validation of coupled codes using vver plant measurements

Nuclear Engineering and Design 235 (2005) 507–519

Validation of coupled codes using VVER plant measurements

T. Vanttolaa,∗, A. Hamalainena, S. Kliemb, Y. Kozmenkovb, F.-P. Weissb,A. Kereszturic, J. Hadekd, C. Strmenskye, S. Stefanovaf, A. Kuching,

P. Hlbockyh, D. Sikoi, S. Danilinj

a VTT Processes, Nuclear Energy, P.O. Box 1604, FI-02044 VTT, Finlandb Forschungszentrum Rossendorf (FZR), Germany

c KFKI Atomic Energy Research Institute AEKI, Hungaryd Nuclear Research Institute Rez plc (NRI), Czech Republic

e VUJE Trnava a.s., Slovakiaf Institute of Nuclear Research and Nuclear Energy (INRNE), Bulgaria

g Scientific and Technical Centre on Nuclear and Radiation Safety (SSTC NRS), Ukraineh NPP Bohunice, Slovakiai NPP Mochovce, Slovakia

j Kurchatov Institute (KI), Russia

Received 29 March 2004; received in revised form 5 June 2004; accepted 31 August 2004

Abstract

n kinet-i 40 typea ight in-s hydraulicc 8. Theg of coolantm e differencesb onsidera-t y of datac lso taughtt©

f parte on

0

A data set of five transients at different VVER type nuclear power plants was collected in order to validate neutrocs/thermal hydraulics codes. Two of these transients ‘drop of control rod at nominal power at Bohunice-3’ of VVER-4nd ‘coast-down of 1 from 3 working MCPs at Kozloduy-6’ of VVER-1000 type, were then utilised for code validation. Etitutes contributed to the validation with 10 calculations using 5 different combinations of coupled codes. The thermalodes were ATHLET, SMABRE and RELAP5 and the neutron kinetic codes DYN3D, HEXTRAN, KIKO3D and BIPReneral behaviour of both the transients was quite well calculated with all the codes. Even an elementary modellingixing in reactor pressure vessel under asymmetric transients improved correspondence to the measurements. Sometween the calculations seem to indicate that fuel modelling and treatment of VVER-440 control rods need further c

ion. The simultaneous validation interacted with the data collection effort and thus improved its quality. The complexitollection systems and sometimes conflicting data, however, called for compromises and interpretation guides that ahe analysts balanced plant modelling.

2004 Elsevier B.V. All rights reserved.

∗ Corresponding author. Tel.: +358 9 456 5020;ax: +358 9 456 5000.

E-mail address:[email protected] (T. Vanttola).

1. Introduction

Best estimate modelling of NPP behaviour isof modern safety analysis. In order to add relianc

029-5493/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.nucengdes.2004.08.047

508 T. Vanttola et al. / Nuclear Engineering and Design 235 (2005) 507–519

Nomenclature

CR control rodFA fuel assemblyMCP main circulation pumpNPP nuclear power plantSPND self powered neutron detectorRPV reactor pressure vesselVVER pressurized water reactor designed in

Russia (water/water energetic reactor)

safety analysis codes various approaches are needed,such as numerical benchmarks and comparing to sep-arate effects tests or integral experiments on specificfacilities. The credibility of the codes may further beimproved if real NPP data is available for testing. Us-ing of such data is, however, challenging because plantoperation is complex and the available data is not inall respects optimal for validation. The efforts in thecoupled neutron kinetics/thermal hydraulics code val-idation concerning VVER reactors have already beengoing on for a number of years using numerical bench-marks. The occurred VVER plant transients suitablefor code validation were for the first time scanned inthe previous PHARE project SRR1/95 (Mittag et al.,2001; Hamalainen et al., 2002). In one of three part of

the VALCO project, introduced here, this task is ex-tended to new transient types.

In the beginning of the VALCO project data wascollected from five transients, three concerning VVER-440 plants and two of VVER-1000 type. One VVER-440 case and one VVER-1000 case were then chosenfor validation: ‘drop of control rod at nominal power atBohunice-3’ for VVER-440 reactors and ‘coast-downof 1 from 3 working MCPs at Kozloduy-6’ for VVER-1000 reactors. The former is an unexpected event fo-cusing on core power and RPV mixing phenomena,whereas the latter is part of start up tests and empha-sizes loop thermal hydraulics.Table 1summarises thecollected transients in the two projects.

