ML20128M988

Research News.Volume 6,Number 1

ML20128M988
Person / Time
Issue date:	01/31/1993
From:	NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:
References
NUREG-BR-0112, NUREG-BR-0112-V06-N1, NUREG-BR-112, NUREG-BR-112-V6-N1, NUDOCS 9302220424
Download: ML20128M988 (14)

Text

-

NUREGIBR-0112 s k* "8 %

$..(... ) RESEARCH NEWS OFFICE OF NUCLEAR REGULATORY RESEARCH U.S. NUCLEAR REGULATORY COMMISSION VOLUME 6, NUMBER 1 JANUARY 1993 TMl-2 Vessel Investigation Project Prior to the initiation of the VIP, the Department of Energy (DOE) had supported extensivo postacci-dont examinations and analyses of the TMI ,2 Alan M. Rubin' DSR/AED' and damaged core. The primary objectivo of this Edwin M. Hackett, DE/MED TMI-2 Accident Evaluation Program was to de-volop an understanding of (1) core damage pro-grossion in the upper coro region, (2) hoatup of Background the consolidated region leading to extensive molt-ing of the core (3) relocation of approximately 20 As the last of the fuelwas being removed from tho tons of debris to the lower head, and (4) release of Three Mile Island Unit 2 (TMI-2) reactor pressure fiss!on products to the reactor vessel and the con-vessel, it could be seen that quantities of melted tainment, fuel had covered the lower head of the pressur vossol, and there was cencern that the lower head DOE decided to end the DOE-sponsored re- (

of the reactor pressuro vessel might have under- search on the evolution of the TMI- 2 accident fol-gono chemical and thermal attack. The possibility ow ng removal of the damaged core. In October of such an attack raised important safety issues 1987, a meeting was held at MIT to review the re-relating to reactor vessel integrity following a so- sults and conclusions of the work completed up to vore accident. that time. The review made it clear that core degra-dation and molting had posed a threat to vessel integrity. Tho review group, which included Pro-In October 1988, the NRC, in cooperation with 10 fossor Neil Todreas of MIT and Edwin Kintner of foreign countries under the auspices of the Or- GPU Nuclear, rocommended that the issues relat-ganization for Economic Cooperation and Devel-ing to vessel integrity should bo investigated fur-opment's (OECD) Nuclear Energy Agency, began ther. RES agreed and decided to perform addi-a joint research program to examine and analyze tional roscarch, subject to obtaining the coopera-material samples from the lower head of the tion of GPU Nuclear, for the purpose of determin-TMI-2 reactor pressure vessel. The objectives of ing what the margin to reactor vessel failure had this program, called the TMI-2 Vessel Investiga' been during the course of the accident.

tion Project (VIP), were to (1) investigate the con.

dition and properties of materials extracted from The principal conclusions from the DOE-the lower head of the TMI-2 reactor pressure vos- sponsored research were that the TMI-2 core sel, (2) determine the extent of damage to the damage progression involved the formation of a lower head by chemical and thermal attack, and large consolidated mass of core material sur-(3) dotermine the margin of structuralintegrity that rounded by supporting crusts, the failure of the remained in the pressure vessel. supporting crusts, and finally, Lie long term 1

9302220424 930131 PDR NUREO BR-0112 R PDR

cooling of a largo volumo ot molton coro material. pumped through holes in the cutting electrodo The TMI-2 accident demonstrated that a sovero cools the molten material and flushes the resultin0 accident can be terminated and confined to tho re- particles away from the cutting area.

actor pressuro vessel by cooling water before the failure of the lower head. However, there was n quantitativo information that could be used to de- be extracted, different regions of the lower head termino how close the vessel was to failuto. had to be sampled. The following arcas were so-lected for extracting samples.

The VIP examinations described in this article go e As closo as possible to the area directly beyond the work performed under the previous beneath the primary relocation path of TMI-2 cxaminations. Specifically, the VIP plan molton coro material to the lower head, was to obtain and examino camples of the lower head steel, instrument penetrations, and e Toward the radial center of the lower previously molton debris that was attached to the head underneath the maximum debris lower head and use this information to estimato thickness, the vossol margin to failuto. These tasks woro not included in the scopo of the DOE Accident Evalu- e In the quadrant of tho lower head where a ation Pro 0 ram. " wall" of consolidated debris similar to a lava front had been observed, VIP Scope of Work e in an area of the lower head not con-The scope of work of tho VIP consists of throo ma- ag Wo mokn com maMal b ad jor components-samplo extraction, samplo s a control samplo, and examination, and analysis of results. The sample . Arcas that include one or more instru-extraction phaso involved developing and testing ment penetrations, especially in the ar-g an extraction tool and procedures, extracting the cas noted above.

samples from the lower head, and transferring the samples from the TMI-2 sito for examination. The Since this was a firct-of a kind process using a criteria and constraints that woro considered in specially designed MDM cuttin0 head, tho exact developing the extraction tool included: number of samples could not be predicted in ad-vanco, it was hoped that from 8 to 20 samples e Obtaining tholargest number of samplos could be obtained. As it turned out, the samplo ex-possible during the 30 day window that traction was very successful, and 15 reactor vos-was available for the extraction process, sol steel specimens,14 incoro nozzles, and 2 in-coro guide tubes woro extracted from the lower e Not broachin0 or significantly weakening head over a 30-day period ending March 1,1990.

