scaling in nuclear reactor system thermal-hydraulics

Nuclear Engineering and Design 240 (2010) 32673293

Contents lists available at ScienceDirect

Nuclear Engineering and Design

journa l homepage: www.e lsev ier .com/

Scaling rau

F. DAuriDipartimento d , 5612

a r t i c l

Article history:Received 13 FeReceived in reAccepted 4 Jun

nginnstitget acaling-the aactoare dinformncep

l pathp forabiliteafte

properly scaled ITF and the availability of qualied system codes are identied as main achievements inNRSTH connected with scaling. The roadmap to scaling constitutes a unied approach to scaling whichaims at solving the scaling puzzle created by researches performed during a half-a-century period.

2010 Elsevier B.V. All rights reserved.

1. Introdu

Nucleartutes onetechnologyin the 1950as a responslatory Com(e.g. US AEfollowing thation and DSafety ofNuin 1992 (see

The behsimultaneosient condittemperaturinvestmenttion capabicontinuous

CorresponE-mail add

(F. DAuria).

0029-5493/$ doi:10.1016/j.ction

Reactor System Thermal-Hydraulics (NRSTH) consti-of the pillar disciplines for the safety and designofwater cooled reactors. The disciplinewas establisheds. Its development had a strong impulse since the 1960se to the safety needs put by the US NRC (Nuclear Regu-

mission), previously AEC (Atomic Energy Commission)C, 1971). The word system became of common usee OECD/NEA/CSNI (Organization for Economic Cooper-evelopment, Nuclear Energy Agency, Committee on theclear Installations) Conference held inAix-En-ProvenceOECD/NEA, 1992).

avior of two-phase mixtures in NRSTH involves theus presence or the occurrence of phase change, tran-ions, non-developed ows, heated surfaces at differentes and complex geometries. Notwithstanding huges and thousands of engaged researchers, the predic-lities of current numerical (or computational) tools arely under scrutiny and improvements are requested and

ding author. Tel.: +39 50 2210 359; fax: +39 50 2210 355.resses: [email protected], [email protected]

needed. Namely, question marks are put in relation to the scaling-capabilities of the tools or the scaling-quality of the results. Thisis discussed below.

Two topics (or issues) of the NRSTH testify of the complexity ofthe discipline and have been part of its half-a-century history:

The stability of natural circulation in boiling systems, see e.g.SOAR on BWRS (State of the Art Report on Boiling Water ReactorStability) (DAuria et al., 1997a).

The scaling, see e.g. J. NED (Journal Nuclear Engineering andDesign) (Special Issue, 1998).

Both topics (related connection also mentioned in the paper)have been the challenge for the investigation by individual scien-tists who contributed to the development of the NRSTH disciplineat least one time during their professional life.Moreover, thewordsclosure-of-the-issue are not part of conclusions of researches per-formed so far in relation to each of the two topics.

The present paper dealswith the second of the above challengeswith the generic objective of proposing a roadmap suitable forthe closure of the issue. Namely, the target of scaling is the watercooled nuclear reactors whose nominal operating conditions arecharacterized by high pressure (of the order of 16MPa), high ther-mal power (up to 4500MWth), high power density (either linearpower up to 50KW/m in one-cm diameter fuel rod, or ux up to

see front matter 2010 Elsevier B.V. All rights reserved.nucengdes.2010.06.010in nuclear reactor system thermal-hyd

a , G.M. Galassii Ingegneria Meccanica, Nucleare e della Produzione, University of Pisa, Via Diotisalvi 2

e i n f o

bruary 2010vised form 25 May 2010e 2010

a b s t r a c t

Scaling is a reference key-word in ecooled nuclear reactor technology coscaling-issue, i.e. the impossibility tois discussed at rst. The so-called sc(or scaling state-of-art;) is given ofscaling-issue in the area of Nuclear Refactors for Integral Test Facilities (ITF)a two-fold nature: (a) classifying thewhich are relevant to scaling: the cointroduced; (b) establishing a logicapuzzle). In this context, the roadmaissue when demonstrating the applicplants. The code itself is referred herlocate /nucengdes

lics

6 Pisa, Italy

eering and in physics. The relevance of scaling in the waterutes the motivation for the present paper. The origin of thecess to measured data in case of accident in nuclear reactors,controversy constitutes an outcome. Then, a critical surveyttempts and of the approaches to provide a solution to ther System Thermal-Hydraulics (NRSTH): dimensionless designistinguished from scaling factors. The last part of the paper hasation about achievements in the area of thermal-hydraulics

ts of scaling-pyramid and the related scaling bridges areacross the scaling achievements (represented as a scalingscaling is proposed: the objective is addressing the scalingy of system codes in the licensing process of nuclear powerr as the key-to-scaling. The database from the operation of

3268 F. DAuria, G.M. Galassi / Nuclear Engineering and Design 240 (2010) 32673293

1Mw/m2) and the volume bounding the two phase mixture is aslarge as a few hundred cubic meters. Those gures of merit refer tothe typical primary circuit of a LWR (Light Water Reactor); if oneincludes the containment the volume to be considered comes to avalue closepaths havindistinguish

Whatevescaling issuprohibitivethe ability oof nuclear r

An agreethe framewtainty) appto supportJ. STNI (Jou(Special Issthermal-hyThe relevanby the US Nproposing tmethodolog

A spot-tcept of scal

(1) Archimity of ascaling-history)

(2) Galileoweightand prothis is aTower o

(3) The Budimensof equaperformless parprovideand wiltems forthe invesystem

(4) In Alamnuclear

(5) Dimensdimenslike Re(Grashoas repoGuidelinthe text

The scaNamely, funumber ofthan condufaith towardreects to tadded:

(a) Lookingtical pip

and dispersed regime are (only) some of the regimes thatcan be distinguished. Empirical equations are (still) neededto characterize the ow-regimes and the ow regimes tran-sitions: some are empirical but others are physically based, i.e.

elvinatiopipetheerimkingucles ofe. N

d muvelopd in

positatioationd wvali

scaliny anon athe spapeility>calinmajundeingedgeermathanLevycodementffecttegrre, s

anslitiesn asn is rcoolerywtionutesl-hyne inmajocalinentient ar reaer m

bac

ling,teriztecteCondto 105 m3. Moreover, inside the bounding volume, owg hydraulic diameters in the range 103101 m shall beed.r is the target for the scaling analysis, the origin of the

e can be characterized as follows: it is infeasible (or cost) to perform meaningful experiments at full scale andf numerical tools designed to simulate the performanceeactors can be proven only at reduced scale.able solution for the scaling issue is mandatory withinork of the so-called BEPU (Best Estimate Plus Uncer-roach where realistic computational tools are appliedthe licensing of NPP (Nuclear Power Plants), see e.g.rnal Science and Technology of Nuclear Installations)ue, 2008). In this case, the qualication of systemdraulic codes against scaling remains an open question.ce of the issue canbe recognized fromthe acronymusedRC when planning the basis for the BEPU, i.e. when

he CSAU (Code Scaling and Applicability Uncertainty)y (see e.g. J. NED, 1990; USNRC, 1989).

ype historical excursus may serve to focus on the con-ing:

edes (287212 B.C.) was aware of the heating capabil-small mirror when exposed to the sun light. Then theup idea came (whether or not this is a legend or theto burn the Roman ships by large mirrors.(15641642) is reported to have dropped a ten-poundand a one-pound weight off the Leaning Tower of Pisa,ved that both fall at the same speed. Whether or notlegend or the history, he used a large-scale facility (thef Pisa) and real (full scale) cannon balls.ckingham Theorem (Buckingham, 1914) allows theionless combination of variables which are part of a settions. The set of equations may be representative of theance of a physical system. The choice of dimension-ameters is not unique: the Buckingham Theorem onlys a way of generating sets of dimensionless parametersl not choose the most physically meaningful. The sys-which these parameters coincide are called similar andstigation performed in a down-sized (or scaled down)is valid for a larger system.ogordo (1945), the rst atomic bomb test, the involvedscientists decided to perform a full-scale experiment.ionless or scale-independent quantities (however,ionless is not a synonymous of scale-independence)(Reynolds), Pr (Prandtl), Nu (Nusselt), Fr (Froude), Grf), are of widespread use in thermo-uid-dynamics,rted in any text-book (e.g. Todreas and Kazimi, 1990).es for scaling complex nuclear systems are provided in-book by Levy (1999).

ling controversy can be drawn from the items above.ll scale tests (prototypical) have been preferred in asituations for addressing technological issues, rathercting scaled experiments, thus testifying of the lack ofs scaling (i.e. the scaling controversy). The controversyhe area of NRSTH in which case the following can be

at the predictability of two-phase regimes in a ver-e in steady conditions: bubbly, slug, churn, annular

a Kequtheingexp

(b) Looa ntencodeDeuseopequicusethe

So,ing anbasedwithinevanttest facThus, sfor twolack of

Havknowltem thother1990;puterexperiarate Etens Inthermouid.