Eight institutes participated in code validation withfive different coupled codes. Six teams applied ATH-LET as thermal-hydraulic code and five teams DYN3Das neutronics code. The combination of ATHLET andDYN3D was applied by four teams. The participants,codes and calculated transients are summarized inTable 2.

In the validation each participant had to interpretthe given plant data, which may add discrepancies inthe comparison of the results. Hence some recommen-dations for data interpretation were needed. A furtherlimitation emerged from the fact that each participantapplied a ready made model, where the level of de-tail varied. Some of these issues are discussed in thepaper.

Table 1The transients used for data collection in the VALCO and SRR1/95 pr

Plant type PRI/SECa

1. Transients reported in SRR1/95VVER-440 3 Drop of one turbine to 1997

1 Shutdown of 3 from 6 986

VVER-1000 4 Turn-off of one from t 19933 Decrease of the turbo-

load level at Zaporosh1 Switch-off of two neig 992

1. Transients reported in VALCOVVER-440 2 Drop of control rod no 9

1 Outage of three main 01 Shutdown of 2 from 6

V ump 3off (1 ocontrol 001

turbine

by reactor tripVER-1000 1 Main coolant p

MCP 1 switch-2 Several single

a 1 = Main coolant pump trips, 2 = control rod movement, 3 =

ojects

Year

house load level experiment at Loviisa-1working main coolant pumps at Dukovany-2 1

wo working SG feed water pumps at Balakovo-4generator power from 1000 MW down to the houseye NPP

1996

hbouring main coolant pumps at Kozloduy-6 1

. 287 at 100%Nnom at Bohunice-3 199coolant pumps atN= 95.4%Nnom at Mochovce-1 200working MCPs at 100% at Dukovany-2 followed 1997

switch-off (1 of 4) at 100%Nnom followed withf 3) at 65%Nnom at Kozloduy-6

1992

rod insertions/withdrawals at Rivne-3 2

trips and 4 = feed water pump trips.

T. Vanttola et al. / Nuclear Engineering and Design 235 (2005) 507–519 509

Table 2Participants, used coupled codes and calculated cases

Organisation Country: Coupled code VVER-440 Bohunice VVER 1000 Kozloduy

FZR Germany ATHLET-DYN3D (Teschendorff et al., 1996;Grundmann et al., 1995)

+

VTT Finland HEXTRAN-SMABRE (Kyrki-Rajamaki, 1995;Miettinen and Hamalainen, 2000)

+ +

KFKI Hungary ATHLET-KIKO3D (Hegyi et al., 1998) +NRI Czech Republic ATHLET-DYN3D +INRNE Bulgaria ATHLET-DYN3D +VUJE Slovakia RELAP5-DYN3D (Strmensky, 2001) +SSTC NRS Ukraine ATHLET-DYN3D +KI Russia ATHLET-BIPR8 (Lizorkin et al., 1992) + +

2. Bohunice transient

2.1. Transient description

In the Bohunice unit 3, control rod no. 287 fromgroup 2 dropped during normal full power operation6.1.1999. The power at first decreased to 89%Nnom.The protection system prevented full power recoveryby blocking control group withdraval. The operatorthen reduced the power to 85%Nnom, where all theparameters were stabilised. These first 1000 s after roddrop were included in the code validation.

The external ionization chamber recordings showedthat the power distribution was remarkably skewed.This also reflected to marked variation of the hot legtemperatures, the fuel assembly outlet temperaturesand self powered neutron detector (SPND) signals.The observed phenomena enabled model evaluationof reactivity effects of rod movements and consequentpower redistribution calculations. The changes in thehot leg temperatures also allowed evaluation of mixingprocess in the upper plenum.