the reactor vessel, and Following extraction, the vessel stool samples

• Working on a shioided platform mounted were decontaminated, sectioned, and distributed 40 feet abovo the lower head and re- to Argonne National Laboratory (ANL), Idaho Na-motely extracting the samplos that wore tional Engineering Laboratory (INEL), and sovon covered by highly borated and buffered participating countries for mechanical and metal-wate' lographic examinations. Also, the nozzles and guide tubes woro cut and distributed for The cutting technique that was selected for ex' examination to ANL, INEL, and the CEA in Saclay, tracting samplos of the lower head steel was a Franco, metal disintegration machinin0 (MDM) proces .

This process usos a series of electric arcs to melt a Since the extraction of the test specimens in 1990, small amount of material in the cutting area. Water metallographic examinations of the vessel stoo!

2

samples, including microstructural examinations United Kingdom) and the Electric Power Research and hardness measurements, have been com- Institute.

pleted to determine the temperature distribution in the lower head that existed during the accident.

Testing of the mechanical properties of thr sam, Future Plans pies was also performed to provide data on ulti-mate strength and change in resistance to failure Results of the ongoing TMI-2 lower head exami-by creep rupture for calculating the margin to nations are expected to provide additional infor-failure. mation on the physical properties of the speci-The VIP team removed more samples from the mens, temperature distributions in the instrument nozzles, and interactions between the molton g

damaged reactor vessel than they originally ex- core material and the vessel. These results will be pected, and the preliminary examinations indi. used to perform scoping analyses of potential re-cated that comprehensive studies of these sam. actor vessel failure modes, such as global or local plcs would provide important data on the behavior failure of the reactor vessellower hNd and pene-of reactor vessels during severe accidents. As a tration tube failures (l.o., tube heat-up and failure result, in September 1991 the VIP Management resulting from the flow of molten core materialin Board decided to amend the original agreement the instrument tube, as well as tube ejectica fol-to extend the VIP program from September 30, lowing heat up and failure of the weld on the inner 1991, to March 31, 1993, and to increase the surface of the reactor pressure vessel), More de-budget by approximately $1.5 million. The objec- tailed analyses of the most likely failure mecha-tives of the amended program are to (1) perform nisms will be performed to estimate the margin to-more detailed testing and examination of the in. failure of the lower head A final project report that core instrument tubo nozzle penetrations and the integrates the results of all the sample examina-in-core instrument guide tubes that were ex. tions and analyses will be issued at the comple-tracted from tho lower head, (2) perform additional tion of this program in June 1993. The results of analyses of potential reactor vessel failuro modes this project, along with the results of previous based on data from sample examinations, and (3) TMI-2 ctudies, will contribute to an increased un-assess the margin to-failure of the lower head of derstanding of core melt scquences and the im-the reactor vessel. pact of such sequences on reactor vessel behavior.

Metallographic and scanning electron micro-scope (SEM) examinations of the instrument tube The Smithsonian institution is planning to exhibit nozzles have been performed to estimate the noz- tools and equipment that were used to extract zles' temperature and interaction of the inconel samples from the TMI-2 reactor vessel under the 600 nozzle material with molten core materlats. VIP. The TMl-2 extraction tool exhibit will be in-Results of these examinations have been used in cluded in a section on nuclear science that is conjunction with estimated temperature distribu- planned as part of a major permanent exhibit on tion-s to assess potential failure modes of the in- Scienco ln American Life at the Smithsonian's Na-l strument tube penetrations. tional Museum of American History. Equipment for this exhibit may include the spare MDM cutting The Accident Evaluation Branch and Materials En- tool and articulating arm, practice cutting sam-gineering Branch of RES are jointly funding ap- ples, a cut and polished metallographic specimen proximate!y 50 percent of the cost of this $9 million from one of the samples, protective suits used program. The remaining funds come from the during the sample extraction (all uncon-OECD member countries participating in this pro- taminated), and a videotape taken during the ac-joct (Belgium, Germany, Finland. France, Italy, tual extraction process. Current plans are for the Japan, Spain, Sweden, Switzerland, and the exhibit to open around April 1994.

3

Instrumentation and Controls also ahead in advanced development except for Tcchnology Study architecturo and instrumentation. In basic ro-scarch, European plants are ahead in four of the A study of the instrumentation and controls (l&C) seven cato00 ries; however, the United States is technology used in nuclear power plants in about equal to Europo in instrumentation and in Europe was conducted recently by a panel of U.S. analog-to digital transition, and the United States specialists. This study included a review of the lit- is ahead in architecturo in general. In other words, erature on the subject, followed by visits to some U.S. computers are being purchased and used in vendors, utilities, nuclear power plants, and re- all countrics that the panel visited, but the devol-search organizations in Europe that are leaders in opment and imptomontation of the computers for the field of nuclear l&C. nuclear power plant l&C is more advanced in Eumpo and Canada.

