the coolant in the reactor vessel downcomer, the fluid-film heat trans'.** coefficient adjacent to the vessel wall, all as a function of time, and the vessel properties. Although licensees were required to use the method of determining RTNDT (RTPTS) specified in paragraph 50.61(b)(2) when evaluating their vessel properties with respect to the screening limits, in performing these plant-specific calculations, they are encouraged to use any alternative methods / data / correlations for which they provide justification of applicability to their specific plant. The cal-culations should be performed with a probabilistic fracture mechanics code such as OCA-P or VISA (Refs.19 and 20). } An acceptable procedure to be followed in the fracture mechanics analysis is as follows: A one-dimensional thermal and stress analysis for the vessel wall should be performed. The effect of cladding should be accounted for in both the thermal and stress analyses. The fracture mechanics model can be based on linear elastic fracture mechanics with a specified maximum value of KI, to account for upper-shelf behavior. Plastic instability should be considered in the determination of failure. Acceptable types of material properties are given in the study of the H. 8. Robinson reactor (Ref. 3). In the Monte Carlo portion of the analysis, as a minimum, each of the following must be assigned distribution functions: KI = Static crack initiation fracture toughness c KI,= Crack arrest fracture toughness NDT = Nil-ductility reference temperature Cu = Concentration of copper, wt-% Ni = Concentration of nickel, wt-% F = Fast neutron fluence 21

The functions used must be justified. Examples of these distributions are found in Reference 3. The following additional information must be supplied: 1. Flaw density - The number of cracks per unit surface area must be es-t blished for use in the calculations and must be justified. A value of 0.2 flaw per square meter (one flaw / cubic meter) was selected in References 1, 2, and 3. 2. Flaw depth density functic1 - The flaw depth density distribution must b2 established. The function to be used can be that specified in References 1, 2, and 3. 3. Flaw size, shape, and location - Axial flaws with depths less than 20 percent of the wall thickness and all circumferential flaws should be modeled in two dimensions. Axial flaws with depths greater than 20 percent of the wall thickness should be modeled in two or three dimensions depending on the relative trughness of the weld regions and plate material. For instance, the length of { an axial flaw in an axial weld that suffers severe radiation damage relative to the plate can be Ifmited to the length of the weld. 4. All regions of the beltline must be considered. This includes axial and circumferential welds as well as the base material. Th] following relationships are required: KI = f(T,RTNDTo,dRTNDT), and c KI, = f(T, RTNDTo,dRTNDT) where T = Wall temperature RT = Initial nil-ductility reference temperature l NDio 22

dRT = Increase in nil-ductility reference temperature due to radiation NDT damage, f(Cu,Ni, fluence). If plant surveillance data meet the criteria for credibility given in Reference 21, they may be used as described therein. Examples of these functions are described in References 3 and 21. In reporting the results, the methods used for the probabilistic vessel-integrity analysis should be described, theli limitations for this analysis identified, and the impact of uncertainties in J1e resulting vessel failure probabilities estimated. ~ Discussion of the analy.*s should include a listing of the assumptions used, their bases, and a discussion of the sensitivity of the results to variations in the assumptions. Vessel dimensions and material properties used should be given. The results of each transient should include a set of critical-crack-depth curves (see Refs. 1, 2, and 3), i.e., a plot of crack depths corresponding to initiation and arrest events vs. time. This plot should also have curves indi-I ccting when warm prestressing is effective, the depth at which upper-shelf tough-ness is effective, and the depth at which upper-shelf toughness is reached. These results should correspond to -2(sigma) values for KI and KI,, +2(sigma) c values for RTNDT, mean values of all other parameters, and an appropriate number cf effective full power years (i.e., corresponding to the RT screening cri-NDT t;rion or greater). D 23

6. INTEGRATION OF ANALYSES l In this chapter, the event frequencies are coupled with the results of the fracture mechanics analysis to obtain an integrated frequency of vessel through-wall cracking due to PTS. An example of one acceptable method is presented in Chapter 6 of Reference 3. A table that supplies the following information for each specific sequence and residual group identified in Section 3.5 should be provided. These results should be provided for the operating time at which the reactor will reach the PTS screening criterion and for any additional operation life being requested: Sequence identification. Type of initiator (small-break LOCA with low decay heat, large steam line break at full power, etc.). Estimated sequence frequency. Method used to determine conditional through-wall crack penetration probability. Sequence conditional through-wall crack penetration probability.* Frequency of through-wall cracking due to sequence obtained by the l product of sequence frequency and sequence conditional through-wall crack penetration probability. For each dominant sequence, a section or table should be provided that sup-1 plies (1) specific reference to the graph of temperature, pressure, and flow as provided in Chapter 4, " Thermal-Hydraulic Analysis"; (2) a time-line description of the accident sequence noting important operator actions, control actions, protection system actions, equipment faults, and vessel failure; and (3) fre-quency of through-wall crack penetration as a function of effective full power years. Results should then be summed within each initiator type to provide a fre-quency of through-wall crack penetration as a function of initiator type.

• The conditional through-wall crack penetration probability is the probability of a through-wall crack as determined by the fract.re mechanics analysis, given that the event occurs.

24

h dc" "' " " "'d

• a' '" *"x c" '" ' " 'vae " " ' '=a r-F tant to PTS.