Thecapabibe taketigatiowater

A veinnovaconstitthermawas dooncilewith stive idto presnucleaconsid

2. The

Scacharacters deInitialHelmholtz instability limit for stratication. Relatedns are a function of (at least) pressure and diameter of. Therefore, the possibility of scaling-up, i.e. of conrm-scaling validity, shall be excluded in the case of lack ofental data.at the predictability of the transient performance of

ar reactor where a numerical code is needed: severalclosure equations (see e.g. Levy, 1999) are part of theow, the paradox: closure equations in order to be quali-st be derived in the conditions of Steady State and Fullyed (SS and FD) ow; however, they are unavoidablytransient (as opposite of SS) and non-developed (as

e of FD) owconditions. Thus, even though the (closure)ns are qualied, no scaling-up outside the range of qual-is justied. The argument at the last sentence can be

ithin the present context, i.e. the difculty to conrmdity of scaling laws.

g is a-priori questionable inNRSTH, i.e. before attempt-alytical derivation, and it is a-posteriori questionablerguments at items a) and b). This is basically agreedcientic community as given in the conclusion of a rel-r on the subject, Ishii et al., 1998: . . . the design < of acannot completely satisfy all the scaling requirements.

g distortions are inevitable . . . Distortions are encounteredor reasons: - difculty to match the local scaling criteria; -rstanding of the local phenomenon itself. . ..this in mind, use is made hereafter of the establishedand of the recognized achievements in nuclear sys-l-hydraulics. The knowledge and thendings inNRSTH,in an dozen text-books (see e.g. Todreas and Kazimi,, 1999, already mentioned), are embedded in the com-s. The basis for the knowledge is constituted by theal data-base from the operation of a few hundreds Sep-Test Facilities (SETF, see OECD/NEA, 1993) and a few

al Test Facilities (ITF, see OECD/NEA, 1987, 1996). Fur-caling refers to space, pressure, power and nature of the

wer to the licensing relevant question how the scalingof currently available system codes can be evaluated? canthe specic objective for the paper. The eld of inves-estricted to the predictability of transient scenarios ind reactors.ide literature exists in relation to scaling. Thus, bringingin terms of new data or new elaborations of equationsan awkward goal at a timewhen investments in systemdraulics are largely reduced (i.e. compared with whatthe 1970s to 1990s). Thus, an attempt is made to rec-r approaches and achievements in NRSTH connectedg, keeping in mind the scope and the specic objec-ed above. Namely, it is not the purpose of the papercomprehensive and consistent review of the scaling in

ctor system thermal-hydraulics, although an attempt toain approaches has been made.

kground information

see the denition given for the scaling issue, can also beed as the capability to transpose the values of parame-d or calculated under an assigned set of Boundary anditions (BIC) to a different set of BIC. In the present case

F. DAuria, G.M. Galassi / Nuclear Engineering and Design 240 (2010) 32673293 3269

BIC shouldbe intendedashavingawidemeaning including thegeo-metric data. Thus, in Nuclear Reactor System Thermal-Hydraulicstypical parameters which may need to be scaled-up are either uidand solid material temperatures, uid velocities, void fractions andmass ow-set of geomduced per u

Althoughcarried outthe use of NRSTH, i.e.mentionedof using Frcooled reacthe pressurplex accidethermal-hypressure drof Freon asfrom thoseuse of the local phenin relationnarios of pafrom watercredit (or canalyses wh

A compr(SOA) in theof several sNeverthelesthe currentas a non-co

Differenapproachestively.Moreand the devand 2.4. Otperformingfrom the di

2.1. Differe

It is wenuclear rearepresentatet al., 2010)geometric d

The macrthe contaconditionvoid fractshall be cotions beloinventorishall be c

The comprst, onee.g. the puthe uppercomer, (c)at the breUpper Tieregion. Ththe pheno

dow for the evolution of the phenomenon, shall be identied.Finally, suitable links shall be established between evolutions ofthe variables in the model and in the prototype.

The micro-scale, or local-scale. The attention is focused to oneetri

e ordessedfromtuallyxpec. In

rimenalueion oe PCTbove

mprddres ab

jectivrst tScaliacili, letpliesdingexpcipletestolatioady rpropversy

e cla

toLWRy 10ble eutese prion 1

cylintitutehavirod sble fuel roture.ng stientamed), bfracttrantitutese ofing Sm, sreatavityr is te: thrates; the relevant BIC that vary due to scaling are theetric dimensions, the power of the core, the power pro-nit length, the pressure of the system.a large amount of important research work has been

in relation to uid-to-uid scaling (e.g. Ahmad, 1973)uid different from water appears to be not justied inwater should not be considered as a BIC in the sensein the previous paragraph. For instance, the advantageeon instead of H20 in experiments simulating watertor conditions derives from the possibility of loweringe and thus the cost of the research. However, com-nts exist where temperature differences may drive thedraulic scenario evolution in conjunction with localops (also affected by local voids). In this case, the useworking uid may drive the transient to situations farpredictablewhenwater is circulating in the system. Theuid-to-uid scaling technique is more justied when

omena are of interest. However, caution remarks applyto the extrapolation possibility to water related sce-rameters measured in experiment with uid different(see e.g. Tain et al., 1995). In the following, scaling-redibility for scaling) is given only to experiments orere water constitutes the working uid.ehensive and suitable synthesis or a State-Of-the-Artarea of scaling in NRSTH would imply the engagement

cientists and an effort outside the present framework.s, an attempt is made in this chapter to give an idea ofunderstanding about scaling. This shall be consideredmprehensive scaling SOA.t meaningful scales for a scaling study and differentto scaling are discussed in Sections 2.1 and 2.2, respec-over, the connectionbetween scaling and the ITFdesignelopment of numerical codes is outlined in Sections 2.3herwise, the already mentioned inherent difculty ina comprehensive systemscalinganalysis canbederivedscussion in Section 3.

nt scales for a scaling study in NRSTH

ll established that thermal-hydraulic phenomena inctors are characterized by different scales; connectedive diagrams canbe found in the literature (e.g.McClure. Three sectors for scaling analyses are recognized withimensions playing a major role in the classication:

o-scale, or system-scale. The whole primary circuit, orinment, constitutes the object of scaling. In steady states, uid and solid surface temperatures, uid velocities,ion and pressure drops are example of variables thatrrelated in the model and in the prototype (see deni-w). In transient conditions, system pressure, total masses and level distribution are examples of variables thatorrelated.onent-scale, or zone-scale, or phenomenon-scale. Atof the following should be identied: (a) a component,mp, or the separator, or the core region, (b) a zone, e.g.plenum, or the connection between cold leg and down-a phenomenon, e.g. the Two-Phase Critical Flow (TPCF)ak, or the Countercurrent Flow Limiting (CCFL) at thePlate (UTP), or the Critical Heat Flux (CHF) in the coreen, the variables affecting the component performance,mena occurring in the assigned zone or the time win-

geomof thaddrexistevenare etime?expethe vlocatat ththe a

A coshall alet itemthe obto thetionalEffect f

Nowalso apin worvaluesor prindownextrapas alrecan becontro

2.2. Th

HowMWthimatelnoticeaconstitwith thin Sect

Thinconsrods

Thevariathe fpera

Duritransthe stionevoidheatcons

In caCoolsyste

In a gby gr

Watetotypc point or better to a small zone (volume as small aser of 1mm3) of the reactor system. The basic questionis: does a qualied experiment or qualied calculationswhich data can be derived that have the same values (orscaled-down values with available scaling factors) that

ted in the assigned small zone of the reactor at a givenother terms, do there is the possibility to afrm that, eithertal dataor code calculation results, areable to characterizeof a thermal-hydraulic parameter valid locally (e.g. thef PCT or the steam velocitylocation) with no or small uncertainty?. The answer toquestions is no.

ehensive scaling analysis in nuclear reactor technologyss and actually addresses each of the three sectors (bul-ove). However, reasonable demonstration of fulllinges is provided, from the available literature, in relationwo sectors, as discussed below (see e.g. the use of Frac-ng Analysis including data from Integral and Separateties).us consider the question at the 3rd bullet item whichto the two previous bullet items with proper changes

. Two sides of the question are distinguished: (1) theected at full scale are unknown; (2) there is no theoremwhich allows the attribute qualied either for scaled-

data or for calculation results which, eventually, needn. Therefore, the strict answer to the question is no,eported. This denitely applies to the micro-scale andagated to other two scales, also reecting the scaling mentioned in Section 1.

ssic approaches to scaling

simulate the transient performance of an up-to-4500(also called prototype hereafter) by a less than approx-MWth ITF (also called model hereafter), with thexception of LOFT (Loss of Fluid Test, OECD/NEA, 1991),the classic scaling problem. Key aspects connectedoblem (see also the operating conditions values given) can be summarized as follows:

drical barswith external diameter of the order of 0.01mthe nuclear fuel rods in any LWR (up to about 50,000

ng active length of about 4m are part of the core).urface (or clad) temperature, Tw, constitutes a targetor the scaled design; linear power, q, produced insideds is among the key parameters affecting the clad tem-

eady-state (nominal operation) as well as in case of a(accident), Tw is affected, when the core geometry isfor the model and the prototype, by q (already men-y uid pressure, temperatures and velocities and byion. Namely, the inuence of uid velocities upon thesfer coefcient may impose the need for pumps if Tws an important variable for the simulation.accident conditions theperformanceof EmergencyCoreystems, e.g. affecting the coolant mass inventory in thehall be simulated in the scaled system.number of transient conditions core cooling is ensured: thus consideration of gravitational heads is essential.he accepted working uid in the model and in the pro-us, large changes in the steam and liquid properties


with pressure may impose the condition of full pressure in thesimulation.

The power-to-volume scaling, typically associatedwith the full-pressure, fuas the rstproblem indesign factoance equatiall together

It is easiassociatedthe casewh(a) cost of ttion of therovercome tformed by I

Startingfrom US NRthermal-hySection 2.2.

At somei.e. beginnitools and olated resultuncertainty(Code Scaliwise, the coprimarily in

During tcialists fromand Taiwanthe scalingdesign of adachievemen

Asalreadnumber of tissue and e2004). Seleemphasis isare cross-co

The connSection 2.2.