Leo Beltracchi of the Office of Research was on the panel. Their study was sponsored by the It appeared that the United States is behind in the National Science Foundation, and their findings development of digital systems for nuclear plants have been published by Loyola College in Mary- as well as in experience in using them. Franco has land as " European Nuclear instrumentation and the most experienco with digital safety systems Controls," which is available from the National and has built successful automatic control and in-Technical Information Service.

formational systems, with the original design pur-Nuclear power plants in Europe, Canada, Japan, chased from U.S. reactor vendors.

and the United States are moving toward the uso of digital computers, especially microprocessors, Germany has developed a unique control and for information and control systems. The amount safety strategy that automatically moves reactor of automation and the role of tho operator are un- systems back into the safe operating region. This der discussion in all countries. In Japan and Ger- automatic action minimizes the number of many, plants are moving toward a high degree of scrams, smoothos transients to minimizo compo-automation, whereas in Franco the emphasis is nont stresses, and provides time for operator di-on computer-generated procedures with the deci- agnosos for a broad spectrum of controlfailures, sien to 2nablo being made by skilled operators. from both equipment failuro and human errors.

Sorm Russian plants uso digital systems to help The system presently involves full digital reactor the operator identify problems, decido on the ap- control, but for prevention and mitigation it uses propriato correctivo actions, and aid in the execu- semiconductor-based analog equipment, which tion of these actions. is to be replaced by digital equipment without much increase in function.

The panel mado a qualitativo comparison of the standing of U.S. nuclear power plants relativo to Europe is ahead in the use of fault diagnosis and the countries visited for status and progress in ba. signal validation systems. Work on the use of digi-sic research, advanced development, and prod. tal information programs for fault management uct implementation in seven categories: control systems in nuclear plants is moving more rapidly room design, analog-to digital transition, fault in both Europe and Japan than in the United management systems, control strategies, I&C ar. States.

chitecture, instrumentation, and standards and tools. The hardware for the digital systems used in all countries is by and largo from U.S. computer The panel concluded that European plants are companics butlimited deployment of digital sys-ahead of the United States and moving further tems in U.S. nuclear plants has curtailed the ac-ahead in all seven categories, with the possible crual of experience in the computer system archi-exception of instrumentation. European plants are tecture for l&C systems.

4 I

In Franco and Germany, control strategios have (JCCCNRS) meeting and the workshop on nu-been extended to allow nuclear power plants to be clear power plant aging and life extension.

used f or automatic changes in power to match de-mands from the utility grid. These capabihtios Maintaining tho safety of the older operating nu-have improved control of local power distnbution clear power plants in the Commonwealth of Indo-changes during transient power conditions. pendent States (CIS) was a major concem of the CIS participants. The Russian Federation (RF)

In all countries visited by the panel, instrumenta- places sigMcant importance on the activitics of tion for nuclear power plants is similar to that in tho this group, as evidenced by the activo participa-United States. Somo special instrumentation is being developed; for exarnple, a special neutron tion of the representatives of various institutions and organizations of the CIS as well as the operat-ing plant personnel.

{

detection system is under development in France to provide improved in coro and ex-core power Information exchange through this group over the density and transient power level information.

past 18 months has provided a foundation for Germany has pionocred the use of prompt, in' developing both near-term and long term pro-core cobalt detectors for gathering detailed power density information. grams 1or managing aging in their operating nu-clear power plants with the primary emphasis on safety. It was recommended that the group con.

The European countries are ahead in the uso of centrato on specific technical issues such as computer-assisted software engineering tools, metal fatigue under operating environmental con-and they are more advanced in the development ditions, in situ degradation monitoring of naturally of standards. Standards and guidelines are the aged cablos, and nondestructivo examination.

basis of the design and development of com-puter based safety systems. The U.S. nuclear in-dustry does not now have equivalent standards How Clean is Clean?

and guidelines for the development of computer- Chris Daily, DRA/RPHEB based safety systems; however, an effort to de-velop equivalent standards is under way. NRC licenscos who need to decontaminato land and structures as part of the decommissioning An important point brought out in the survey is that and license termination process must have criteria the United States has been able to learn from the to determino "how clean is clean enough." The mistakes and overcomplexitics of other countries. NRC must be assured that public health and in Franco and Canada, the programmability of the safety and the environment are protected and en-digital systems enticed the users to add complexl. sure that the total dose to an individual ls less than ties that evolved into problems. Efforts must be the public dose limit of 100 mrom/ year. In addi-made to maintain simplicity in systems, and prob. tion, the NRC has set a goal for public dosos at-lems are often not recognized until the review, tributablo to residual contaniination after decom-quality assuranco (verification and validation), and missioning at a fraction of the public dose limit.

final approval stages.

The dose that an individual might recolve from re-sidual radioactivity remaining at a site attor do-Workshop on Nuclear Power Plant commissioning is estimated by first developing Aging and Life Extension in Moscow sgenarios te describe potential future uses of the site. The scenarios allow possible exposures to bo During October, NRC staff and representatives of modeled and associated doses to be estimated.

the Department of Energy and nationallaborato- The modeling and scenarios can becomo ex-rios participated in the US-CIS Joint Coordinating tremely complicated, depending on the complex.