Finally, the results should be summed over each initiator type to provide an integrated frequency of through-wall cracking for the vessel. This integrated value should be reported as a function of effective full power years and plotted with uncertainty values as determined in Chapter 7, " Sensitivity and Uncertainty Analyses of Through-Wall Crack Frequency," and included on the plot. The dis-cussion should identify important operator actions, control actions, and plant features that can cause or prevent vessel failure. I D 25

7. SENSITIVITY AND UNCERTAINTY ANALYSES OF THROUGH-WALL CRACK FREQUENCY In order for the results of the probabilistic analysis to be useful for regulatory decisionmaking, the sensitivity of the results to input parameters and assumptions should be determined, the major sources of uncertainty should be identified, and the magnitude of the uncertainty should be estimated. In this chapter, the results and the procedures used to perform each of these processes are documented. A good example is given in Chapter 7 of Reference 3. 7.1 Sensitivity Analysis Perform a sensitivity analysis to estimate the change in the through-wall crack frequency for a known change of a single parameter. Parameters examined in the sensitivity analysis should include (1) the initiating event and event tree branch frequencies, (2) the thermal-hydraulic variables (temperature, pres-sure, etc.), and (3) the fracture mechanics variables (fluence, flaw density, etc.). Where appropriate, 68th percentile (1-sigma) values should be used to represent the change in the parameter. This should provide a sufficient change l to illustrate the effects of the change, and the use of the 68th percentile value whenever possible will help to define the important variabilities. In the case of temperature and pressure, however, the 68th percentile values may vary from one sequence to another. In this case, it may be easier to identify a representative change in the parameter that could then be used for all sequences rather than to try to use the 68th percentile values. Each variable examined in the sensitivity analysis should be listed along with the change in the variable. In the cases in which changes are represented by using 68th percentile values, some explanation should be provided to document the reasons the value is considered a 68th percentile value. In those cases in which something other than a 68th percentile value is chosen, discussion should center around the reasons for choosing the value used. Sensitivity factors should be obtained by dividing the through-wall crack frequency obtained with the changed variable by the throu0h-wall crack frequency obtained with each variable at its mean value. Supply the sensitivity factors obtained for both positive and negative changes in each of the variables. The I 26

sensitivity factors obtained for changes made in the PTS-adverse direction should F be ranked according to magnitude and provided in table form. 7.2 Uncertainty Analysis 7.2.1 Parameter Uncertainties 1 Each step in the probabilistic analysis should include an uncertainty anal-ysis. This should include uncertainty in frequency of occurrence of a sequence, uncertainty in temperatures and pressures reached during the sequence, and un-certainty in the fracture mechanics model for vessel failure given the transients. For the following reasons, a Monte Carlo simulation is appropriate for PTS uncertainty analysis. The temperature and pressure error distributions are not symmetric. The fracture mechanics results are nonlinear with respect to I variations in input parameters, particularly the temperature and pressure time histories. The results of the Monte Carlo analysis c.*q indicate the shape of the output distribution. The Monte Carlo approach would involve four steps as described below: 1. Develop a statistical distribution for each variable used in the calculation - This step will involve the representation of each variable as a distribution with 5th and 95th percentiles as previously identified. The shapes of the distributions selected should be discussed. 2. Select a random value from each distribution - A random sampling code should be used to sample from each of the distributions. 3. Calculate a through-wall crack frequency estimate based on values D obtained in the previous step - In this step, the through-wall crack frequency 27

is obtained based on the randomly selected variables. This requires under-standing the form of the relationship between each input variable and through-wall crack frequencits. For some variables such as initiating event and branch frequencies and flaw density, this is simple since the through-wall crack frequency is directly proportional to the value of these parameters over the range of variable values considered. Other variables such as temperature and pressure may require the development of an appropriate relationship. In such cases in which the effect of a variable change may be dependent on the value of another variable, response-surface techniques may be used to estimato important I interaction effects. 4. Summarize the resulting estimates and approximate frequency distribu-tion - Steps 2 and 3 are repeated until a statistically valid number of trials have been performed. A distribution of through-wall crack frequencies is then produced from the results of the trials. The 95th and 5th percentiles and the mean (expected value) of this distribution should be identified and discussed. 7.2.2 Modeling Uncertainties (Biases) l During the process of performing the PTS analysis, the analyst will make simplifying assumptions in order to make the analysis tractable. Such assump-tions include decisions on thermal-hydraulic models, fracture mechanics models, grouping of sequences both for thermal-hydraulic analysis and fracture mechanics analysis, etc. These assumptions can introduce conservative or nonconservative biases into the analysis. These biases should be identified and their potential impact on the results discussed. In this section important assumptions made as part of the analysis should be listed. Each assumption should be identified as being either conservative or nonconservative. A discussion should be supplied for each assumption with respect to its impact on the overall value of through-Wall crack frequency. Whenever excess conservatism or nonconservatism is sus-pected to be present in an assumption, an alternative assumption should also be used in the full calculation procedure and the impacts on the overall result con. pared. l 28