2.2. Calculation specification

The calculated transient starts with control rod dropduring the first 12 s,Fig. 1. The initial position ofthe regulating group was 175 cm from the bottom ofthe core. The operator gradually lowered the powerl g-u t re-p wasr and3 ri-

ods that seemed to fit best with the data. The deter-mination of the absolute movement was left for theanalysis teams. Also the modelling of flow by-passroutes had to be agreed upon to enable fuel assem-bly outlet temperature comparisons. No recommen-dations were given for the secondary side modelling,because the level of detail varied a lot between theteams.

2.3. Used codes and assumptions

The Bohunice transient was calculated in five in-stitutes with five different code couplings (Table 2).Table 3summarizes the codes, the coupling types andthe basic models. Parallel coupling means that corethermal hydraulics and fuel heat transfer is calculatedboth in the neutronics and the thermal hydraulics codes.Internal coupling means that the thermal hydraulicscode solution is directly applied in the neutronics code,and external coupling that the core thermal hydraulicsand heat transfer are solved solely in the neutronicscode. All teams modelled all the fuel assemblies sepa-rately, but the amount of thermal hydraulics channels inthe core region varied. Most of the teams also modelledall the six circulating loops individually.

The asymmetric transient also appears in the loopbehaviour, and hence mixing in the lower and upperplena calls for attention. VTT, KFKI and KI appliedspecific mixing models for these volumes, while NRIand VUJE assumed perfect mixing. In the core mixingi em-b

0a epa-r ina-

evel from 89 to 85% by moving downward the relating group, but the needed movement was noorted. For the calculations the regulating groupecommended to be inserted in two slopes (1/4/4 of the total movement) within fixed time pe

s very limited due to the shrouds around the asslies.

Four codes, CASMO-4, HELIOS, KARATE-44nd KASSETA were used for cross section data pration from the nuclear data libraries. The comb


Fig. 1. Layout of the fuel assemblies and essential control rods in the Bohunice core. The approximate thermal hydraulic sectors correspondingto the loops and positions of assemblies chosen for data comparison are shown.

Table 3Basic modelling in the Bohunice calculations

VTT VUJE NRI KFKI KI

Neutron kinetic code HEXTRAN DYN3D DYN3D KIKO3D BIPR8Thermal hydraulic code SMABRE RELAP5 ATHLET ATHLET ATHLETType of coupling of codes Parallel Parallel External Internal InternalNo. of core flow channels in thermal hydraulic code 6 6 – 25 349No. of core flow channels in neutron-kinetic code 349 349 349 – –No. of assemblies in neutron kinetic code 349 349 349 349 349Total core bypass flow in active core (%) 3.64 9.112 3.67 9.12 9.0Mixing of adjacent loop sectors in RPV before/after

core (%)30/0 Perfect across RPV Perfect across RPV 25/0 100/<100

Number of circulation loops 6 6 2 6 6


Table 4Initial state data in Bohunice calculations

Measured VTT VUJE NRI KFKI KI

Neutron kinetic code HEXTRAN DYN3D DYN3D KIKO3D BIPR8Cross section calculations CASMO-4 HELIOS HELIOS KARATE-440 KASSETANeutron power (MW) 1375 1375 1379 1381 1378Total mass flow (kg/s) 8814 8843 8814 8889 8962 8810Core inlet mass flow (kg/s) 8513 8011 8573 8145 8260Average cold leg temperature (◦C) 270.6 270.4 270.4 270.8 270.0 270.5Average hot leg temperature (◦C) 300.4 299.9 299.9 300.2 299.1 300.1Doppler coefficient (pcm/K)a −2.67 −2.64 −2.48 −2.83 (−3.28) −2.54Moderator total temperature coefficient (pcm/K)a −39.1 −53.01 −43.38 −41.8 −41.24Boron concentration (g/kg) 3.0 3.0 3.0 2.72 2.96 2.60

a Uniform temperature change.

tions appear inTable 4, where the main parameters inthe initial states are compared, too.