Committee on Civilian Nuc! car Reactor Safety ity of the site and the level of uncertainty that is 5

considered acceptable in the doso estimate. Do- creased benefits to the public health, public tailed modeling may often bo beyond the techni- safety, or to the environment. Hence, the doso cal and financial capabilities of a large number of conversion factors in NUREG/CR-5512 are licensocs - especially small licensees with limited judged to bo higher than (l.o., overestimate) the resources. most probablo annual dose but rnay be lower than (i.e., underestimato) the bounding annual doso.

NUREG/CR-5512, " Residual Radioactivo Con-tamination From Decommissioning" (October Licensees have some flexibility when applying the 1992), has been developed to provide generic and modeling contained in NUREGICR-5512. If sito-specific doso conversion factors for residual sli0htly increased accuracy or realism of the radioactivity based on generic modeling. Thoso scrooning dose conversion factors is desired, and doso conversion factors ato used in a screening thero is adequate justification, thu generic (do-analysis to determino whether a sito moots the er . f ault) paramotor values may bo replaced with sito-teria for releasc or whether rnoro detailed analysos specific paramotors. Within the modeling frame-work of NUREG/CR-5512, such a substitution of must be performed. The first volumo of the report presents the scenarios, models, mathematical parameters would lead to site-specific derived formulations, assumptions, and justification of pa. doso conversion factors. The site specific doso rameter choices. It also includes responses to conversion factors may then replace the generic comments from the January 1990 draft of the re- doso conversion factors in the screening analysis.

port. The second volumo will be a uscr's manual As another example of the flexibility of tho ap-for an associated microcomputer-based pro- proach, the NRC staff is developing a supplemen-gram, and it will include tables of the generic doso tal technical rationale and method for Incorporat-conversion factors, examplo calculations, and the Ing independent ground-water models into the computer codo listing. A third volumo will contain NUREG/CR-5512 methodology so that a hierar-a sensitivity analysis of parameters used in the chy of scroonin0 can be implemented. Existing modeling and a comparison with previously used groundwater flow and radionuclido transport guidanco such as Regulatory Guide 1.86, "Ter- models used in the performance assessment of mination of Operating Licensos for Nuclear low and high-level wasto facilitics will be exam, Reactors " ined as possible tools in this methodology, and their appropriateness for scrooning sites with re-The scenarios used in this gencric analysis are sidual radioactivo contamination that exceeds the prudently conservative but not necessarily bound- " water uso models" presently stated in NUREG/

in0 or " worst case." Selection of a prudently con- CR-5512 will be ovaluated. The development of a servativo scenario requires a great deal of profes- hierarchical strategy to choose the appropriato sional judgment and common senso. The intent is Groundwater flow and transport model, given the to account for the vast majority of potential uses of site radiological survey inventory and conditions, lands and structures and overestimato the most will also be performod.

probablo annual dose whilo discounting a small fraction of highly unlikely uses that would result in NUREG/CR-5512 is one part of a largo program higher doses. The prudently conservativo ap- the NRC staff has under way to provido informa-proach does not includo low probability scenarios tion and guidance for the implementation of re-that may result in higher calculated dosos but are lease criteria for the accommissioning of lands based on aberrant behavior or unpredictabit and and structures. For examplo, NUREG/CR-5849, highly unlikely circumstances. The alternativo ns

• Manual for Conducting Raolological Surveys in to uso scenarios that would yield an upper limit on Support of License Termination" (Draft, Juno doses, i c., bounding or " worst case," but would 1992), provides information on acceptable meas-unnecessarily limit the usefulness of the resulting urement and survey techniques and procedures.

roteaso critoria without providing significantly in- The draft manual is designed to provido the 6

I

licensco with guidance on planning, conducting, products, including contractor reports. His recent and documenting sito survoys that con be used to assignments include the following:

demonstrate that a site has boon dccontaminated to a levol consistent with the Corr, mission's critc- e In 1986, WA's application for an operating li-ria. It contains specific guidanco t n the rolo of sur- conso for Watts Bar was suspended when, voys in the decommissioning piocess, survey amo..g other problems, numerous defects in planning and design, selection ano use of radio- tho OA records woro found. In 1988. TVA in-logical instrumentation, conducting 'ho survey, stituted a Cortoctivo Action Program (CAP) to interpreting survey results, and docu anting and correct the problems with the OA records.

reporting survey results. It also includes a samplo The goal of the CAP is to demonstrate that Survey Plan and a sample Final Status Survey each of the 186 ANSI record types has no Report prepared in accordance with the proced- more than 5 porcent defects (e.g., missing uros contained in the manual. records) with 95 porcent confidence. To achlove this goal, WA proposed a sampling plan based on a Bayoslan approach.

ASME Award to Bob Bosnak There woro two major flaws in the sampling Robert J. Bosnak of the Division of Engincoring plan. The first concerned the Bayoslan has been elected an Honorary Member of the approach used and the second dealt with the ASME Boiler and Pressuro Vessel Committoo rectification proceduro if defects woro found.