8. EFFEC LOF CORRECTIVE ACTIONS ON VESSEL THROUGH-WALL CRACK FREQUENCY This cha,)ter summarizes the licensee's program of corrective measures. Each corrective measure considered by the licensee should be presented and ex-plained in this chapter. In each case, the reasons for considering the action as a corrective measure should be documented, and the estimated impact of the I action with respect to through-wall crack frequency should be provided. Cor-rective actions that should be considered include, but are not limited to, those discussed in the remaining sections of the chapter. An example can be found in Chapter 8 of Reference 3. 8.1 Flux Reduction Program Early analysis and implementation of such flux reductions as are reasonably practicable to avoid reaching the screening criterion are already being required and accomplished in accordance with the PTS rule, S 50.61. Further flux reduc-tions to critical areas of the vessel wall that would reduce the risk of con-tinued operation beyond the screening criterion should be considered. If such I additional flux reductions are needed, in view of the irreversibility of embrittlement, the licensee shou?d consider early implementation before reaching the screening criterion. For licensees who are considering applications to extend the operating license beyond its present expiration date, it may be pru-dont to implement the reduction as early as possible to avoid the necessity of vessel annealing or replacement. 8.2 Operating Procedures and Training Program improvements Operator actions and associated plant response play a key role in the initiation and mitigation of PTS events. Therefore, ensure that the actions are based on approved technical guidelines that include an integrated evaluation of relevant technical considerations, including, but not limited to, PTS, core cooling, environmental releases, and containment integrity. The evaluation should address the following types of concerns: Frequent realistic " team" training should be conducted, exposing the I operators to potential PTS transients and their precursor events. 29

-m The training should give the operators actual practice in controlling reactor system pressure and cooldown rates during PTS situations. Specific training thould include, but not be limited to, reactor cool-ant pump trip criterion, HPI throttling (.riterion, co,ntrol of natural circulation, recover'y from inadequate cott cooling, recovery from solid plant operations, and use of PORVs to c'tntrol primary overpressure. \\ Instructiot.s should be based on analyses that include consideration of system response delay times (e.g., loop transport time, thermal transport time). Whether or not there is a need for coobiown rate limits for periods shorter than 1 hour should be evaluated. \\ Methods for controlling cooldown rates should be provided. Reference should be made to these methods with respect to the dominant PTS risk sequences whenever possible. l Guidance should be provided for the operator if cooldown rates or pressure-temperature limits are exceeded. These guidelines should take into account potential core cooling, environmental release, or ', containment integrity problems that could exist as a result of respond-ing to the abnormal cooldown rate. These guidelines should leave little doubt as to when PTS concerns cre more important than other safety issues and when other safety issues assume primary importance .s over PTS concerns. s The desired region o'f operation betwee.1 the pressure-temperature limit and the limit determined,by avoidance of saturation conditions should be evaluated to determSAe if it can be revised to minimize total risk due to plant operption fv6m PTS plus non-PTS events. - t Instructions for controlling pressure following depressurization transients should be provided. l 30 s

h '" " ""c "' '" "'d "v"" "* ' " '"e c "d " *"er* ""'"r"' F circulation is lost and the primary system main circulation pumps are not available. 8.3 Inservice Inspection and Nondestructive Examination Program The use of state-of-the-art nondestructive examination (NDE) techniques could provide an opportunity to decrease any conservatism that might exist in the flaw density value used in the analysis. This decrease in conservatism, however, may be less important than the decrease in uncertainty in the actual flaw density that may result from an examination of this type. Existing inservice inspection programs should be reevaluated to consider incorporation of state-of-the-art examination techniques for inspecting the clad-base metal interface and the near-surface area. This includes plant-unique consideration of the clad surface conditions. Consideration should be given to increased frequency of inspections. The reliability of the NDE method selected to detect small flaws should be documented. 8.4 Plant Modifications Plant modifications that may be considered include the following: 1. Instrumentation, Controls, and Operation a. Reactor vessel downcomer water temperature monitor b. Instantaneous and integrated reactor coolant system cooldown rate monitors c. Steam dump interlock d. Feedwater isolation / flow control logic e. Reactor coolant system pressure and temperature monitors f. Monitor to measure margin between vessel inner-surface temperature and current RT at that location D NDT g. Diagnostic instrumentation and displays 31

s h. Primary coolant system pump trip logic i. Automatic isolation of auxiliary feedwater to broken steam lines / generators 2. Increased Temperature of Emergency Core Cooling Water and Emergency Feedwater If plant modifications are proposed to prevent overcooling, the report should include an evaluation of undesirable side effects (i.e., undercooling) and a discussion of steps planned to ensure that the modifications represent a net improvement in safety when PTS and non-PTS related events are considered. 8.5 In Situ Annealing If in situ annealing is part of the licensee's program of corrective mea-sures, the licensee should describe the program to ensure that annealing will achieve the planned increase in vessel toaghness, the surveillance program to monitor vessel toughness after annealing, and the program to ensure that annealing does not introduce other safety problems. L 4 32