2.4. Bohunice results

The rod drop is immediately recorded in the outof core ionization chambers close to the dropped rod(Fig. 2). The plant data is presented as stepwise signals,where the reading is updated only when the change islarge enough from the previous value. In the modellingVTT applied specific response kernels for the ex coreneutron detectors, while the rest simulated the signalwith fast neutron flux in the fuel nodes closest to thedetector location. All the simulations of the close de-tectors at first go below the recorded relative power 0.7,but then recover on some higher level due to feedbackeffects. The other detectors record only the later control

F d thes ient.

action and set close to 0.9 (Fig. 2). The overall asymme-try of the core at the end of the calculation is illustratedin Fig. 3. It appears that the relative calculated radialpower profiles vary considerably close to the droppedrod (±10%), which may also be seen in the simulatedexternal detector signals (Fig. 2). This may imply thatmodelling of the VVER-440 type control rods in nodalcodes still needs attention.

The codes predict power levels from 90 to 95% af-ter the initial drop (Fig. 4). The KFKI, VUJE and NRIcalculations stay on a higher level than the VTT and KIcalculations. The variation can be explained by differ-ences in feedback models such as fuel Doppler effectand worth of the dropped rod. In the KFKI and VUJEcalculations the core inlet temperatures decrease moreand thus increase reactivity in the core compared toother calculations (Fig. 5). The plant data is not veryexplicit, because on the other hand 89% was reportedafter the drop, while e.g. evaluating from loop temper-atures 93–95% could be deduced as well (Fig. 6).

The final power level 85% was then obtained bymoving the regulating group downwards. In the simu-lations the regulation movement varied considerably,ranging from 12.4 to 29.5 cm (Table 5). The mainreason for the differences is the varying intermediatepower level after the rod drop, but other modelling pa-rameters also contribute, such as the reactivity effectsof fuel temperature, coolant inlet temperature and theworth of the regulating group.

The kernel model of the ionization chambers soft-e fastflc all-e the

ig. 2. Three measured ex core ionization chamber signals animulations near and far from the dropped rod, Bohunice trans

ns the simulated signal compared to the simplerux of the closest fuel node (Figs. 2 and 3). In the VTTalculation the power after the rod drop is the smst and the radial power tilt one of the largest, but


Fig. 3. The calculated relative radial power profile (Pfinal/Pinitial ) in the indicated assembly row crossing the dropped rod position and onecorresponding total core map (NRI calculation), Bohunice transient.

simulated signal of the closest detector EP1 is in themiddle of the other calculations.

The measured hot leg temperatures start to divergejust after the rod drop due to asymmetric power gen-eration (Fig. 7). In the VTT and KFKI calculations theupper plenum is modelled without mixing and in thelower plenum only with partial mixing (Table 3). Inthese results the hot leg temperatures in the two loopsnear the dropped rod decrease 7◦C more than in theopposite loops, which is in agreement with the obser-vation. In the KI result the divergence is 4◦C, whilein the other calculations with perfect mixing the phe-nomenon is naturally lost.

Fig. 4. Calculated neutron power, Bohunice transient.

The plant signals from the in core neutron detectorsand the thermocouples are damped because of physi-cal inertia or filtering of the data. In order to make thecalculations comparable with the measurements sim-ilar filtering was needed, as was already done in theprevious project (Hamalainen et al., 2002). The sametime constants, 10 s for loop temperatures and 30 s forfuel assembly outlet temperatures also seemed to fit tothe Bohunice case. On the other hand, to make reason-able predictions the two time constants of the in coreRhodium neutron detectors had to be increased artifi-cially.

Fig. 5. Calculated and measured core average inlet temperature, Bo-hunice transient.


Fig. 6. Calculated and measured average temperature difference be-tween hot and cold legs, Bohunice transient.

Two fuel assembly outlet temperatures are shown inFigs. 8 and 9. The measured temperature decrease closeto the dropped rod, 17◦C, is a bit larger than the sim-ulations. The KFKI and VTT results are closest to themeasurements. The temperature drop due to the latercontrol action is somewhat larger in the calculationsthan in the measurements. An obvious explanation isthat the calculated power did not decrease enough dur-ing the first phase.