Withinrecommend(however, tall researchwith Zubercode is putprocess. Th

2.2.1. The pThe nee

capable of constitutedtradictory rconstraintsNRC (Nuclea suitable eCooling SysCoolant Acc

Primordfying the dein the US, s(1969) andarchival for

Karwat (1983), and Kiang (1985). Till the TMI-2 accident, i.e. 1979,the attention for the simulation was focused toward the LB-LOCA(LargeBreak LOCA), however in the post-TMI the above referencedpapers also considered SB-LOCA (Small Break LOCA) phenomena

onalen inilitief) wor thI-2 aand

this ptheight,ling. Howandi

e ree pralize

rstototyion otransionsionlactu

desigplicadim

itiesles once,ex baobtadesigynthmeDAuimulaof d

rdwaity vahar

ing zessel); (e

v (HLPRZ)LegGen

Outlees, Prary(Uppperinfo

ed thconimpingase oeter armorll-height and time-preserving requirements canbe seen(in historical terms) proposed solution to the scalingNRSTH. This is discussed in Section 2.2.1. In that section,rs derived from basic principles (i.e. not even the bal-ons) and suitable for building NPP simulators are listed.ly seen that the power-to-volume scaling together withrequirements has two main drawbacks, specically inen transients governed by gravity heads are concerned:he facility; (b) strong impact upon the transient evolu-mal power release from passive structures. In order tohese drawbacks, fundamental research has been per-shii and is outlined in Section 2.2.2.from the 1960s, Zuber, primarily with the supportC, provided a continuous contribution to key issues indraulics including the scaling. His ideas are discussed in3.point in the history of system thermal-hydraulics,

ng of 1980s, the issues of validation of computationalf evaluation of uncertainty associated with any calcu-s became relevant. The connection between scaling andis discussed in Section 2.2.4 dealing with the CSAU

ng, Applicability and Uncertainty) procedure. Other-nnection between scaling and validation is mentionedSection 4.2.

he last two decades, i.e. 1990s and 2000s, reactor spe-Russia and Far East Countries (primarily South Korea

) made available their contributions to the solution ofissue and the same issue became important for thevanced reactors. Thus, a short survey of related scalingts is provided in Section 2.2.5.ymentioned (e.g., see J.NEDSpecial Issue, 1998), a largeop level scientists were engaged to address the scalingxcellent scaling reviews can be found (e.g. Yun et al.,cted ndings are summarized in Section 2.2.6, wheregiven to selected-representative scaling factors thatrrelated with design factors from Section 2.2.1.ection between scaling and licensing is emphasized in7.the present framework and consistently with Zuberations the need of hierarchy among scaling factorshis idea is shared or acceptedby themajority, or evenbyers in the area), is recognized. Hereafter, inconsistentlyrecommendations (discussed below) the computer

at the center of the attention for performing the scalingis is characterized as the key-to-scaling.

ower-to-volume scaling and the design factorsds to design and to construct experimental facilitiessimulating the NPP performance, noticeably the PWR,a challenge for the involved scientists due to the con-equirements coming from safety relevance and cost. This happened the rst time in the 1960s when the USar Regulatory Commission, AEC at that time) requestedvidence of the capabilities of the ECCS (Emergency Coretems) to keep cooled the core following LOCA (Loss ofident).ial set of scaling parameters were introduced for justi-sign and the operation of Semiscale and Loft facilities

ee e.g. the pioneering works by Carbiener and ChudnikYbarrondo et al. (1974), that later on were put in anm by Navahandi et al. (1982), Larson et al. (1982),

(additiare giv

Facthe LstJapan fthe TM(1994)

At1990s,full-hethe scationedNavah

(1) Tim(2) Tim(3) Ide

Thethe prreductpowerdimendimencriteriato thethe ap

Theof facilprincipdifferecomplfactorsderive

A sto-volu1985;PWR sgroupsITF haquant

Thefollowsure VPlenum(f) HL-Line ofh (Cold(SteameratorU-TubSecondUP-SEPDRY (U

Thebe notstrictlycriteria(accord

In cparamFurthedetails related to the role of Loft in relation to scaling,Section 3.4 with focus to LB-LOCA).

s like Lobi, Spes, Bethsy and the Rosa series (noticeablyere designed and built (or modied) in Europe and ine simulation of SBLOCA in PWR, after the occurrence ofccident; see e.g. the data base collected by DAuria et al.Ingegneri et al. (1997).oint in the history of NRSTH, i.e. decades 1980s andpower-to-volume scaling approach, time-preserving,full-pressurewas thepreferredpracticalway to addressissue within the area of ITF design, as already men-ever, three kinds of principles were distinguished (e.g.

et al., 1982)

ducing or linear scaling.eserving or volume scaling.d time preserving.

set implies the reduction of the linear dimensions ofpe by a given factor and, in transient conditions, thef time by the same factor. In this case, the amount ofsferred to the uid is reduced as the square of the linearfactor. The second set of criteria is at the origin of theess design factors discussed below and the third set ofally resembles the second set, with more exibility leftner, being the time preserving the main objective fortion.ensionless design factors are applicable for the designand experiments. They can be derived directly from thef conservation of mass, momentum and energy. As athe scaling factors are derived from the more or lesslance equations for single and two phase ow. Scalingined from simplied balance equations can be used ton factors.

esis list of design factors that characterize the power-scaling can be found in Table 1 (e.g. DAuria and Vigni,ria et al., 1988, in these papers, where a hypotheticaltor and an experiment are designed, respectively). Twoesign factors K can be distinguished dealing with there and the test BIC, respectively. Here, K is the ratiolue in the model/quantity value in the prototype.dware of the hypothetical ITF is subdivided into theones: (a) LP (Lower Plenum) of RPV (Reactor Pres-); (b) CO (Core); (c) CO-BY Core Bypass; (d) UP (Upper) UH (Upper Head); (f) HL-h (Hot Leg, horizontal part);, vertical part); (g) PRZ (Pressurizer); (h) SU-LI (Surge; (i) LS (Loop Seal); (j)MCP (MainCoolant Pump); (k) CL-, horizontal part); (l) DC (Down-Comer) of RPV; (m) SGIerator Inlet Plenum, Primary Side); (n) SGO (SteamGen-t Plenum,PrimarySide); (o) SGUTPS (SteamGenerator,imary Side): (p) SG UT SS (Steam Generator, U-Tubes,Side); (q) SG DC (Down-Comer, Secondary Side); (r) SGper Plenum and Separator, Secondary Side); (s) SG UH-Head and Dryer, Secondary Side).

rmation in the table is self-explanatory. However it canat the Kv value is the key for the design and its value isnected with the available nancial resources. Adoptedly thatmass andenergy in themodel areproperly scaledto Kv) in relation to the prototype.f LOCA, the break area should be scaled according to thet row6 in the table, i.e. the ratioAR/V shall be preserved.e, the break location should also be preserved consid-


Table 1Synthesis of dimensionless design factors characterizing the power-to-volume scaling (top level hierarchy factors).

No Design factor Scaling of: Zone of the ITF Notes

1 Kv =Kv-max Volume Each Zone The largest is the best.2 Kh -LI, LS

, SG D

3 KN.4 Kh RY5 KL =

6 KA7 KD. h zon

8 KW9 Kt =10 Kq

Kq 11 Kf.r

12 KT

13 KP14 Kr.s15 Kr.s16 Kp.h

17 Kth18 Kh.l19 KG

K20 K

21 KG.22 KT.E23 KG.24 KG.25 KT.F26 KRR27 KM

28 KVL

ering the prvs loop leng

The scaldesign factoparametersapply (i.e. ithe table):

Even thoudent amolist of par

The desigThe key rremainingminimizin

The reqution set ashall be gtric poweof cases)trically heshould beis properlenergy to(namely,energy at

The paramditions, al=1 Elevation change CO, CO-BY, UP, HL-v, PRZ, SUSGI, SGO, SG UT PS, SG UT SS

L. = 1 Number of loops System1 Elevation change LP, UH, SG UP-SEP, SG UH-D1* Length of horizontal

componentsHL-h, CL-h

=KAbr =Kv Cross and break area Each vertical zoneeq. = 1 Hydraulic diameter CO and SG UT, primarily. Eac

possible.=Kv Power CO, SG UT1 Time System=1 Linear power CO, SG UT

=1 Volumetric power.g.m. = 1 Fuel rod geometry and material CO

=1 Fluid temperature Each zone

=1 Pressure.T. = 1 Rod surface temperature CO (primarily) and SG UT.h.f. = 1 Rod heat ux.t.a. =Kv Passive heat transfer area Each zone

.p.s. = 1 Thickness of passive structures.e. = 1 Environment heat losses=Kv Mass ow rate=1 Fluid velocityp =1 Pressure drops (friction and

local)ECC =Kv ECC ow ECCSCC =1 ECC temperature

SL =Kv Steam line ow-rate SG UH-DRYFW =Kv Feed-water ow SG DC

W =1 Feed-water temp=1 Recirculation ratio

CP.ch = 1 Non-dimensionalcharacteristics for pumps andvalves

MCP

V.ch = 1 Valves of PS and SS

oper geometric location and the curve pressure-dropth, i.e. criterion at row 20.ing factors which give origin or are consistent with thers in Table 1 shall be classied as the primary scaling, as discussed in Section 2.2.6. The following remarksn addition to the notes provided in the last column of

gh an attempt is made to list scaling criteria indepen-ng each other, priority is given to get a comprehensiveameters suitable for the design of the PWR-ITF.n factors at rows 6, 7 and 20 may appear inconsistent.equirement is represented by item 20 (Kp =1). Thetwo parameters should be adjusted accordingly, i.e.g other (unavoidable) scaling distortions.

irement (target) at row 14 is ensured by the condi-t row 10, primarily. However, proper considerationiven to the materials and to the control of the elec-r supply when fuel rod simulators are (in the majorityelectrically heated. The material conguration of elec-ated rods (in the typical ITF, i.e. fuel rod simulators)considered in order to ensure that energy to the uidy scaled down (e.g. Carbone et al., 1980). The thermalthe uid shall be evaluated during the entire transienta LB-LOCA) evolution time, also considering the storednominal operating conditions.eter at row 20 includes the MCP head in nominal con-

so reported at row 27. The parameter at row 20 should

be checkeas far as p

The paramdrop distr

The paramparamete

2.2.2. The IRemarka

issue cameSchlegel etIshii et al. (1from this laacterizes th

(1) Recognan intecomprising conconsideing, (b)the locain the level.

(2) Identifyadvantabuilding, MCP, DC,C

In the case of LS, the pipe axis or the bottomedge of the pipe in the prototype should beconsidered.To study asymmetries.

* Not possible in general. Friction pressuredrop must be preserved (see also row 7).Related to uid ow.

e as far as Froude number scaling is preferred forhorizontal pipes.Energy to uid shall be scaled.Key target for transient simulation.Number of heated rods and of U-Tubes scaledas Kv . Volumetric power is a target.

Target for the design together with previouscondition.Initial condition. Target for transientsimulation.

Target for transient simulation.

Impossible to achieve. Resulting distortions tobe characterized.

Impact to be minimized.

Flexibility (e.g. orice optimization) needed.

ECCS injection ports shall be preserved.