9. PREDICTION OF VESSEL FAILURE MODE In developing the PTS rule (S 50.61), the NRC staff used analyses that pre-dicted the expected frequency of through-wall crack penetration due to PTS events. It was implicitly assumed that a through-wall crack would likely cause a core melt, which was assumed to be an unacceptable event that should be pre-vented, regardless of the resulting risk in terms of person-rem, fatalities, etc. Analyses that were performed subsequent to the rule development tended to confirm the assumption that a through-wall crack is indeed likely to cause a vessel failure massive enough to cause core damage (Ref. 22 provides an example of these calculations). The acceptance criteria described in the Introduction to this guide state that analyses using acceptable techniques such as those outlined in this guide and resulting in a PTS-related through-wall crack penetration frequency less than 5 x 10 8 per reactor year are acceptable without further analyses of risk. However, if continued operation is requested and the predicted through-wall crack frequency is greater than 5 x 10 6 per reactor year, consequence and risk I calculations are also necessary to judge the acceptability of continued plant operation above the screening limit. The consequence and risk analyses appro-priate are described in this chapter and in Chapter 10. " Likelihood of Core Melt and Prediction of Risk." This chapter should present methods that can be used to determine the mode of vessel failure that is expected to occur when a crack penetrates the wall of the vessel. The results to be obtained include the final length and opening area of the crack, the size and velocity of possible missiles, and predictions regarding whether such missiles will be arrested. These results are needed to predict the core melt conditional probability given through-wall crack penetra-tion, which in turn is needed as input to the risk calculation presented in Chapter 10. Input for these analyses include the vessel wall temperature distribution cnd pressure at the time of occurrence of the through-wall crack, the probability cf existence of such cracks for each weld in the vessel, initial toughness pa-rameters for each weld and plate in the vessel, and the fluence at each location 33

in the vessel. All major contributing seqt.ences to vessel failure should be addressed. The output of the analyses will be the conditional probability for each of the significant failure modes. Examples of such modes include the range from crack arrest at the end of the affected axial weld to extension of the axial crack into adjacent plate material to a complete fracture of a circumferential weld. The vessel failure mode calculations should be performed with a probabil-istic fracture mechanics code such as described in Reference 22. As an alter-native, the licensees may conservatively assume the failure mode that has the greatest consequences on predicted risk. The following methods and assumptions are described in more detail in Reference 22 and are acceptable to the NRC staff: The initial length of all axial cracks is the full length of the axial weld. All circumferential cracks should extend the full 360 degrees around the vessel unless an analysis can justify arrest at a shorter length. The initial state of each through-wall crack in the vessel should be treated as a running crack that must be arrested to preclude further crack propagation. Thermal stresses can be neglected in analyses of propagation of through-wall cracks after the crack has initiated. The effect of fluence and temperature gradients can be treated by calculating an average toughness for the vessel wall by a root mean square approach. l 34

The effects of fluid structural interactions and structural dynamics for opening cracks should be considered. However, simplifying approx-imations as described in Reference 22 may be used. The effects of irradiation on ductile brittle transition temperature, upper-shelf toughness, and flaw stress should be included in any ana-lyses. The methods should be consistent with those used to predict the growth of part-through flaws. Elastic plastic fracture mechanics should be used for growth predic-tions of through-wall cracks in the vessel. Acceptable methods and approximations are described in Reference 22. The possibility that axial cracks may turn and follow circumferential welds should be part of the failure mode analyses. In the calculations of probabilities for vessel failure modes, the random variations in the parameters governing material toughness can I be assumed to be independent for the different welds and base metal that make up the vessel. The masses and velocities of vessel fragments that become missiles should be estimated. Analyses should be performed to determine if such missiles can escape from the vessel cavity and if such missiles can cause loss of containment integrity. Core movement and deformation caused by missiles should be examined to evaluate core damage and prcbability of core melt. The methods and assumptions used to predict vessel failure modes should be described. The output of the analyses should describe the location and size of the final opening of the through-wall crack. The licensee should document the sources of vessel material characteristics D such as chemistry and value of toughness properties for the irradiated vessel. 35

Uncertainties in these properties should be discussed. Conservative values { should be assumed when data are lacking. Analyses have shown that axial cracks may very often turn and produce a fracture of circumferential welds. In the analyses of vessel failure modes, crack tip stress intensity factors have been estimated for those cracks that may turn and follow a circumferential weld. The stress intensity factor decreases by a factor of two when the crack changes direction from axial to circumferential. Fluid-structural interactions can be approximated in the analyses of reac-tor pressure vcssels by the use of static stress intensity factor solutions. However, one must assume that the full pressure remains during the static crack opening process. For the long through-wall cracks of concern to the failure mode analyses, the use of elastic plastic fracture mechanics is required. When cracks run beyond the irradiated zone into ductile materials with good upper-shelf tough-l ness, crack arrest can often be predicted if one takes credit for the increased tearing resistance associated with stable crack growth. An important part of the evaluation is to assemble data that describe the chemistry and initial toughness of all welds and base metal in the vessel shell. (These data are generally available to the licensee or through industry data bases.) When vessel-specific data are lacking, it will be necessary to make conservative estimates based on data available for similar vessels. It will be difficult to estimate the size of vessel fragments that can become missiles. For horizontal missiles, it may be possible to apply ballistic penetration equations to predict missile arrest by the vessel cavity concrete. It may be found that the upper bound velocities for the spectrum of missile sizes show consistent arrest. The other class of missile of concern is that of the upper-head assembly. For these missiles, the restraint provided by the attached primary coolant piping may be the major factor contributing to arrest of the missile. l 36