In the Bohunice core there are Rhodium typeSPNDs at seven elevations in 36 fuel assemblies. Thesignals of eight assemblies were compared to the mea-surements (three of them shown inFig. 1). The signals

Fig. 7. Calculated and measured temperatures in six hot legs, Bohu-nice transient.

decrease almost uniformly near the dropped rod at allelevations, by 0.55 in relative units. On the other sideof the core the rod drop is not seen, but only the latercontrol action. The general behaviour of the calculatedand filtered SPND signals is quite well in accordancewith the measurements (Fig. 10). It may be noticedthat in spite of total power decrease some calculationsactually indicate slight local power increase in thelower part of the core on the opposite side. The increasecan be explained by added reactivity due to coldercore inlet flow in those calculations. The detectorrecordings were too insensitive to prove or disprove thephenomenon.

Table 5Control rod and fuel rod modelling in Bohunice calculations

VTT VUJE NRI KFKI KI

Total control rod movement (cm) 12.4 21.0 29.5 18.0 16.0Worth of dropped rod in hot zero power

critical state,T= 260◦C, no Xe (%)0.1337 0.1143 0.12646 0.165 0.1312

Critical boron concentration in hot zeropower,T= 260◦C, no Xe (g/kg)

6.19 7.00 6.67 5.93a

Fuel average temperature (◦C)Initial state 527 679 616 593Final state 488 622 571 537

Maximum fuel centreline temperatureInitial state (◦C) 925 1228 1217 950 1129Final state (◦C) 888 1246 1189 858

Used gas gap model, constant (W/m2 K),dependencies on temperature or burnup

Temperatureand burnup

Constant 5600 Temperature Constant 3000 Constant 3000

a Control rod position 200 cm.

and burnup


Fig. 8. Calculated and measured fuel assembly outlet temperature inposition 06–45, Bohunice transient.

Fig. 9. Calculated and measured fuel assembly outlet temperature inposition 20–43, Bohunice transient.

3. Kozloduy transient

An experiment of two successive pump trips wasconducted in the start-up tests of the Kozloduy NPP unit6 of VVER-1000/V-320 type in 1992. In the VALCOproject the latter trip was chosen for code validation,but the whole transient was documented in the datacollection phase.

3.1. Transient description

In the first phase MCP no. 3 was switched off at fullpower, after which the automatic reactor power regu-lator decreased power to about 65% and the flow in

Fig. 10. Measured and calculated SPND signals at elevation201.5 cm in the Bohunice case. All signals are scaled to 1 in thebeginning of the transient. TOP: position 6–41 beside dropped rod,MIDDLE: position 7–48 beside regulating group and BOTTOM:position 15–56.


Table 6Basic modelling in the Kozloduy calculations

VTT KI FZR SSTCN INRNE

Neutron kinetic code HEXTRAN BIPR8KN DYN3D DYN3D DYN3DThermal hydraulic code SMABRE ATHLET ATHLET ATHLET ATHLETType of coupling of codes Parallel Internal External External ExternalUsed symmetry in core 60◦ 360◦ 60◦ 60◦ 60◦No. of core fluid channels in thermal hydraulic code 4 163 – – –No. of core fluid channels in neutron kinetic code 28 – 28 28 28No. of assemblies in neutron kinetic code 28 163 28 28 28Total core bypass flow of active part of core (%) 3.0 4.04 3.0 3.0 3.0No. of circulation loops 4 4 4 4 4

the tripped loop reversed. In the second phase, 90 minlater, MCP no. 1 was tripped, after which also this loopreversed. The regulator reduced reactor power furtherto 51.5% by first moving the control rod group no. 10in and half a minute later out. The primary pressurewas regulated by the feed and bleed system and expe-rienced a temporary rise of maximum 0.25 MPa during40–160 s, when flow reversed in the tripped loop. Thepressurizer spray valve opened twice. The steam headerpressure decreased by max 0.07 MPa during 20–80 sbut was also recovered. The turbine controller unloadedthe turbine from 565 to 415 MW within 60 s. The timedependent core data was limited, but the initial and fi-nal states of the transients were documented. The plant

functions and measurements were documented exten-sively and sometimes, in case of conflicting informa-tion, additional data evaluation effort was needed.