Quality at steam line inlet shall be the same.Boundary condition.Target for the design.Boundary condition.

d all around theprimary loop and, in the secondary loopossible.eter at row26 should be a consistentwith the pressureibution in the secondary side of the steam generator.eter at row 28 can also be taken as a consequence of

rs at rows 6 and 20.

shii scalingble four decades contributions related to the scalingfrom Prof. Mamoru Ishii (e.g. Ishii and Jones, 1976;

al., 2009), with key document constituted by the paper998). Two topics/subjects (underlined below) are takenst paper and discussed hereafter; the second one char-e so-called Ishii approach:

izing the scaling hierarchy (i.e. when designing an ITF):gral system scaling is performed whose componentse the rst two levels, and the phenomenological scal-stitutes the third level. More specically, the scaling isred as follows: (a) the integral response function scal-the control volume and boundary ow scaling, and (c)l phenomena scaling. A top-down approach is pursuedrst two levels and the bottom-up approach in the third

ing the reduced length (and height) as a scalingge, also in connection with cost reduction whenup a facility.


Related to item (1), sub-item (a), eight established numbers areproposed to be preserved from non-dimensionalising the balanceequations: the Zuber (or Phase-change), the Sub-cooling, theFroude, the Drift Flux, the Friction, the Orice, the Timeand the Tconsistenttwo numbdepends upthese constupon the thare not partdesign leveInertia num

When pthe basis founder the iitem (b) arein Table 1.

A varietythe sub-itemnomena areequations, cotherwise,criteria in Tbreak owalso below)1/4, the breuidowarupstream thTable 1.

Related tequation [fize that, inthe prototythe equivaltributes tomthe considehereafter as

The auththe basis ofis reducedthe two-phevents are aby (KL)1/2 (to the verysingle andscaling is no

The resuto sub-itemdesign paraBoilingWat(Ishii et al.,

Kp =1 Kh =1/4 Kt = (Kh)1/ K = (Kh)1 Kq = (Kh) KA =KvKh KAbr =KA(

It shall bthe paramewhen Kh =1

The dirework is con

consequent demonstration of quality for the value of those factors.The demonstration can be achieved from the comparison betweencorresponding experimental data measured in ITF designed fol-lowing the Ishii-scaling and ITF designed based on Kh =1 and time

ving

The peciangagroegon to-authof h

eratithe

ing ot anorigd topleana

(200y me, or k

e (coerro

wreqvastethea se

nedationd Tetes ce toplema

e achationthis id merimknoeemed d

th exdingowlenotcesse me/or s

e useily cnesscod

the enallyin fosoles, whormapar

uirementhermal Inertia. The rst six (over 8) numbers arewith the design parameters listed in Table 1. The lasters include the parameter conduction depth whichon the thickness and the material of solid structures:itute a target for scaling and not a parameter dependingermal-hydraulic designer. Therefore, these parametersof Table 1: this implies that an attempt is made at the

l to fulll the targets put by the Time and the Thermalbers, but distortions are unavoidable.

erforming the scaling analysis under the sub-item (b),r reduced height scaling is introduced as discussedtem (2). Other parameters introduced under the sub-basically consistent with the design parameters listed

of local phenomena are considered when performing(c): when the variables characterizing the local phe-

the same as the independent unknowns of the balanceonsistency with design criteria in Table 1 is achieved;Ishii requirements are not consistent with the designable 1. An example of this is the design factor for thearea which is connected with the height (or length, seescaling criteria: in case the length scaling criterion is

ak ow area results 2 times smaller than the (ordinary)ea. This implies the considerationof homogeneousowe break, which cannot be among the design criteria in

o item (2) the authors emphasize the need to satisfy theL/D+]R =1 (symbols are given in Table 2). They real-any model, the ow area is necessarily smaller than inpe and, apart for the core region (as the authors note),ent diameter D is also smaller. Thus, reducing L con-ake easier the fulllment of the above equation. This is

ration at the basis of the Ishii-scaling and ismentionedIshii scaling attribute.ors found, consistently with the balance equations atthe design criteria in Table 1, that . . .if the axial lengthin the model, then the time scale is shifted inase ow natural circulation loops. In such a case, theccelerated in the scaled-down model by a factor givensymbol in Table 1) over the prototype. . . . This leadsimportant conclusion that for systems involving bothtwo-phase ow in a reduced length model, real-timet appropriate.lt from the overall scaling analysis, namely connecteds (a) and (b) of step (1) is constituted by the followingmeters for the Puma ITF, simulator of the Simplieder Reactor (the same symbols as in Table 1 are adopted)1998), already mentioned:

2

/2

1/2

Kh)1/2

e noted that all design parameters would coincide withters in Table 1 in the situation (not suggested by Ishii)(2nd bullet item in the list above and row 2 in Table 1).ct impact of the Ishii scaling upon the present frame-nectedwith theapplicationof the scaling factors and the

preser

2.2.3.A sp

term efrom Ktributithe coimpactconsidtion toincludrelevan

Thebe usethe comscalingZuberalreadpivot

(1) ThingnethetogingtuculClaeracodcomand

(2) Thculinuexpareagrtunwifunknaresucabland

(3) Theasfultheto

(4) (Fiityobityinfcomreqoplaws, as mentioned in Section 3.5.

rogress due to Novak Zuberl mention is deserved to Novak Zuber owing to his long-ement in the area of system thermal-hydraulics (e.g.er and Zuber, 1968; Zuber et al., 2007, see also con-CSAU in Section 2.2.4). In addition, the reputation ofors of his papers about scaling (see below), and theis personality upon the scientic community justify theon hereafter. An excursus through the Zuber contribu-scaling issue is given in two parts hereafter, the formerriginal writings from his paper, the latter summarizingalytical achievements.inal writings taken from the paper (Zuber, 2001), cansynthesize the author opinion toward the results of

x computational tools. Thus, simple and straightforwardlysis was proposed. Four paragraphs from the paper by1), are relevant within the present context (where, asntioned, the thermal-hydraulic code is proposed as theey-to-scaling, for performing scaling analysis):

mpulsive) vision of code complexity and of compensat-rs. Thedifculties inmodifying codes to accommodateuirements stem, in largepart, fromtheir complexity, i.e.numberof closure relations (the constitutivepackage),r with transition criteria and splines, each introduc-t of coefcients (dials). The latter may be adjusted orto produce an acceptable agreement between code cal-s and a specic set of experimental data (say, the Peak

mperature). However, this tuning procedure also gen-ompensating errors, which limit the applicability of thea different set of requirements or design. Thus, throughxity, these codes have already become more inexibleladaptive to changes.ievement and the evaluation of results from code cal-s. . . .one of the greatest concerns of any professionaleld should be the indiscriminate use of two- or three-

odels, which invariably claim a good agreement withental data. Yet, some of these formulations and codes

wn to be inadequate, awed and/or incorrect. The goodent may be explained only in terms of the carefullyials hidden in the code . . . Although good agreementperimental data may ensure the continuation of project, such formulations cannot contribute to the fund ofdge. Laws of variable coefcients and tuning dialsyet laws of physics. Yet these continuous claims tohave institutionalized the art of tuning as an accept-thodology for addressing and resolving technical issuescientic problems..r is (guilty and) code-jockey. Although cultures are nothanged, I see no reason to promulgate the inane waste-of the code jockey attitude. . . .Organizations that allowe jockey culture to prevail, will be inevitably relegatedvolutionary junk heap... . .) scaling may help. . . .An ever-increasing complex-rmulating and analyzing problems leads to inefciency,cence and evolutionary failure. By contrast, simplic-ich allows for parsimony, synthesis and clarity oftion, ensures efciency, survival and replication. Thisison (complexity versus simplicity) also provides thements and guidance for a success path in T-H devel-. . . . Then . . . scaling provides the means to process


Table 2Sample list of dimensionless scaling factors characterizing various approaches (High [H] and Low [L] level hierarchy scaling factors).

No. Scale factor Meaning Ref. * **Criteria in Table 1, rows:

1 (Gp/W)(cvk/) Mechanical compliance Wulff (1996) H 1, 8, 12, 19, 202 uG/W3 WHL,ma4 pPPG5 h0 G6 (f/g7 G2 (L/A8 G2(NZu9 1,1,0 W ressu10 GADS 11 (tph.w12 [fL/D+13 [(L/)/14 [Ach =A15 (gWL16 hfg(117 WDC/(18 (GADS +19 tph.w20 tph.w21 1/Sx

Nomenclatu

A= cross secAHT =heatedcp,v =uid spD=equivalenf= friction prG=mass owGADS =ADS (AGCR = criticalGECC = ECC (eGIRWST = IRWh, hfg = Enthajk,x;y =Superk= thermal cL= length (oL2 = two phM= (system)NSubc = sub-cNZu = Zuber nP=pressureSx = slip ratioV= (system)W=power (cWDC =powerWHL,max =he

* Hierarchy** Or reason# More prec

informa(and, thScalingquantifyinforma

Signicaby Zuber (12007), ZubeTwo acronyH2TS (HieraAnalysis). R

2.2.3.1. H2Tof the H2TS

To be comable and tThermal compliancex/W Effect of heat losses/W Effect of pumping powerECC/W Thermal effect of (ECC) injection) GCR/G Effect of break ow)/g L2V0 Flow inertia in 2- owNSubc)/pPP fA2HT Momentum ux changes in 2-tph.w/P Ratio of pressure change due to int. energy change to ref. p

tph.w/M Ratio of integrated ow-rate to reference mass/V)(g L/p)1/2 Ratio of core make-up ow to vessel volume]R Friction and local pressure drop(2/s)]R Transport time/conduction time]R Break ow area scaling

)/(f cp f3A)# 1- nat. circ. (operational mode))fA/W# 2- nat. circ. and sump circ.