The prediction of vessel f1ilure modes may be performed on a probabilistic basis using Monte Carlo simulations. The approach may use the probabilistic fracture mechanics codes for prediction of through-wall cracks as a guide. For the failure mode, the calculations must involve through-wall rather than surface cracks. The toughness of each weld and plate in the vessel should be simulated. The step changes in toughness from material to material are the basis for pre-dicting the arrest of the through-wall crack. i I D 37

l 10. LIKELIHOOD OF CORE MELT AND PREDICTION OF RISK l 10.1 Risk Analysis A factor to be considered in determining the acceptability of operation beyond the screening criterion for plants with a predicted through-wall crack frequercy greater than 5 x 10 8 per reactor year will be the estimatad risk in terms of the frequency of core melt due to PTS-induced failures, mul.iplied by the consequences in terms of public exposure to radiation. To calculate risk in these terms, one must understand the core melt frequency due to PTS-induced failure, the effectiveness of containment, the behavior of radionuclicbs within the primary system and containment, and the transport of radionuclides beyond the site boundary. This chapter gives guidance for the estimation of those quantities. 10.2 General Guidance The purpose of severe accident analysis is to obtain a realistic estimate of risk. Consequently, it is appropriate to use best-estimate methods, data, I and assumptions. However, because many of the phenomena are not well understood, it is necessary to account for the impact of analysis uncertainties on the con-clusions being drawn from the analysis results. Analyses should be plant and site specific. Generic results may be used if the generic data are shown to be applicable or bounding for the plant under analysis. Many of the phenomena that govern severe accident risk are currently under-going intense study by the NRC Severe Accident Research Program, the Electric Power Research Institute, and foreign research programs. Because no existing plants are expected to approach the screening criterion in the near future, it is likely that much of the ongoing research will have been completed when the first PTS risk calculation is performed. Better information and calculation methods may therefore be available for use in those first calculations. The discussion below describes the methods and assumptions that are currently ac-ceptable to the NRC staff and identifies areas where ongoing research is likely 38

to change current methods. The licensee should use the methods and assumptions accepted by the NRC at the time of the analysis in order to minimize the review and approval process. 10.3 Core Melt Frequency The frequency of through-wall cracks in the reactor vessel should be deter-mined from thermal-hydraulic, metallurgical, and fracture mechanics analysis according to the guidance described in the sections above. The frequency of through-wall vessel cracks should be estimated as a function of crack location and size, including the case of a complete crack around a circumferential weld. Based on break location and size, the licensee should determine whether or not the pressure vessel will depressurize because reactor vessel pressure can affect containment performance. The licensee may choose to assume that all through-wall cracks lead to core melt. Alternatively, best-estimate analyses of the emergency core cooling I system performance may be submitted to demonstrate that one or more classes of cracks would not lead to core melt. Based on these analyses, the licensee should determine the frequency of core melts (per reactor year). Given a core melt, the timing and mode of con-tainment failure depend on the nature of the accident sequence and the status of containment protective systems at the time of core melt. Together, these factors determine the plant damage state. For PWRs, the principal containment safety feature is containment heat removal by sprays and fan coolers. Operation of the sprays or fans can delay or prevent containment overpressure failure. Furthermore, the sprays, and to a lesser extent the fan coolers, can reduce the suspended fission product inventory in containment. In determining the probability of fan and spray operation, the licensee should account for multiple independent failures and common cause failures such as sump blockage or clogging of the fan cooler heat exchangers by aerosols. The staff's current perception is that operation of I 39

sprays and/or fans is highly likely for those sequences involving potential PTS Concerns. The possibility exists that containment will be bypassed by a consequen-tial break in an injection line outside containment or by a steam generator tube rupture. The former may be postulated to result from violent motion of the primary coolant system during blowdown via a through-wall crack. The latter can also result from primary system motion but, in addition, may have been the transient that initiated the PTS event. There is also preliminary evidence that tube ruptures could result from the superheated steam produced during core meltdown. All plausible mechanisms should be accounted for in estimating the likelihood of containment bypass. The staff's current perception is that containment bypass in a PTS event is unlikely. j 10.4 Containment Failure Given plant damage state frequencies, the timing of containment failure and the magnitude of the radiological release depend on the response character-istics of the containment. These characteristics are numerically embodied in the containment failure matrix (C-Matrix). Determining the C-Matrix requires a best-estimate analysis of the 'ailure pressure of containment. In general, this pressure will be higher than the design pressure. Threats to containment integrity can come from many sources, including missiles, steam explosions, direct heating, failure to isolate, gradual over-pressurization, hydrogen burns or detonations, preexisting or induced leakage, basemat meltthrough, and the bypass modes discussed above. Each mode of failure should be assessed for each plant damage state. The probability of gradual overpressure failures and hydrogen burns should be based on plant-specific cal-culations of accident phenomenology in the reactor vessel and containment, using a methodology acceptable to the staff at the time the analyses are performed. l 40

i The probability of leakage or failure to isolate can also be based on generic results and should account for operational experience with containment isolation during normal operation. Basemat meltthrough should be assessed with a model acceptable to the staff at the time the analyses are performed for core-concrete interaction. The output of the containment failure matrix is a set of probabilities of release categories that characterize the timing of containment failure and severity of the radiological release. In the WASH-1400 study, these core melt release categories are designated PWR-1 through PWR-7. The contributions from several plant damage states and containment failure modes can be lumped into a single release category if their characteristics are similar. 10.5 Source Term The radiological source term associated with each release category speci-fies the fraction of each type of radionuclide released, the energy of release, the chemical form, the particle size, the release time, the warning time, and the duration of release. Calculations of these parameters are based on estimates of the rate of release of fission products from the fuel, the transport of radionuclides in the reactor vessel and containment, and various mechanisms for deposition and retroval from the atmosphere. Currently, there is a major research cffort devoted to a better understanding of the radionuclide release fractions and release characteristics. It is possible in the future that the NRC will adopt a new methodology. For the purpose of risk calculations for PTS, the licensee should use the source term methods and assumptions that are acceptable to the NRC at the time of the analysis. 10.6 Site Consequence Analysis For each radiological release category, an assessment of offsite consequences should be performed. The CRAC-2 code for consequence calculations is an example of an acceptable methodology. It has been tested in numerous applications and l 41