(B&PVC). Bob received this award in Anaheim, While a Bayoslan approach based on California, at the November 1992 mooting of the subjectivo export judgment is extensively ASME Council on Codos and Standards. Bob used in nuclear probabilistic risk assessment served on the Committee for more than 20 years, (PRA), tho situation at Watts Bar is different.

first while working for the Coast Guard and then for First, contrary to the usual PRA situation in the AEC/NRC. Ho recently resigned from the which export judgment must be used Committoo as he plans to retiro from the NRC in because adequato data is unavailable, February 1993. Bob's efforts greatly contributed unlimited data ls available at Watts Bar simply to the uso of ASME codes and standards by both by sampling the OA records. Second, the Federal agencios as a means of satisfying regula- goal in a PRA is usually to estimato a quantity tory requirements. Bob's know how and appro- (e.g., coro damago frequency); at Watts Bar clation for tho credibility of tho voluntary standards the goal is to test a hypothesis. This programs have resulted in genuino respect from difference is crucial. The goal of a PRA is to all who have served with him on the Committoo. gain insight into the risk; the goal of the Watts Bar CAP to is make a decision. Because the Watts Bar sampling plan can be very sensitive Expert Statistical Review to the subjective choice of a prior distribution a on ch (a nocessary stadng The NRC deals with statistical issues in many point for a Bayoslan analysis), the NRC would forms all the timo. In a recent lotter to the Chair-man of the NRC, the Advisory Committoo on Re-nt be able to validate the Bayesian appt ach. Instead, a classical statistical actor Safoguards emphasized the importanco of export statisticai review of NRC products involving appr ach to moet tho 95/5 acceptance critorion was recommended.

a statistical analysis. RES has an export statisti-clan on staff, Leo Abramson of the Division of A second flaw in 100 sampling plan Safety issue Resolution, who also serves as an concerned rectification. In order to minimize in-house statistical consultant to other offices the required samplo size, TVA opted to use a within the NRC His role is to detect and correct sampling plan with an acceptanco number of any statistical deficiencies or problems in NRC zero. In this plan, a random sample of 60 is 7

chosen and the population is accepted only if Performance Assessment " This document no defects are found in order to satisfy the presents a quantitativo f ramewor k f or defining 95/5 acceptanco critorion, the population SCC in terms of an upper bcund on tho must be rejected it any defects are found in number of failed wasto packages and a (

the samplo. However, to avoid rejecting the confidenco level for attaining the bound. In population outright if a defect is found, TVA order to deal with the lar00 scientific proposed to "roctify' the population by uncertainties involved, NUREG/CR-5639 removing all records with the same type of discuscos a number of methodolo0ies for defect from the population as well as from the uncertainty analysis. For this study, a now sample. An additional random camplo is method was developed for combinin0 drawn to replaco the defects found in the bounds on ono subset of input variables with samplo of 60 and the population is accepted it distributions on another subset of input ,

no further defects are found. While this variables to produce bounds on tho  ;

rectification proceduro is intuitively plausible, distribution of the output variables.

it turns out that it will satisfy tho 95/5 I acceptanco critorion only if the defoct types to in a report entitled *'Substantially Completo be rectified are completely specified beforo Containment' Feasibility Assessment and the initial samplo is drawn. If this condition Alternativos Report," the CNWRA identified cannot bo satisfied (which it rarely can bo in four altomativo regulatory positions cased on practico), the samplin0 plan must be revised the quantitativo framework developed in by increasing the samplo sizo. NUREG/CR-5639. One alternativo is Lased on the current qualitativo ruto, ono is based in responso to the NRC position paper on a modified qualitativo rule, and two are crit'cizing their CAP, TVA scrapped their based on quantitativo versions of the rulo. All Dayosian and rectification approach and alternativos would requira detailed technical adopted an approach based on classical Guidance to DOE. In order to provido NMSS statistics. NRC has accepted the revised with insight as to which rogulatory position to CAP. recommend, a prioritization exercise based on the methodology developed to clicit o Under current rogulations, tho containment of export judgment for NUREG-1150 was high-lovol tadioactive wastos in the proposed carried out. A set of objectivos and attributos Yucca Mountain repository is iequired to be to rato the alternativos was developed, and a "substantially completo" for a period of panel of nino NRC staff members was clicited 300-1000 years, with tho preciso period to bo to rato the alternativos. To the surpriso of chosen by the Commission at a later dato, some panel members, sovon of the nino NMSS is considering how to interpret this panelists preferred the quantitativo qualitativo rule in a quantitative f ashion. They altomativos. A description of th6 prioritization have tasked the Conter for Nuclear Waste study and its results was published as .