3.2. Used codes and assumptions

The Kozloduy transient was calculated in five insti-tutes, VTT, KI, FZK, SSTCN and INRNE with threedifferent coupled codes. The used codes and the basicmodels are summarised inTable 6and the reactor datain Table 7. All the code coupling types were in use.This kind of transient needs modelling of all the circu-lation loops, but symmetry of the core may be applied.Only KI used the whole core modelling. The interme-

Table 7Reactor data in the Kozloduy calculations, the second phase

VTT KI FZR SSTCN INRNE

Neutron kinetic code HEXTRAN BIP8 DYN3D DYN3D DYN3DCross section calculations CASMO-4 KASSETA NESSEL NESSEL NESSELInitial neutron power (MW) 1947 1945 1947 1948 1931Initial xenon Non-stationary Non-stationary Stationarya Non-stationary Stationarya

Initial boron concentration (g/kg) 3.15 2.60 2.89 2.63 2.98Doppler coefficient at initial state

(pcm/K)b−2.55 −1.74 −2.50 −2.74

Moderator total temperature coefficientat initial state (pcm/K)b

−24.1 −34.2 −27.5 −27.3

Fuel average temperature (◦C)At initial state 514 506 566 633At final state 480 470 518 571

M ◦836

U Consta

aximum fuel center line temperature (C)At initial state 756At final state 701

sed gas gap model: constant htc ordependencies on

Temperaturedependent

a Xenon imbalance compensated by boron change.b Uniform temperature change.

829 960 960749 863 864

nt Temperaturedependent

Temperaturedependent

Constant


Fig. 11. Calculated and measured axial position of control rod group10, Kozloduy transient.

diate time of 90 min is not quite enough to stabilize thewhole primary system, but such an assumption maybe used in the calculations, except for the core xenoncontent.

3.3. Kozloduy results

For power regulation FZR, SSTCN and INRNE usedmeasured CR positions directly as boundary condition,whereas VTT simplified it slightly and KI modelledthe controller itself. The measured and calculated con-trol rod group positions appear inFig. 11. The KI re-sult differs by 3–4% units from the data in the laterphase of the transient. The core power is measured onlya few times (Fig. 12). Generally the calculations fol-

F tran-s

low the real behaviour, even though there is some un-derestimation during the stabilization phase. The dif-ferences between the measurements and the calcula-tions are largest before the control group withdrawal at40 s.

The main primary loop parameters are shown in theinitial state inTable 8. In the circulating loop figurestypically three curves are compared: the value of theloop number 3 with initially reversed flow, loop 1 withthe stopping MCP and the average of the loops 2 and 4with still operating MCPs.

In the simulation of flow behaviour it is importantto model flow friction in different parts of the pri-mary circuit properly, and to use as correct homolo-gous pump curves as possible. The data included mea-sured pressure differences during normal operation andsome characteristic pump coast down curves, but plantspecific pump curves were not available. Some gen-erally used models in VVER-1000 applications could,however, be obtained. Based on this material, it wasthe responsibility of each team to prepare the model.The measured and calculated pressure differences overthe MCPs are compared inFig. 13. It may be noticedthat the general behaviour during the transient is wellreproduced, but some deviations appear, such as thehigher than measured pressure differences in the run-ning pumps of SSTCN and VTT or the slightly deviat-ing time behaviour of the FZR pumps. Also the calcu-lated core mass flows and loop flows, that are derived

F in loop1 theb oduyt

ig. 12. Calculated and measured neutron power, Kozloduyient.

ig. 13. Calculated and measured pressure increase in MCPs(pump trip, flow reverses), loop 3 (pump off, reversed flow fromeginning) and average of loops 2 and 4 (pumps running), Kozl

ransient.


Table 8Measured and calculated initial data and main coolant pump parameters, Kozloduy transient

Measured VTT KI FZR SSTCN INRNE

Neutron power (MW) 1949 1947 1945 1947 1948 1931

Cold leg temperature (◦C)Average in loops 1, 2 and 4 286.5 285.5 285.9 284.8 285.4 282.6Loop 3 with stopped pump 284.4 285.5 285.5 284.8 285.1 282.8

Hot leg temperature (◦C)Average in loops 1, 2 and 4 310.7 307.8 310.9 307.8 308.7 303.9Loop 3 with stopped pump 277.0 276.4 276.6 279.2 275.9 277.9