subjected to international peer review. The licensee should use a code that { embodies an acceptable set of models for fission product transport and deposition, emergency actions, and dose response. The licensee is encouraged to use codes and methods accepted by the NRC at the time the calculation is performed. The licensee should use plant-specific data for weather and other site characteristics. A site-specific population distribution should be used for the year with maximum population during the remaining projected life of the plant. Consequences should be integrated over an appropriate area. The licensee must justify the choice of warning time, emergency response delay time, radius of limitation, and speed of evacuation. These should be in agreement with an evacuation plan and model accepted by the NRC at the time the PTS analyses are performed. The consequence calculation should be performed for numerous weather condi-tions and averaged based on the actual plant meteorology. Results should in-q clude conditional point estimates of early fatalities, latent cancer fatalities, and public radiation exposure (in person-rems) for each release category. Results should be consistent with NRC acceptance criteria in effect at the time of the PTS analyses. 10.7 Risk The risk contributions for each release category should be calculated as the product of the release category frequency per reactor year and each of the three consequences determined, i.e., early fatalities, latent cancer fatalities, and person-rems. The overall PTS core melt frequency and risk is the sum of the release category contributions and should be calculated using models approved by the NRC at the time the analyses are performed and should be compared to acceptance criteria in effect at the time. l 42

10.8 Uncertainties The uncertainty in risk reflects the uncertainties in every aspect of the sequence of calculations, including the through-wall crack frequency, vessel failure mode, core melt frequency, containment response, radiological release fractions, and offsite consequences. The licensee is expected to make reason-able estimates of the uncertainties associated with each phase of the calcula-tion. Based on those uncertainties, the licensee should place reasonable upper end lower bounds on the risk estimates. I 43

11. RESULTS AND CONCLUSIONS REGARDING PTS RISK This chapter should summarize the models used and the results obtained and should provide the conclusions reached with respect to continued cperation of the plant. 11.1 Summary of Analysis In this section the major findings of each aspect of the PTS analysis, as described in the previous chapters, should be presented. These should include: Expected (mean) value of frequer.cy of reactor vessel through-wall crack penetration vs. time, with uncertainty bound '95th percentile). Identification of dominant accident sequences. e If sensitivity / uncertainty analysis shows that slightly different assumptions could lead to different dominant seodences, identification of these assumptions and discussion of the impact on results given the different assumptions. Identification of important operator actions, control actions, anc ] plant features that can increase or decrease the frequency or consequences of overcooling transients. Major sources and magnitudes of uncertainty in the analysis. The relative effectiveness of potential alternative corrective measures in reducing the expected (mean) value of through-walt crack penetration. The program of planned corrective measures. For cases in which the predicted through-wall crack penetration frequency is greater than 5 x 10 6 per reactor year, this should also include: 44

D Expected (mean) value for risk from PTS-related core melt, person-rems, and fatalities. Identification and explanation of differences if dominant sequences are different for through-wall crack, core melt, and early release of major radioactivity. The likelihood that the accident can cause a missile to rupture containment and cause core damage and/or disable containment cooling. The relative effectiveness of potential alternative corrective measures in reducing the expected (mean) value of vessel failure, core melt, person-rems, and fatalities. I 11.2 Basis for Continued Operation Finally, as part of the plant-specific analysis package, the licensee I should provide a basis for concluding whether or not continued plant operation is justified. The basis for continued operation should include comparison with NRC's PTS acceptance criteria given in the Introduction to this guide. D 45

REFERENCES 1. T. J. Burns et al., " Preliminary Development of an Integrated Approach to the Evaluation of Pressurized Thermal Shock Risk As Applied to the Oconee Unit 1 Nuclear Power Plant," Oak Ridge National Laboratory, U.S. Nuclear Regulatory Commission (USNRC) Report NUREG/CR-3770 (0RNL/TM-9176), to be published. 2. D. L. Selby et al., " Pressurized Thermal Shock Evaluation of the Calvert Cliffs Unit 1 Nuclear Power Plant," Oak Ridge National Laboratory, USNRC Report NUREG/CR-4022 (ORNL/TM-9408), November 1985. 3. D. L. Selby et al., " Pressurized Thermal Shock Evaluation of the H. B. Robinson Unit 2 Nuclear Power Plant," Oak Ridge National Laboratory, USNRC Report NUREG/CR-4183 (0RNL/TM-9567), November 1985. 4. USNRC, " Pressurized Thermal Shock (PTS)," SECY-82-465, November 23, 1982. 5. L. Potash, " Confusion Matrix," Section C.l.2 of Appendix C in "0conee PRA," Electric Power Research Institute, Palo Alto, CA, and Duke Power Co., Charlotte, NC, NSAC/60, Vol. 4, 1984. 6. R. A. Bari et al., " Probability Safety Analysis Procedures Guide," Brookhaven National Laboratory, Revision 1 to USNRC Report NUREG/CR-2815, August 1985. 7. A. D. Swain and H. E. Guttmann, " Handbook of Human Reliability Analysis with Emphasis on Nuclear Power Plant Applications," Argonne National Lateoratory, USNRC Report NUREG/CR-1278, October 1983. 8. B. Bassett et al., " TRAC Analysis of Severe Overcooling Transients for the Ocoree 1 PWR," Los Alamos Scientific Laboratory (LASL), USNRC Report NUREG/CR-3706, August 1985. 9. C. D. Fletcher et al., "RELAP 5 Thermal-Hydraulic Analysis of PTS Sequences l for the Oconee 1 PWR," EG&G, USNRC Report NUREG/CR-3761, July 1984. 46