Gh)DVI Down-comer uid heat-upG)/GIRWST IRWST injectionjk,x/L Momentum conservation along xjk,y/L Momentum conservation along y

Slip ratio along x

re Greektional area =Void fractionsection heat transfer area s = Thermal diffusivecic heat at constant pressure/volume =Thermal expansiot diameters 1,1,0 = Coefcient deessure drop coefcient = Conduction depth-rate h0 = Excess injectioutomatic Depressuriz. Syst) ow-rate p=Pressure drop dow-rate (core if no subscript) pPP = Pressure gainmergency core cooling) ow-rate tph.w =Duration of pST (in-reactor water storage tank) ow u= Internal energylpy, Latent heat of vaporization =f gcial velocity for phase k along x or y 0 = Function of NZonductivity = local pressure drof one component) f , g = liquid and stease (or boiling) length =velocity (un-chocmass 1-, 2-= single-phaooling numberumber Subscripts

ch= critical ow condalong x DVI =direct vessel involume f= liquidore, if no subscript) R= ratio model/protofrom down-comer walls

at losses, maximum value

scaling factors, [H] or [L].for non-compliance.isely dened in the original paper.

tion in an efcient manner, as required by competitiveereby, replicating) systems. . . . The Fractional Change,and Analysis approach . . . offers a general paradigm foring the effects that an agent of change has on a givention system..

nt scaling achievements can be found in the document991), and in the papers by Zuber et al. (1998, 2005,r (2001), Wulff et al. (2005), and Catton et al. (2005).ms can be used to characterize those achievements, i.e.rchical Two-Tiered Scaling) and FSA (Fractional Scalingelated key ideas are synthesized below.

S (primarily, Zuber, 1991; Zuber et al., 1998). At the basisthere are two main objectives (or needs):

prehensive, systematic, yet practical, auditable, trace-echnically justiable.

To createrequiremprovidingfactors.

The 1stergistic effbetween thplex proceIdenticatio1990) procalso bringsaddressedequations acriteria.

The 2ndcriteria forow depenanalysis is1, 12, 19L 18H 27, 20, 8, 19

21, 22L TPCFH 19, 5, 1, 13

19, 13, 27,10re Banerjee et al. (1998) L TPCF

H 1, 9, 20Ishii et al. (1998) H 20

L Struct. thicknessIf Kh =1/4

Reyes and Hochreiter (1998) H 2,6,8,13,196,8,13,19

L Struct. heat releaseTPCF

Yun et al. (2004) L Three-dimensional effectsityn coefcientpending upon break ow, see Ishii et al., 1998n enthalpy above initial enthalpyue to frictiondue to pumping powerhenomenological windowchange

u and NSubcp coefcientam densityked ow)se, two-phases

itionsjection (line)

type

a hierarchy among scaling factors and scaling design orents, thus eliminating the arbitrariness in scaling anda quantitative estimate of the importance of the scaling

tier, or top-down scaling analysis, examines the syn-ects on the system caused by complex interactionse constituents. Therefore a complex system or a com-ss is subdivided into pieces. The PIRT (Phenomenan and Ranking Table, e.g. see J. NED, Special Issue,

edure is applied, thus requiring expert judgment. Thisto the identication of the important processes to bein the bottom-up scaling analysis. Fluid conservationt a given scaling level are considered to obtain similarity

tier, or bottom-up scaling analysis, provides similarityspecic processes such as ow pattern transitions anddent heat transfer. The focus of the bottom-up scalingto develop similarity criteria to scale individual pro-


cesses of importance to system behavior as identied in the stepabove. Closure equations, typically addressing local phenomena,are considered to obtain similarity criteria.

The result from performing the top-down and the bottom-upanalyses ishierarchy abullet abov1998, time1990, as onless groupssystem andarchy, thusThis isdonefrequency bmass, momThe latter qprocess to o

2.2.3.2. FSAal., 2005; Caprevious pasystem thervirtual linkNamely, theofmechanisorigin of a nare heavily

The FSAof three papet al. (2005)list of referwork documin the papeto other autorder to appplied (lumthe reporteof Change (ered: the fo(or agent otion of itsthe rankingfor scaling tfor scaling tpressure: ndifferent tranot appearFRC and theratio have

The FSAcomputatioanalysis ofet al. (2005However, threduce theaccording ttitative hiergenerated Pthe concernof view, specapabilitiesinfrastructu2007; Ayde

The dirpresent framena scaling

scaling. More specic connections are:

- the use of balance equations in the 1st tier of the H2TS and inSA, is consistent with the design criteria in Table 1;ision by Zuber about code applicability anduser impact uponesults contributes to provide additional reasons for the scal-ontroversy mentioned in Chapter 1, thus supporting thetive

strandent papmpu-users of

The sCSApionscie

d. Sixinge) wed todisc

wordmpoCSAUiremn uncis beire eitem

senincl

1,ed in

er-1the cle-uphe scee scato d

envivideder-1ct ofiffers toecesproc. It idatasentlies,7)

er-2achck. . .stata large number of scaling ratios. In order to create themong thoseparameters, e.g. the objectives at the seconde, time ratios are introduced (according to Zuber et al.,ratios were proposed in previous years, e.g. by Moody,e outcome of the methods for establishing dimension-). The numerical values of time ratios for an assignedan assigned process are used for establishing the hier-for identifying the most inuential scaling parameters.by consideringeach timeratio as theproductof specicy the residence time. The former quantity denes theentum or energy transfer rate for a particular process.uantity denes the total time available for the transferccur within the control volume.

(primarily, Zuber, 2001; Zuber et al., 2005, 2007; Wulff ettton et al., 2005). The H2TS procedure discussed in theragraph connects the scalingmethod and the (complex)mal-hydraulic phenomena. Thismaybeused to create awith the (complex) structure of related computer codes.bottom-up approach of the 2nd tier,may need the useticmodelswhose implementation in the codes is at theumber of degrees of freedom for the code-users thatcriticized e.g. by Zuber (2001) and reported above.for system thermal-hydraulics was established in a seters by Zuber et al. (2005),Wulff et al. (2005), and Catton, presented in the same conference (Avignon, 2005, seeences). However the origin of FSA is attributed to theented in Zuber (1991) by Zuber (2001); furthermore,

r by Zuber et al. (2005), the authors make referencehors who introduced the words fractional analysis. Inly the FSA method, balance equations are needed: sim-ped parameter model) balance equations are used ind application. Key parameters named Fractional RatesFRC) and Fractional Change Metric (FCM) are consid-rmer quanties the intensity of each transfer processf change), which changes a state variable by a frac-

total change during an assigned transient and allowsof the agents of change; the latter is representative

he fractional change of a state variable. The FCM is usedhe fractional change of a system state variable such asoticeably, state variable trajectories representative ofnsients reported versus FCM show similarities that dowhen the same variables are reported versus time. TheFCM have the role that specic frequency and time

inside the H2TS, respectively.procedure appears as an attempt to get rid of complexnal tools and to create a different framework for thecomplex scenarios, even though, according to Wulff) . . . FSA is a scaling analysis, not a prediction tool.e combination of FSA and scaled experiments can sharplyneed for computer code calculations.. Rather, againo Wulff et al. (2005), FSA can be seen as . . .the quan-archy that supersedes the widely used, but subjectivelyhenomena Identication and Ranking Table.. Denitely,ed attempt is highly desirable from a scientic pointcically in case of succeeding in duplicating the same(e.g. by system codes) by an independent analyticalre, as demonstrated by recent applications (e.g. Wan,mir, 2009).ect impact of the Zuber scientic activity upon theework, other than thedeepunderstanding of phenom-processes, is constituted by the importance given to

the F- the v

the ring cobjec

Theindepepresencate coby endaudito

2.2.4.The

tutes ato themethon varyvolumdevoteBefore

Thethe i

Therequup aThisrequ(this

Keybelow,papersreport

(a) Papingscato tcodhowarepro

(b) PapEffeat dtionis nupNPPthepreappitem

(c) Papof echelastfor the present effort.

ightforward scaling analysis based on open and user-t code application, e.g. the procedure proposed in theer, may help in tackling the aversion toward sophisti-tational tools, as well as, toward the results obtainedrs of the codes including independent assessors andcalculations.

caling requirements in CSAUU (Code Scaling Applicability and Uncertainty) consti-eering effortmadebyUSNRCaimedatmaking availablentic and technology society a suitable uncertaintypapers (eachpaper ishereafter recalledaspaper-nwith

from 1 to 6, connected with the page numbering in there published in J. NED, Special Issue, 1990 (pp. 1117,

CSAU) which provide the full description of the method.ussing the scaling requirements, two notes apply:

Scaling is part of the CSAU acronym: this testies ofrtance of the issue, according to US NRC.

shall be considered as a basket of (very) valuableents for performing an uncertainty study or for settingertainty method, rather than an uncertainty method.cause of the large number of steps of the method thatxpert judgment and are left to the user of the methodis not discussed any further in the present paper).

tences or paragraphs dealing with scaling are listeduding paper identication (e.g. 16: note that only2,3and4are consideredhereafter), originalwritingitalics and comments from the present authors.

, related to the step 6 of the CSAU aimed at determin-ode applicability: . . .whether the code . . . accuracy andcapabilities . . . are adequate tomodel processes importantnario. . .. It is requested to evaluatewhether adequatele-up capabilities exist, but no way is suggested abouto, neither acceptability threshold for a possible processsaged, see item 7) in Section 4.1 for a possible answerby UMAE.

, related to the step 10 of the CSAU titled DetermineScale: Differences for similar physical processes, but

ent scales, should also be quantied for bias and devia-establish a statement of potential scale-up effects . . . Itsary to evaluate the capability of the BE code to scale-esses from reduced scale test facilities to the full scales also necessary to quantify the effects of limitations onbase used to develop models and correlations that arein the code.. Same comment as in the previous itemexcept for the last sentence that is not dealt with byin section 4.1., related to step 5 of the CSAU: The scale-up capabilitycorrelation (. . .ideally the result of a line by line coding) should be part of the code documentation. Thus theement in italics of the previous item is reiterated.


(d) Paper-3: The (incomplete) table-list of experimentswhichmustbe at the basis of the qualication of a code includes a columndevoted to counterpart test.

(e) Paper-3, related to steps 8 to 10 of the CSAU: Parameters . . .such as Hmust hadata . . .TF [Testconservathe extesibility t

(f) Paper-5ing requwhere nuate wherange; [1ships; [1range. Ifto the resub-stepSecondlynegativeFinally,how togiven).