Ro0ulatory Analysos (CNWRA) to delinoate CNWRA 92-016. "'Substantially Complete the technical considerations involved in sub- Containment' (SCC) Elicitation Report," in stantially completo containment (SCC), to August 1992. I identify and develop uncertainty evaluation methods for SCC analysis, and to identify al-

• The Radiation Protection and Health Effects ternativo regulatory positions. Branch of RES is developing a

• Manual for Conducting Radiological Surveys in Support 3 Throo reports woro producad for this task, of License Termination," NUREG/CR-5849, including NUREG/CR-5639, " Uncertainty prepared by Oak Ridge Associated Universi-Evaluation Methods for Waste Package ties. Tho manual usos a statistical hypothesis I

8

- - _ - _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ - - _ - _ _ ~

testing procedure to determine whether the -

The proposed methodology for calculat-guidelino value is exceeded and additional ing P,e was based on an inappropriato cleanup is required. For a set of measuro- use of regression plots and their associ-monts in a survey area, the samplo mean plus ated confidence bands Therefore,itwas an appropriato multiplo of the standard devia- recommended that DOE consider devel-tion is compared with the guideline value. If oping a model for the spatial distribution any ci the measurements is a localized " hot of volcanic centers in the vicinity of Yucca spot" (i.e., an area with elevated radioactiv- Mountain and a model for the probability ity), the proceduro may bo overly conserva- of repository disruption as a function of tivo and Icad to requiring a cleanup for a the location of a volcanic conter, and survey area whose mean activity level is be- then combine these two models to calcu-k low the guideline value. A less biased esti- lato Pro.

mate of the mean activity lovel can be ob-v tained by using a weighted averago (instead -

In order to handle the uncertainty as to of the samplo mean)in which the weights are which of several competing models ap-proportional to the areas of elevated activity. plies, DOE proposed using a weighted average, with the wclghts to be chosen by expert opinion. This approach is in-e As a part of the structural aging research pro- tended to reduce bias and to provide a gram, the Structural Seismic Engineering summary assessment useful for regula-Branch has produced a report on "Nondo- tory decisionmaking, However, bias is structivo Evaluation of the in-Place Compres- not necessarily reduced by weighting al-sive Strength of Concreto Based Upon Lim- ternativo models. If the correct result lies ited Destructive Testing," NISTIR 4874, pro- somewhere in the middle of the model pared by the National Instituto of Standards results, weighting models would tend to and Technology (NIST). The report takes reduce bias. Nevertheless, there is no available nondestructivo concreto testing assurance that the correct result is in the data as reported in the open literaturo and at- middle-it might be at one of the ex-tempts to establish a correlatbn between tremos or even beyond. If this were the nondestructivo and destructive testing (which caso, weighting could give a false assur-is more accurate) for residual concrete ance that bias had been reduced. In strength. Sinco the amount and quality of the order that the results reflect the full rango availablo data wet e limited, a complex statisti- of scientific l'ncertainty as expressed cal procedure was used to establish the re- by the exports, it was recommended quired correlation. The correlation was vali- that the methodology developed for dated by splitting the data sets in half and NUREG-1150 be used as a basis for checking each data act half against the other eliciting and aggregating expert opinion.

half. It is expected that the recently released NIST report will generato considerable inter- In addition, using weighted models may est in the civil engineering community. destroy information essential for a proper regulatory decision. Suppose that Model A leads to a release that exceeds the e DOE has submitted a study plan for sito char- regulatory requirements while Model B acterization studies to calculato Pre , the prob- leads to a release that does not exceed ability of magmatic disruption of the Yucca the requirements. Suppose further that j Mountain candidate repository site. Several half of an expert panel prefers Model A statistical problems with the proposed study while the other half prefers Model B, so plan were identified, including: that each is given equal weight. If the 9

weighted rolcaso does not exceed the Modification of the ROSA Large-regulatory requirements, the NRC may Scale Test Facility for AP600 be lod to believe that the rcpository design is acceptable. Since only one Confirmatory Safety Testing model, A or B, can be correct, this Gono S. Rhee, DSR/RPSB conclusion is j,ustified only if thoro is high assuranco that Model B is the correct Westinghousc Electric Corporation has submitted model. However, sinco half of tho exports the Advanced Passivo 600 MWo (AP600) nuclear profor Model A, thero is no such power plant design to the NRC for design cortifica-assuranco. in this situation, the NRC's tion. The Offico of Nuclear Regulatory Roscat ch is conclusion should be based on the conducting confirmatory testing of AP600 safety realization that neither Model A nor systems. Such confirmatory testing is conducted Model B can be ruled out. Accordingly,it to help the NRC staff ovaluato tno safoty of the was recommended that the model AP600 teactor systems, results be presented separately, to0cthor with their associated weights in contrast to the current generation of reactors, and tho accompanying justifications. this now design features passivo safety systems This will allow the regulatory decision to for mitigating accidents and operational tran-reflect tho full rango of scientific sients. Since thoso passivo safoty systems rely on uncertainty. gravity-driven flow, the driving forces for the safety functions are small compared to those available under conventional pumped systems, Thus, the performance of thoso now safety systems may be g

20th Water Reactor Safety dv rs ly affected by small variations in thermal Information Meeting hydraulic conditions. Also, the operation of the HES hold its 20th Water Reactor Safety Informa. passivo safety systems posos challenging com-tion Mooting at the Bethesda Marriott Hotel on Oc. putational problems for curront thormal-hydraulic tober 21-23,1992. The mooting was attended by system analysis codos in that the current codos more than 500 participants representing the U.S. wero not sufficiently assessed for conditions of Government, the nuclear industry, and foreign low pressuro and low driving heads and for the governments. Commissioner E. Gail do Planque multiple flow paths used in the AP600 design.

was the keynoto speakor, and Eric Bockjord, Di- Thoroforo, a full-height, full-pressuto integral of-rector of the Office of Nuclear Regulatory Re. focts test facility was nooded for confirmation of search, spoke about NRC roscarch, accomplish. AP600 safety system performanco and for indo-monts, and prospects. pondent assessment and validation of computer analysis codos.