i 10. J. Koenig, G. Spriggs, and R. Smith, " TRAC-PF1 Analysis of Potential PTS Transients at a Combustion Engineering PWR," LASL, USNRC Report NUREG/CR-4109, April 1985. 11. C. D. Fletcher et al., "RELAP 5 Thermal-Hydraulic Analyses of PTS Sequences for the H. B. Robinson Unit 2 PWR," EG&G, USNRC Report NUREG/CR-3977, April 1985. 12. C. D. Fletcher, C. B. Davis, and D. M. Ogden, " Thermal-Hydraulic Analyses of Overcooling Sequences for the H. B. Robinson Unit 2 PTS Study," EG&G, USNRC Report NUREG/CR-3935, July 1985. 13. T. G. Theofanous et al., " Decay of Buoyancy Driven Stratified Layers with Application to PTS," Purdue University, USNRC Report NUREG/CR-3700, May 1984. 14. T. G. Theofanous et ., " REMIX: Computer Program for Temperature Transients Due to High Pressure Injection in a Stagnant Loop," Purdue University, USNRC Report NUREG/CR-3701, January 1986. 15. T. G. Theofanous et al., " Buoyancy Effects on Overcooling Transients Calculated for the USNRC Pressurized Thermal Shock Study," Purdue University, USNRC Report NUREG/CR-3702, January 1986. 16. Bart Daly, "Three Dimensional Calculations of Transient Fluid Thermal Mixing in the Downcomer of the Calvert Cliffs-1 Plant Using SOLA-PTS," LASL, USNRC Report NUREG/CR-3704, April 1984 17. Martin Torrey and Bart Daly, "SOLA-PTS: A Transient 3-D Algorithm for Fluid Thermal Mixing and Wail Heat Transfer in Complex Geometries," LASL, USNRC Report NUREG/CR-3822, July 1934. 18. F. X. Dolan et al., " Facility and Test Design Report: 1/2 Scale Thermal Mixing Project," USNRC Report NUREG/CR-3426, Vols. I and II, September 1985. 47

19. R. D. Cheverton and D. G. Ball, "0CA-P, A Deterministic and Probabilistic Fracture-Mechanics Code for Application to Pressure Vessels," Oak Ridge National Laboratory, USNRC Report NUREG/CR-3618 (0RNL-5991), July 1984. 20. D. L. Stevens et al., " VISA - A Computer Code for Predicting the Proba-bility of Reactor Vessel Failure," Battelle Pacific Northwest Laboratories, USNRC Report NUREG/CR-3384, September 1983. 21. USNRC Regulatory Guide 1.99, " Effects of Residual Elements on Predicted Radiation Damage to Reactor Vessel Materials." 22. F. A. Simonen, " Vessel Behavior Following a Through Wall Crack," PNL SA12547, presented at the 12th Water Reactor Safety Research Information Meeting, Gaithersburg, MD, October 1984. G 48

1 DRAFT REGULATORY ANALYSIS D The pressurized thermal shock (PTS) rule, S 50.61 of 10 CFR Part 50 (July 23, 1985--50 FR 29937), requires collection and reporting of material properties d;ta, analyses of flux reduction options, and detailed plant-specific PTS risk cnalyses for those plants that reach the screening criterion based on RT as

NDT, specified in the rule, during the term of the operating license.

The proposed r;gulatory guide addresses the detailed plant-specific risk analysis requirement, providing recommendations regarding how licensees should perform and how the NRC staff should review those analyses. l Neither the PTS rule nor the proposed regulatory guide requires specific j corrective actions. The proposed guide merely provides guidance for the per-l formance of the analyses required by the rule to identify and select necessary l l corrective actions. Therefore, in accordance with the Commission's Regulatory Analysis Guidelines (NUREG/BR-0058, Revision 1), this regulatory analysis does not provide extensive and detailed assessment of required, specific corrective actions. I The background material, nature of the problem, objectives, and costs, etc., of the PTS rule's requirements are covered in the regulatory analysis prepared as part of the rulemaking proceeding (Enclosure B to SECY-83-288, Proposed Pressurized Thermal Shock (PTS) Rule, July 15, 1983, and Enclosure D to SECY-85-60, Final Pressurized Thermal Shock (PTS) Rule, February 20, 1985). This regulatory analysis therefore addresses only (1) the need for publishing guidance regarding how licensees should perform the required plant-specific analyses, (2) the appropriateness of this particular guidance, and (3) the basis for the NRC staff acceptance criteria provided in the subject guide. 1. Need for Guidance The NRC staff has obtained considerable experience concerning PTS risk cnalyses. This experience has ccme from performance of analyses by the staff, " Reference Temperature for the Nil Ductility Transition, a measure of the temperature range in which the materials' ductility changes most rapidly with changes in temperature. 49