The impof CSAU, coNoticeably,suitable solthe presentattempt to

2.2.5. Scalinreactors

The pursubject, as avarious studto show theTo this aimthe rst tw(b) providedand Taiwaning also genAP-1000 an

(a) Russianoperateof this pgary, thewas alsand conpreserviSzabadoet al. (2counteret al. (20

(b) Far Eastand opein Taiwreducedbasicallyin the caBaek et

(c) Scalingleast thr

have been modied to perform a few or a number of experi-ments relevant to the new generation reactors. These includeSpes, Lstf, and Piper-one. Moreover, among ITF mentioned atitems (a) and (b), Pactel and Iist have been modied for simu-

ng p. Tuung Aa-APR is to

Yadig98),ted

Derivs secetened botbetwed inscal

ouldhy leas sfact

lingthorsed in

f (19rjeeet al.s andscaliet al.

w hule-inhe p, botfactre,itertiosant cchre, inccritireleaminin T

ginatimeistog in mion.furt

takenof dito t

TF (nDS=

tio of Ctio of Atio of ATC [Heat Transfer Coefcient], pump performance . . .ve their uncertainties from . . . SET [Separate Effect Test]The comparison must encompass data from a range ofFacility] scales, to facilitate extrapolation . . . Otherwisetive estimates . . .. Following this statement implies

nsive use of conservatism in input data, owing to impos-o fulll the (scaling) requirement.devoted to scaling, or to step10of theCSAU:The follow-irements that also constitute sub-steps (i.e. 10-n below, identies the sub-step) in themethod are: [10-5] Eval-ther the database [for code assessment] covers the NPP0-7] Evaluate the scale-up capability of closure relation-0-9] Evaluate whether the closure relations cover the NPPany of the above evaluations is negative, then add biassults. Firstly, it appears that sub-step [10-9] is part of[10-7] and sub-step [10-7] is part of sub-step [10-5]., it seems clear that the outcome of any evaluation isin the sense thatNPP ranges arenot covered, in general.

there is no systematic procedure or suggestion aboutdetermine the bias (even though some examples are

ortance given to the scaling issue, for the applicationnstitutes the nal remark from the short overview.keyproblemsassociatedwith scaling are envisaged, butution is not provided. Therefore, the nal results fromeffort, i.e. the Road-map to Scaling, can be taken as an

address the scaling issue in the CSAU.

g studies in selected countries and for advanced

pose here is not to present a State-of-the-Art on thelready mentioned, rather to connect key ndings fromies with the topics discussed in the present paper andbackground for building-up the Roadmap to Scaling.

, a review of scaling contributions is given below, ino cases focusing on ITF: (a) related to Russian reactors;

from Far-East Countries, noticeably from South Korea; (c) connected with the design of new reactors (includ-eration III plus, AP-600 [Advanced PWR] and, later on,d SBWR [Simplied Boiling Water Reactor]).

Reactors scaling. Three main ITF have been built andd (all of these are still in operation at the time of issueaper) to simulate the VVER reactors: the Pmk in Hun-Pactel in Finland and the Psb in Russia (the Isb facility

o constructed by the same Institution that designedstructed Psb). The power-to-volume, full-height, time-ng scaling is adopted in all cases, as reported bys et al. (2007), Tuunanen et al. (1998), and Melikhov003), respectively. Proper attention has been given topart testing (discussed in Section 3.5), see e.g. Szabados09) and Blinkov et al. (2005).Countries scaling activities. Twomain ITF have been builtrated in Far East Countries (excluding Japan): the Iistan and the Atlas in South Korea. The reduced-height,-pressure scaling and reduced-height, full-pressure,pursuing Ishii scaling criteria, see above, is adopted

se of the two ITF, as reported by Hsu et al. (1990), andal. (2005), respectively.activities connected with the design of new reactors. Atee ITF simulators of PWR or BWR, listed in Section 2.3,

lati(e.glatiRosSBWtionby(19rela

2.2.6.Thi

compltop anlishedreport

Thetors shhierarcsideredscaling

Scaing aureport

- Wulf- Bane- Ishii- Reye

Ishii- Yun

A fea samp

In t(1998)designthermoHochretime-radominand HoTable 1wherepower

Sumcriteriaare orithree-dthose dkeepinvalidat

As abelowvaluesITF andthree ITank; A

(1) Ra(2) Ra(3) Raassive phenomena relevant to new generation reactorsnanen et al., 2000; Chang et al., 2006). Thus, ITF simu-P-600 or AP-1000 include Spes and Lstf (also known as600), already mentioned, and Osu-Apex. ITF simulatingnclude Panda, Puma and Giraffe. Signicant contribu-scaling dealing with AP-600 and SBWR, are providedaroglu (1993), Ishii et al. (1998), Reyes and Hochreiter

Banerjee et al. (1998), and Cho et al. (2005), the last oneto a specic phenomenon.

ations of scaling factorstion has been built-up having in mind the purpose ofss for the present survey and aiming at counter posingtom level scaling factors. To this aim, a parallel is estab-een the design factors in Table 1 and the scaling factorsTable 2.ing factors which are at the origin of the design fac-be considered as primary scaling requirements (or topvel scaling factors). The remaining ones should be con-econdary scaling requirements (or low hierarchy levelors).studies considered in Table 2 are by due the follow-(the adopted scaling methodology and attribute areparentheses, if applicable):

96),et al. (1998),(1998) (Ishii scaling attribute),Hochreiter (1998), (H2TS Methodology by Zuber, and

ng attribute),(2004).

ndred scaling factors are obtained in those papers anddicative fraction of these is reported in the table.apers by Ishii et al. (1998) and Reyes and Hochreiterh scaling and design factors are presented: namely,ors are basically those presented in Section 2.2.2. Fur-an evaluation is made in the paper by Reyes and(1998), of so-called characteristic-for the dominant-processes: it is worth noting that allharacteristic time ratios (e.g. Table 3 of paper by Reyesiter, 1998) are satised by the design factors listed inluding the noticeable exceptions of those time ratioscal ow-rate (discussion in Section 3.1) or thermalse from structure appear.g-up, scaling distortions associated with the designable 1 and expected due to the scaling factors in Table 2,ed by TPCF, thermal power release from structures andnsional effects. Designers of ITF may act to minimizertions, without impacting other design parameters and

ind a key objective of the ITF construction that is code

her investigation of scaling data, lets consider the datafrom the paper by Banerjee et al. (1998). These are the

mensionless factors (relevant to scaling) related to threehe reference prototype. This is AP-600 for each of theomenclature for the following is: CMT=Core Make-upAutomatic Depressurization System).

Prototypevalue

Range of value inmodels

MT ow to vessel volume 0.82 0.771.21DS ow to vessel volume 0.96 0.982.00DS ow to CMT ow 1.18 0.812.58


The following remarks apply to the data above and to the infor-mation in Table 2.

(A) The usu(e.g. takcharactwere simprocesseOther, mare: diffto justifirrelevathe sent

(B) The desappliedfactorsdesign oconictthose at

(C) The ansamountno praing activdistortio

(D) The agrsentencwith thescenario

(E) Then ththe appamonging. Thicode saolds ofscaled mby testigeneticequatiocabilityto situations cohere conscalingcapabilifactors aanalyst.

The follofrom the patrative for tsimilitude, hof scaling rein which arrates and volength and produce conicting scaling requirements..rms, what are called here detailed scaling parame-(or introduce the potential to imply) conicting scalingts related to theparameters inTable1. Therefore scalingarried out during the early conceptual design of experi-ts usefulness be limited later to post-factum quanticationscale distortions.. Thus, at a time (even 1996) when

ve been built and operated and codes have been qual-as possible, the use of detailed scaling parameters isthe characterization of previously avoidable scaling

withinlink be

In gparts wset byinclud

Witbeyondthe prethe curUS NRments2.2.4, alicensi

2.3. Th

Sevbeen dDAuritify anITF (e.gmary opublishin Sect

Reldesignexampnomendocumcapabiexperithat mof NRS

It isevant OECD/ment,for the

TheincluLstf,Pipe

The oductof Biimpoby n

Arouandmen

Amo(dediffeing trepo

Lessumeevenis the(rougalready mentioned BEPU approach, brings to the directn Scaling and Licensing.ic terms the Licensing can be seen as formed by twomust reach suitable consistency: (a) the requirements

roper Authority, and, (b) the answer from the applicant,e support from the research side.entering the details of the licensing process that is

scope here, in simplied terms, it can be remarked thatpaper deals with a proposal for item (b). Furthermore,situation for item (a) is summarized by the documents05) and IAEA (2010). In both documents the require-y the CSAU in relation to scaling, discussed in Sectionsically translated into the legal-safety language of the

erimental facilities

tens experimental facilities characterized as ITF havened built and operated in the area of NRSTH (see e.g.01). Cooperative efforts have been completed to iden-racterize the research programs associated with thoseOECD/NEA, 1996; Addabbo et al., 2001). Valuable sum-

pertise from the operation of individual ITF have beenanumberof cases, seealso summary informationgiven.2.5.t information related to scaling can be derived from thethe operation of SETF (e.g. see OECD/NEA, 1993). Anthis is given in Section 3.2 related to the CCFL phe-ere it may be sufcient to notice that in the mentioned

more than 200 SETF suitable for demonstrating codehave been classied that are at the origin of about 2000al datasets. All over the world it can be easily estimatedhan 1000 SETF have been or are in operation in the area

ide the scope to summarize here even the key scale rel-gs and the interested reader can easily access the two

documents. However,making reference to the ITF docu-facts can bementioned tomake clearer the perspectiveent paper:

considered in the mentioned OECD/NEA documentsWR simulators like Semiscale, Loft, Lobi, Spes, Bethsy,ist, Umcp and Uptf and BWR simulators like Fist, Fix-II,, Rosa-III and Tlta.ll cost associated with all experimental programs con-imarily during the decades 19701990 is of the order-Dollars (value of the year 2000). This testies of thece given to the experimental scaled-evidence, namelyr safety regulators.000 ITF tests have been performed with data recordedronically stored, typically 1001000 signal per experi-h recording frequency of the order of 1Hz.ose experiments, counterpart tests and similar testsns provided in the reference document) performed iny scaled facilities till the year 1996, i.e. directly address-caling issue, are classied in the OECD/NEA (1996),e also Section 3.5 below.10% of the above tests have been analyzed, with doc-post-test studies, by system thermal-hydraulic codes,gh the key goal for the execution of those experimentslication of code-nodalizations. The cost of the analysesone man-year per experiment) constitutes a justica-


tion for not analyzing the experimental data, thus preventing thepossibility to improve the understanding in relation to the scalingissue.