Ei0hteen sessions woro scheduled in which re- For confirmatory testing, it was determined that searchers prosented findin0s in such areas as so- the most cost-offectivo route was to modify an ex-vero accidents, primary system integrity, thermal isting full height, full-pressure test facility rather hydraulics, advanced reactors, human factors, than build a new one. Thus, all the existing integral advanced control systems technology, aging, offects test facilities, both in the United States and earth sciences, probabilistic risk analysis, and abroad, woro screened to select the best candi-structural and seismic engincoring. Also included dato. The critoria for tho initial screening included was a special session presented by the Electric the size, f acility configuration similarity, availability Power Roscarch Institute (EPRI) on its nuclear schedulo, willingness to share the coot, and the safety research and development programs. The ability to ontor into a confidential agroomont proceedings will be published early in 1993. with Westinghouse for handling proprietary 10

i information. This szeeninC revcated that the best (ECC) components, typical of those in the refer-candidato was tM Rig of Safety Assessment enco PWR, are included in ROSA V. The current (ROSA) Largo Scale Tost Facility in Japan. To ROS A V f acihty is very similar to the ROSA IV f acil-confirm those initial resurts and to determino the ity described in a January 1989 JAERI report, extent of modification necessary to simulato the JAERI-M-84-237,

• ROSA-IV Largo Scalo Test AP600, the Idaho National Engineering Labora- Facility System Description."

tory (INEL) was contracted to perform a compara-tive study between ROSA and AP600 using the Facility Modification RELAPS/ MOD 2.5 code. This study was published as NUREG/CR-5853. " Investigation of the Appli-cability and Limitations of the ROSA large-Scato A comparison betwoon the existing ROSA facility and the AP600 design showed that ROSA did not g

Test Facility for AP600 Safety Assessment" (Do- contain the key components important for safety cember 1992); the main points of the study are responso of the AP600. It was not obvious how presented here. much hardware modification to the ROSA facility would be needed to simulato the AP600. The fidel-ity of simu'.ition must be balanced against the as-Existing ROSA Facility sociated cost. The fidelity should bo high enou0h to result in a facility capable of producing data for The ROSA test facility is located at the Japan codo assessment covering the major AP600 phe-Atomic Energy Research Institute (JAERI) in nomena in the correct sequence. At the samo Tokal Mura, Japan. The existing facility, called time, the cost and the schedule have to be afford-ROSA-V LSTF, is a 1/48 volumetrically scaled, full-height, full-pressure conventional Westinghouso able. To make an optimum choico, INEL was E asked to consider four levels of modifications in j four-loop pressurized water reactor (PWR) simula- progressively more extensive stages. The first tor. The referenco PWR used for the ROSA-V facil- level of modifications was the absoluto minimum, ity design was very similar to the Trojan Plant. and the fourth level was tho most inclusive among

, When compared to AP600 ROSA V represents the four levels. To judge the fidelity of simulation of 1/30-volume scaling. The ROSA facility includes each level of modification, the following steps two primary loops, each containing one cold leg, were followed, one hot leg, an activo inverted-U tube steam gen-erator, and an activo reactor coolant pump. Each Criteria lised for Evaluating Each Level of ROSA V loop represents two of the reactor loops Modification lumped together. The loop horizontal legs are sized to conservo the scaled volume as well as the in evaluating each level of modification, the ratio of length to the squiuo (oot of diameter, RELAP5/ MOD 2.5 code was used as a primary tool LJDO.5, in order to correctly simulate the two- for comparing the predicted behavior of ROSA phase flow rogime transitions. The inverted U with that of AP600 for selected accident scenarios.

tube steam generators are fulllength and contain This approach is based on the assumption that 6 141 tubes. Tube thickness, outsido diameter, and RELAP/ MOD 2.5, although not assessed against length aro identical to those of the referenco PWR. AP600 systems test data, will show major trends in A pressurizer is connected to one of the hot legt overall behavior in such global parameters as de-The ROSA V vessel includes an annular down- pressurization rate, mass inventory, and energy comer and contains 1064 full-length electrically distribution. The validity of this assumption is par-heated rods capable of operating at 10 MW, or tially supported by the fact that the RELAPS code 14 percent of scaled full power for the referenco reasonably matched experimental data from PWR, The heater rod dimensions and pitch are the many different facilities, of different sizes, which same as for the 17x17 fuel assembly used in the were designed to simulato current PWRs. Since referenco PWR core. Emergency core cooling the thermal-hydraulic processes involved in 11

current reactors and passive mactors are funda- ance lines, a passivo residual heat removal montally the Samo, it is likely that the RELAP5 system with simulated secondary cooling, code will also showthe majo trends in AP600 and automatic depressurization system with ROSA, even though the predictions may not be as stages 1 through 3 on top of the pressurizer, accurato until further improvements are made in stago 4 on tho hot log, and minimization of such areas as mathematical modeling of condon- the pump loop scallengths.

sation in the presence of noncondensibio gases, boron transport, and the computation of level 11. Second-level modifications were derived tracking and thermal stratification in a tank, from the analysis of the first level modifica-tions and included the addition to the first lovel c'a properly scaled AP600 pressurizer, Accident Scenarios surge lino, and surgo lino connection.