from prototype plant-specific analyses performed by national laboratories and sponsored by NRC, and from review of industry-sponsored analyses. The proposed regulatory guide reflects the lessons learned from this experience. Avail-ability of the guide will aid licensees in performing analyses that will effi-ciently derive risk estimates in the form the NRC needs for use in evaluating their conformance with the regulations. This need for guidance is particularly acute since the plant-specific PTS analyses should use a probabilistic risk analysis (PRA) approach, as opposed to the more traditional design basis accident (DBA) approach, as explained below. The PTS risk is developed as the sum of the small risks resulting from each of a large number of possible (but unlikely) PTS events. The proposed regulatory guide accordingly describes acceptable methods to identify as many as possible of the potential PTS events, group them, calculate the frequencies and consequences of each group, determine the risk due to each group by multi-plying the predicted frequency by the calculated consequences, and then sum the results from all groups to obtain total PTS risk estimates that can be compared with the acceptance criteria given in the proposed regulatory guide. l The DBA approach, on the other-hand, would attempt to define a worst cred-ible event (the " design basis accident") and then show that (1) consequences from that event are acceptable and (2) all other credible events are less severe and therefore acceptable. The staff has determined that this DBA approach is not appropriate for plant-specific PTS analyses because the total risk from all credible PTS events can be significant even though each event individually is less severe than the DBA. The NRC staff therefore believes that it is necessary to publish this guide so that licensees will use the acceptable PRA approach and not waste time and resources on the more traditional DBA t.pproach. 2. Justification of This Particular Guidance The NRC staff has performed prototype plant-specific analyses for three plants. They constitute the most detailed, thorough analyses performed to date, and the lessons learned in their performance are reflected in the guide. The NRC staff has incorporated into the guide descriptions of the best methods found { l 50

i D d; tails should be included), how to use event tree methodologies to identify rcgarding how to assemble details of a plant's design (and to what level those ] and group potential PTS events, how to calculate severity of the events, how to integrate the resulting risk, and many other subjects. The staff believes that the benefit of this experience is presented in this guide, and its use by li-c;nsees will enable them to avoid many of the false starts and errors made by the staff and their contractors in performing the prototype analyses, thereby saving time and resources. 3. Justification of Acceptance Criteria The guide states that in judging the acceptability of continued operation beyond the PTS screening criterion, the staff will accept, without further analyses of risk, any analyses performed with acceptable methods such as those discribed in the subject regulatory guide that predict a through-wall crack penetration frequency less than 5 x 10 6 per reactor year. For analyses that predict a through-wall crack penetration frequency above that value, additional analyses predicting core melt frequency, person-rems, and fatalities are also required. These additional analyses will be used in determining acceptability of continued operation and, in addition, will be compared to any quantitative criteria in effect at the time the analyses are performed. The mean frequency of reactor vessel through-wall crack penetration is included as the principal acceptance criterion because the staff's analyses predict that there is a high likelihood of core damage in the event of such cracks. Core damage events have potential public health and safety consequences that are difficult to analyze with certainty. They also would have severe eco-nomic impacts upon the licensee and the public who will pay for cleanup and rcplacement power. For all these reasons, reactor vessel through-wall crack p;netration frequency is included as the principal acceptance criterion. The particular value of 5 x 10 6 mean frequency per reactor year was selected as an achievable, realistic goal that will result in an acceptable level of risk. It is believed that this value is acceptably low considering that pressure vessel fcilure is not part of the design basis of the plant and therefore must have a frequency low enough to be considered incredible. When the various (unquanti-I fiable) biases that are inherent in the analyses are taken into account at least 51

qualitatively, such as the implicit assumption that " core damage" is equivalent I to " core melt," this value probably results in a core melt mean frequency close to one per million reactor years. Additionally, the staff believes that in order to make a thorough assess-ment of risk due to PTS events, for cases with a marginal predicted through-wall crack penetration frequency, the analyses should include prediction of risk in terms of predicted core melt frequency, person-rem exposure, and early and late fatalities. This is because, given a reactor vessel through-wall crack, the vessel failure modes will vary from plant to plant, and, given vessel failure and core melt, containment performance and resulting risk to the surrounding site-specific population density will be different. Therefore, a risk assess-ment in those standard terms is required for cases in which the through-wall crack penetration frequency is not by itself low enough to obviate the need for these additional analyses (i.e., is greater than 5 x 10 8 per reactor year). It is possible at this time only to state that the results of these additional analyses will be used to evaluate the acceptability of continued operation and, in addition, will be compared to whatever goals are in place or in general use I by the NRC (if any) at the time the plant-specific PTS analyses are required. In the opinion of the NRC staff, there are no practical quantities on which to base the acceptance criteria other than reactor vessel through-wall cracks (i.e., vessel failure) ard/or the standard risk measures of core melt frequency, person-rem exposure, and fatalities. l 52

UNITED STATES rinst conss unit NUCLEAR REGULATORY COMMISSION

• 5 ' 8,',* "c' 5 '^'"

WASHINGTON, D.C. 20555 was,w o c,, OFFICIAL BUSINESS PENALTY FOR PRIVATE USE, $300 bf05 5064235 N C ADM. 3 3 DOCU IY DF 11SAlop 04p ENT DC 15 CON TR WASHINGTot; L DESK'RECSB C 2g555 l l l e -__1}}