It will be difcult to nd individual analysts (or code-users) whohave perfit will bedevelopedITF and pe

A concluexperimentthe viewpobiased by th

2.4. The com

Noticeabization (or iand theuserelevant inand, conseqas the thre

Relatedography exto synthesizlet items bean idea of tresults of eapoint for in

System coNRSTH, ations mosteam phtutive mocodes. Rement andfound inet al., 198The qualitutes a diOverviewmay alsoDAuria aand DAuresourcesresources

The numetion. Thethe use oresult in umeshes todecreasinthe unsta

The nodaconnectsthermal-hof the codization. Ththe so calbased onsible the chand requconditionhigh levetion that

control volume dimensions is mainly a consequence of the same[CV+ J] approach; the empirical equations implemented into thecode may also have a connection with the choice of control vol-ume dimensions. Thus, an optimal nodalization exists valid in

ipleil (e.gt in thvelothe oty oflizatccelllso bode-(seeestandormaespoexis

mpacQualing (suitawith

8), re5), folly a

ics r

scopossiovidedgeproacf sopreies, etive

tions

n-sc

mpred uyOEtoSEsentmentan eon,ter, bhe skITF i-shoCA sreasnot-

imeon dea shocalcuingd (aormed analyses of more than 10 experiments in ITF andvery difcult to nd analysts who have modeled (i.e.input decks suitable for code analyses) more than 5

rformed test analyses accordingly.

sive remark from the above bullet items is that theal data base from ITF is not fully exploited, namely fromint of scaling, or that common views about scaling aree lack of detailed knowledge of existing databases.

putational tools

ly, two computational tools, i.e. the code and the nodal-nputdeckor interfacebetween the codeand the reality),r (or codeuser, or nodalizationdeveloper, or analyst) areNRSTH and affect the quality of any calculated databaseuently, the scaling process. These can be identied heree components of a calculation.to each of the components of a calculation, wide bibli-ists and it is not within the scope of the present papere the state of the art on the subject. Rather, in four bul-low, key literature reports are recalled that may give

he features and of the importance upon the calculationch of the four components and do constitute a startingterested reader to enter into more details:

des are obviously at the center of the development ofs already mentioned in this paper. Six balance equa-del, i.e. mass, momentum and energy for liquid andases, with proper set of around one-hundred, consti-dels (and/or closure equations) are at the basis of thoselap5, Athet, Trac, Cathare are familiar names. Develop-description of the main features of those codes can bethe related manuals and even in textbook (e.g. Forge8), or international reports (e.g. OECD/NEA, 2004ac).cation of the system thermal-hydraulic codes consti-scipline by itself and the scaling issue is part of this.papers dealing with procedures for qualication (onend V and V, Verication and Validation) are those bynd Galassi (1998), DAuria et al. (2006a), and Petruzziria (2008). As a key remark here is the note that theneeded for code V and V can be even larger than theneeded for code development.rical solution method affect the result of any calcula-lack of convergence of some system codes comes fromf ill-posed non-hyperbolic systems of equations whichnstable problems. The equations require large enoughprevent instability through numerical diffusion. When

g the mesh size, numerical diffusion is decreased andble nature of equation is revealed (Ferreri et al., 1996).lization is the results of a brainstorming process thatthe reality, i.e. the NPP, with the code, i.e. the systemydraulics code. Suitable understandings of the NPP ande nature and structure are needed to develop a nodal-e nodalization in the concerned NRSTH code suffers of

led [CV+ J], i.e. Control Volume plus Junction approach,a modular approach that on the one hand makes it pos-onstruction of mountain by simple stones, on the otherires a suitable denition by the users of the interfaces among the stones. The [CV+ J] approach requires al of expertise by the code user. The empirical observa-results of a code calculation are affected by changes of

princdetamento debe ofplexinodaBonucan a

The coperthe bitiesthe fthe raturethe iUserTrainciesdealt(199(200be fu

3. Top

Thefar as pmay prknowling apwork ofor thestrategapplicaderiva

3.1. No

Acobe basparedbrelatedthe prein the

Asnomenhereafsider tBethsyor snapa SB-LO

Thefrom alated tsituatital datnot be

Duraffectefor each code application, such that an increase in the. in the number of nodes) does not bring to improve-e results (i.e. better simulation of the reality). The effort

p a nodalization for a water cooled nuclear reactor mayrder of a fewman-years thus giving an idea of the com-such an endeavor. Thus, qualication is needed for theion to a similar extent of what needed for a code (e.g.i et al., 1993; Aprile et al., 1996); suitable informatione found in the J. STNI Special Issue, 2007.user (or the analyst) is typically the nodalization devel-the denition given above) and is responsible to set-upnodalization taking into account of the code capabil-deciencies. The analyst should be seen, not only asl responsible for the result of a calculation, but also asnsible of the quality of the same results. A wide liter-ts in relation to the User-Effect (UE), i.e. documentingt of users upon the results (UE), the Requirements forication (RUQ) aiming at reducing the UE, and the UserUT), dealing with the recommended list of competen-ble for an analyst. The issues of UE, RUQ and UT areby Aksan et al. (1993), Ashley et al. (1998) and DAuria

spectively, see alsoAksan et al. (2000) andPetruzzi et al.r supporting information. Obviously, the analyst mustware of the scaling issue.

elevant to scaling

e for the present chapter is to put together objective (asble) facts and ndings that are relevant to scaling ande a support or may justify, together with the standardsynthesized in theprevious chapters, theproposed scal-h (Chapter 4). Mentioning or giving emphasis to theme authors in the previous chapters, does not implysent authors, neither endorsing the concerned scaling.g. H2TS or FSA in the case of Zuber, neither the relatedresults, e.g. Kh


.

vapor pull tcavitation.system parinto scalabnozzle diamrepresentatthat are relthe connecters and coITF and at dthe connect

The situcomplex ifmixture levlower plenutarget TPCFof the ITF onomena are

At thisreproduciblaffect the TPnario. Thuscase) can beshould be c

The onlis the valiboundary cthis conneccode-user iparametersbe avoided.

3.2. Phenom

As a commay add thsee the CSNmatrices (Odent. For inbecause ophenomenadependent.

geogatioe con

demn-comt leg(e.gat, 1is maagesationn. Asicald ochanthe lisafetme

e of L., 197Fig. 1. Scaling of TPCF

hrough and sub-cooled vapor formation by sharp edgeThe last phenomena, at their time, are affectedby (local)ameters that do not appear into balance equations orle mechanistic models, like: (a) ratio between breaketer and vessel diameter, (b) geometrical parametersive of the conguration of internals in lower plenumeasing thermal power to the uid, (c) rounding edge attion between break nozzle and vessel. Those parame-nnected phenomena are not reproducible in differentifferent scales (i.e. simply because the rounding edge ationbetweenbreaknozzle andvessel is notmeasurable).ation during the concerned snap-shot is even moreone considers heat losses primarily in the exit nozzle,el formation and uid temperature stratication in them of the facility. All those phenomena also affect thephenomenon and cannot be controlled by the designerr by the designer of the experiment. Again, those phe-not reproducible in different ITF and at different scales.point it is straightforward to conclude that non-e parameters and non-reproducible phenomenaCFphenomenonwhich affects the overall SB-LOCA sce-

Theinvestiples ar

Thedowof hosionsKarwtionpassformregiois baliquifueland

TheECCSphaset alneither TPCF nor the overall scenario (SB-LOCA in thisscaled-up or extrapolated. Rather, the concerned issue

onsidered when performing uncertainty analysis.y use of experimental information like what discusseddation of computational tools where all relevantonditions are modeled. An important role is taken intion by the user of the code and by the exibility of thenterface, where tuning (i.e. unjustied change of inputwith the goal of matching the experimental data) shall

ena and accuracy scaling

plement to the introductory remarks in Section 3.1, oneat, in principle, all thermal-hydraulic phenomena (e.g.,I [Committee on the Safety of Nuclear Installations]ECD/NEA, 1993, 1996)) are geometry scaling depen-stance, two-phase pressure drop depend on diameterw regimes depend on diameter. Then, a number ofdepending on pressure drops are geometry scaling

scale exporigin of tments alldesign of

Thendinural circufunctiontion of thof steam

As a conare, in geneof data from

Let us cocalculate pof the modone of the pe.g. the ratitative of thThe possibimetry scale dependence has been the subject of directn in relation to selected phenomena. Noticeable exam-stituted by:

onstration that liquid penetration (i.e. down-ow) iner, or across the upper core plate namely in the caseinjection, strongly depends upon the system dimen-

. Wolfert, 2008 and CCFL occurs, see also Glaeser and993). In case of small scale experiments liquid penetra-de difcult by the uid interaction with wall in narrowand by the high steam speed. In this situations levelmay occur in the upper region with voided bottom

t a large scale (e.g. UPTF, see below) this phenomenonly prevented by three-dimensional effects: downwardw occurs preferably in selected zones (e.g. low powernels in the case of core), level formation does not occursquid directly contributes to the core cooling.y issue raised because of inadequate effectiveness ofasured in small scale facilities during the blow-downOCA, caused by the ECC bypass phenomenon (Beckner9, see also Sudo, 1984). In this case, Semiscale smalleriments performed at beginning of 1970s were at thehe issue. Later on, the availability of large scale experi-owed thedemonstrationof the suitability for the currentECCS.g that natural circulation performance, namely the nat-lation mass ow-rate in the core region reported as aof the mass inventory of the primary system, is a func-e number of U-tubes (then of the volume scaling ratio)generators (DAuria et al., 1991).

clusion from the above, thermal-hydraulic phenomenaral, geometry scale dependent. Thus no extrapolationsmall scale experiments is acceptable.

nsider now the possibility to scale-up the capability tohenomena. In this case we can consider as a measureeling capability (or capability to calculate phenomena),ossible denitions of accuracy or error in the prediction,o Yexp/Ycalc. Here, Y is the generic parameter represen-e thermal-hydraulic phenomenon under investigation.lity to scale-up the capability to calculate phenomena


is considered in the works by DAuria and Karwat (DAuria et al.,1988, see also below). The evaluation of such possibility is con-sidered by Belsito et al. (1994), Bovalini et al. (1993), and by anumber of papers dealing with the bases or the development of theUMAE and C1995; DAuracronyms aracy is not afor the appFurther eviupon scalinFFTBM (Fasal., 1990). TSB-LOCA Co(1997b). Adcan be founThe accuracculation res

Deniteproper analnon-dimenferentwordcalculationevidence wdeciency rconstitute ato be contin

3.3. The expaccident

A milestbility to rescenario rein a NPP. ThrespectivelyLstf is the stime-preserscaling. TheNPP is close

The usetest SB-SG-Watanabe (sured similduration ofthe time ofin the Lstfwevent takesization of thopening cycare also mesponding tidetected ordecreasing

The resulation of daunderstand(and the dethe transpoimental simthe scaling

3.4. The sca

Any disfacilities lik

operated to gather experimental data that have been used todemonstrate the quality level of computational tools.