In determining the ability of the ROSA facility to simulato tho AP600 roactor, tho following accident 111. Third level modifications included all the above plus the splitting of one cold log into scenarios were analyzed with RELAPS in both the two to incorporato two CMTs. A CMT is con-AP600 and ROSA.

nocted to each split part of the cold 100, as in e A 3-in. diameter break in a cold log the AP600.

e A 1-in. diameter break in a cold log IV. The fourth-level modifications resulted from the initial analyses, more in-depth inspection e A 3-in. diameter break in a pressure balanco of the plant design differences, and discus-line betwoon the coro makeup tank (CMT) sions with representatives of the Japan and a cold 100 Atomic Energy Research Instituto, who owns the ROSA facility. These included the first e One and throo tubo ruptures in a steam gon- and second levels coupled with appropriato erator upper head flow paths and adding an in-containment refueling water storage tank e A main steam lino break and two CMTs. Since thero is only one cold leg in each loop, CMT cold log pressure bal-Those transients were selected because they anco lines are connected to the same cold challengo the passivo safety features of the log for most transients when asymmetry be-AP600. The processes and governing mecha- tween the two CMTs is not expected, but nisms participating in these transients span a connected to a different cold leg for a reasonably complete range of impoitant non symmetric pressure-balance-line-break phenomena- scenario.

Different Levels vi Modifications The comparisons among RELAPS calculations for different levels of modifications showed that the The four levels of facility modifications that were first level modifications woro capable of reason-considered are defined below. ably representing AP600 behavior during the early portion of most transients when asymmetric be-I. First-level modifications woro determined havior between the two CMTs was not expected.

merely by inspection of the two designs, with The behavior in slow transients, or the latter part of only essential modifications considered, in- fast transients, was distorted panly because of the cluding the addition of the passivo safety fea- larger friction and metal mass to volume ratio tures not present in ROSA: the coro makeup used in the calculations and partly because of the tank (CMT) and appropriato pressure bal- other differences in hardware, most of which were 0

12

i eliminated as the Level I to Level IV modifications thero were additional differences that could havt woro mado. been eliminated, o g., completo elimination of the primary coolant pump loop seal, different steam Sinco the first-level modifications have only one generator tubo thickness, and different hot leg CMT, it can not simulato a situation in which two volumes. Lovel IV modifications aro being CMTc act differently, e.g., a break in the pressure implemented under a contract with Sumitomo balanco lino to one of the CMTs. On the other Heavy industries, which designed, constructed, hand, splitting a ROSA cold leg into two to be ablo and has been operatin0 the ROSA facility as a to attach a CMT to each part of the split cold leg contractor to JAERI. The facility modification is (Level lil modification) did not produco good re- scheduled to be completed within a year, and a sults becauso, unlike AP600, the split cold legs series of tests will bo initiated in early 1994. The had to be merged before they entor the vessel funds for testing will be provided by JAERI.

sinco another largo holo could not be drilled into the vesschvall. Therofore, in the Level IV modifica-tion, splitting a cold log was not incorporated. In- RESEARCH NEWS is published by the USNRC stead, both of two CMTs woro connected to the Offico of Nuclear Regulatory Research, Ann F.

samo cold leg when asymmetry between the two Beranek, Editor.

CMTs was not expected, and one of the two CMTs is connected to a different loop when asymmetry Comments, sug0estions, and artic!cs for futuro is-is expected. This arrangement produced reason. sucs should bo directed to the Editor, RESEARCH abloapproximationof thebehaviorof twoCMTsin NEWS, Office of Nuclear Regulatory Roscarch, AP600. U.S. Nuclear Regulatory Commission, Washing-ton, DC 20555.

Conclusion Copies of NRC and other Government publica-tions may bo purchased from the Government Level IV modifications resulted in an acceptable Printing Office at the current GPO prico, Informa-fidelity of AP600 simulation in terms of overall tion on current GPO prices may bo obtained by system behavior as judged from the global contacting the Superintendent of Documents, parameters such as depressurization rato and U.S. Government PrintinD Offico, Post Offico Box liquid mass inventory. Further modifications were 37082, Washington, DC 20013-7082, telephone not considered to be cost effectivo even though (202) 512-2249 or (202) 512-2171.

13

Printed on recycled paper E

Federal Recycling Program E

UNITED STATES HRST CLASG Malt NUCLEAR REGULATORY COMMISS!ON POGTAGE AND FEES PAID WASHINGTON, D.C. 20555-0001 usNac PERMIT NO. G 67 OFFICIAL BUSINT.3S PENALTY FOR PRtVATE USE. $300 12055513c531 1 1401AF19L19F U S t!C C-0 A D

DIV FOIa & CUHLICATIONS -VCa TPS-PCC-NUCEG w$f 7: G T T: DC 20555 1