LoftITF (50Mof a

ne slooptor aof thavereaccleardedWS (, 198ary aed watedr rearangece recialscaleraturo ontively2.

eralblicsystarizebovebetwmainentam ralyseet

the. MoCD/Nndreby Dqualncluheprses t

cod wits (alictio

test (of o

tion,llingthees, dictinsafetthe

ertainertainIAU uncertainty methodologies (DAuria and Pellicoro,ia andGiannotti, 2000), respectively (see below, for thend for more details). The demonstration that the accu-scale dependent parameter constitutes a prerequisite

licability of the concerned uncertainty methodologies.dence of the same nding, e.g. accuracy independentg, is obtained by adopting the accuracy denition of thet Fourier Transform Based Method) (e.g. Ambrosini ethis evidence can be found, for the specic case of theunterpart Test (see below) in the paper by DAuria et al.ditional information connected with accuracy scalingd hereafter in relation to Fig. 3 and related discussion.y is the measure of the discrepancy between code cal-ults and (relevant) experimental data.ly, accuracy does not depend upon scaling and, thusytical denitions of accuracy shall be considered as asional, scale-independent statistical parameters. In dif-s, onemay conclude that there is no demonstration thataccuracy is connected with scaling. In case a differentould appear, this should be considered as a modelingequiring code improvement. This conclusion does notprinciple neither a theorem; rather it is an observationuously validated against the experience.

erimental simulation of the Mihama NPP SGTR

one in scaling is constituted by the demonstrated capa-produce with an experiment the complex transientsulting from a SGTR (Steam Generator Tube Rupture)e involved NPP and ITF are the Mihama-2 unit and Lstf,. The Mihama-2 unit is a 1475 MWt, 2-loop PWR. Thecaled model of a PWR according to power-to-volume,ving, full-pressure, full-height (see also Section 3.6)volume scaling ratio between Lstf and the Mihama-2to 1/21.of fundamental scaling criteria in the design of the06 and the analysis of experimental data, Hirano and1992), demonstrated qualitative andquantitative mea-arity between the model and the prototype for a timemore than one hour after the transient start. Namely,pressurizer emptying and start of rell were measuredithin a difference of a few seconds, although the secondplace nearly 3000 s after the transient start. Depressur-e intact steamgenerator and of the primary system andles of the relief valve in the affected steam generatorsasured in the experiment and strictly match the corre-me trends of the NPP. No dry-out condition was eitherexpected in any of the two measured scenarios, thus

the complexity level of the simulation.lts discussed above should not authorize any extrapo-ta from any ITF to any NPP. However, they conrm theing by the scientic community that the scaling lawssign factors) discussed in Section 2.2.1 are suitable forsition of phenomena between NPP and ITF. The exper-ulation of the Mihama-2 transient constitutes one ofbridges introduced in Section 3.6.

ling role of Loft and Uptf

cussion about scaling cannot forget the facts thate Loft and Uptf have been designed, constructed and

3.4.1.Loft

aroundsectioncore. Ointactgenerabreaksmentsof theby nuble Enthe ATDivineboundequipp

Relnucleain thereferenthe ofpowertempeequal trespecaround

Sevthe pution ofsummof the amentsuffer,documcriticis

AnaCerulloing of(1998)the OEeral hupaperdozenand coupon taddres

Systemcisionimplie- pred

thetimevenful

- thatvalupredthe

- thatuncuncOECD/NEA, 1991), is actually a Nuclear Plant producingwth power. Six fuel bundles having the same cross-ctual nuclear fuel and reduced height constitute theteam generator and one coolant pump are part of the while (roughly) pressure drop simulators of the steamnd of the pump are part of the broken loop, wherehe primary circuit are located. A few tens of experi-been performed during the around 20 years of the life

tor, where the word experiments can be substitutedaccidents. Noticeably, in this list there are the Dou-

Guillotine Break LOCA (further discussion below) andAnticipated Transient Without Scram) (e.g. Bayless and2). Both these two experiments are unique in terms ofnd initial conditions that cannot be achieved inside ITFith electrically heated rod simulators.to scale, Loft can be considered either a 1:1 (small)ctor or an ITF, having volume and power scaling ratio1/15 to 1/50 (roughly) depending from the size of theactor, i.e. either a 2-loop or a 4-loopNPP. This last one isreferenceprototypeplant, thus the ofcial volumeandis 1/50. Design-scaling parameters related to pressure,e, time, rod-surface- temperature and linear power aree (e.g. fullling conditions at rows 13, 12, 9, 14 and 10,, in Table 1). However, length is reduced for a factor

hundreds or even thousands of papers and reports inliterature deal with Loft experiments and the applica-em codes. It is outside the scope of the present effort torelated ndings. Moreover, all (or the large majority)papers and reports conclude about satisfactory agree-

een measured and calculated trends and unavoidablyly because of size, of non-adequate or non-traceable

tion. The related results and ndings are prone to theised in Section 2.2.3.s of Loft experiments are documented in the paper byal. (1985), including the LB-LOCA test L2-5. Revisit-same experiment was performed by Bedrossian et al.re recently the L2-5 test was selected at the basis ofEA BEMUSE Project (e.g. Petruzzi et al., 2004). The sev-d pages report by Petruzzi and DAuria (2005), and thee Crecy et al. (2006), describe calculation results by aied groups of code users. A wide variety of ndingssions from the mentioned researches have an impactesentwork.However, the followingkey-remarkdirectlyhe strategy pursued here:

es are able to predict Loft LB-LOCA with adequate pre-hout any adjustment or tuning. Adequate precisionl together):nof all relevantphenomenaandeventmeasuredduringcritical heat ux, reood, time of pressurizer emptying,ccurrence of peak clad temperature, accumulator inter-etc.)with suitablequalitative andquantitative accuracy,acceptability criteria;errors or differences between measured and calculatedo not impact the safety limits: for instance the error ing rod surface temperature does not cause over-passingy threshold when added to the predicted value;experimental data are bounded by the calculation ofty, without adjustments or tuning when applying thety methods.


The remark from the analysis of Loft experiments can be thesame as from the analysis of any ITF experiments, however, theintrinsic value can be higher due to the dimensions of the facility:errors in predictions exist and shall be bounded by uncertainty. Ascorollary stis meaninglcode in relabeing consiapplicabilit

3.4.2. Uptf,The Upt

1993)was c4500Mwt Nlegs, pressupumps arehowever,

- power isamount o

- the steamfrom the p

- themain cprototype

- the vessel

As in theexperimentcode quali2008, or, resame concluhere. Howeso far makin

Experimare compleand the simcase of Cctfis relevant a

No claimSctf and Ccany NPP. Rfacility, shain reproducshall be eva

3.5. The cou

The idetogether wnote that thplan that ain operationand Loft, exconditionspart or simiperformedand similaras already mand counte

Becauseexperimentinitial and tof the consiity was perf5 ITF and 7

The involved ITF are Lobi (simulator of German four loop PWRby two un-equal loops), Spes (simulator of US three loops PWR bythree equal loops), Bethsy (simulator of French three loop PWR bythree equal loops), Lstf (simulator of US four loops PWR by two

oopsr equs stental tconleg,. Sel

rrenccumm) clatioitionreduado

ountsuredthe oondinracyd ind

dless areor ma anCherse, eDAurparfewtabaporteass intem

ee leof t

dataperimta caan i

ttommentwherthree useexpeiagra) mecaparentlyseA andan awitatte

ng thtestSB-

eri eio, bye alsectioatements: (1) any best estimate computer calculationess if not supported by uncertainty evaluation; (2) thetion towhich the above conclusion applies is eligible fordered within a suitable scaling process to demonstratey in licensing.

Sctf and Cctff (Damerell and Simons, 1992; Damerell and Simons,onstructedwith the internals of thevessel of a (roughly)PP. Other than the vessel, four hot legs and four coldrizer and proper simulators of steam generator andpart of the facility. It is denitely a full scale facility,

produced by a nearby steam plant that feeds suitablef steam in the core region,generator simulator remove energy by removing massrimary system,oolant pumps suitably reproduce pressure drops of thepumps,design pressure allows operation at 2MPa.

case of Loft, a large number of papers deals with thes performed in the Uptf. Those works address eithercation (e.g. DAuria et al., 1999b; Glantz and Freitas,markably, the scaling, e.g. Takeuchi et al., 1998). Thesive remarks reported in the case of Loft are applicable

ver, no deep international activity has been performedg use of Uptf data.ents performed in Sctf and Cctf (e.g. Murao et al., 1993)mentary to Uptf considering the scale of the facilitiesulated region of the vessel, i.e. the core region in theand Sctf. The same conclusion/nding reported for Uptfs far as the present framework is concerned.shall be made that experimental data from Loft, Uptf,

tf researches can be used to determine the behavior ofather those data, like data measured in any relevantll be analyzed by codes and the capability of those codesing the experiments with suited-quantied accuracyluated.

nterpart testing and the similar tests

a to perform experiments at different scales cameith the decision to build the rst ITF: it is enough toe rst ITF even built was named Semiscale, under thelarger IT

scaling in nuclear reactor system thermal-hydraulics

Documents