Core inlet mass flow (kg/s) 13640 13572 12252. 13024 12237 14670

Mass flow rate (kg/s)Average in loops 1, 2 and 4 5210.7 5170.4 4813.7 5015.5 4847.1 5490.3Loop 3 with stopped pump −1569.5 −1511.4 −1651.5 −1602.9 −1712.0 −1515.4

Pressure increase (kPa)Average in MCP 1, 2 and 4 469.8 511.6 451.9 473.4 528.7 510.0MCP 3 162.8 159.5 152.6 148.1 144.8 209.1

Flow reversal time (s) in loop 1 42.0 43.2 50.0 38.6 48.0MCP 1 stopping time (s) 115 116 110 108

from the pressure differences, may deviate as much as10% from the evaluated data (Table 8). The timing offlow reversal varies between 38.6 s of SSTCN and 50 sof FZR.

The cold leg temperature is controlled by the sec-ondary temperature until the flow is reversed. The re-versal brings about temperature minimum in the coldleg, the depth of which varied from 5 to 10◦C in thecalculations (Fig. 14). In the beginning most of the cal-

Fig. 14. Calculated and measured cold leg temperatures in loops 1and 3 and average of loops 2 and 4, Kozloduy transient.

culations were in accordance with the measured coldleg temperature and at the end all except KI are about2◦C below the measurements. The INRNE calcula-tion started from a lower level mainly because of thelarger total flow. The hot leg temperature of the af-fected loop experiences a stronger and permanent de-crease (Fig. 15). In these data comparisons uniformtime constants were applied for the calculated signals.The measured hot leg 1 temperature is probably dis-

Fig. 15. Calculated and measured hot leg temperatures in loops 1and 3 and average of loops 2 and 4, Kozloduy transient.


turbed by loop wall temperature as is indicated in thefigure.

In the evaluation of upper plenum pressure andpressurizer level, it turned out that all the calculatedpressures are more sensitive than the measurement tochanges in the primary loop just after MCP trip, whichoften appears in one-dimensional modelling of multi-dimensional phenomena.

The level of detail in the secondary side modellingvaried also it the Kozloduy transient. As an example,FZR, SSTCN and INRNE used the measured steamheader pressure as a boundary condition. Several sec-ondary side parameters were included in the data com-parison to check proper boundary conditions for theprimary side, that was in the focus of this study. Mostof the parameters were in reasonable agreement withthe observations.

The axial and radial power distribution and the coreoutlet temperature distributions were compared in theinitial and final state. The simulation of xenon transientbetween the two pump stops by VTT and SSTCN leadto more downwards peaked power profiles than in theother calculations. In the FZR and INRNE initial statesthe xenon change is compensated with increased boricacid concentration. The applied boron concentrationvaried between the calculations because of unambigu-ity in the plant data (Table 7).

The initial and final radial power distributions werealso compared to plant data. In the initial state the calcu-lations differed from the measurements on an averageb atet gestd

tem-a enceb eemst ni sientw risei thel heb Them byt Thed iallyc e no-t thet

Fig. 16. Difference of measured core outlet temperatures to calcu-lated as a function of calculated relative bundle power in the begin-ning of the Kozloduy transient.

4. Summary and conclusions

In the VALCO project data was collected about fiveVVER specific nuclear power plant transients. Twoof them, ‘drop of control rod at nominal power atBohunice-3’ for VVER-440 reactors and ‘coast-downof 1 from 3 working MCPs at Kozloduy-6’ for VVER-1000 reactor were used for coupled code validation.Eight institutes participated in the validation with fivedifferent coupled codes.

In the Bohunice case the most essential part of val-idation is the core behaviour during the rod drop and3 min later starting control actions to reach 85% powerlevel. The general behaviour during the whole transientwas quite well calculated with all the codes: both thepower distribution changes and the fuel assembly out-let temperatures were mostly well reproduced. Therewere, however, notable differences in the first powerdecrease, in the axial power profile, in the calculatedcontrol rod worth and in the fuel temperatures. The dif-ferences in the required control rod group movement toreach the final power were large. This may imply thatthe control rods of VVER-440 and fuel models needfurther attention.

The measurements of the individual assembly outlettemperatures and the hot leg temperatures indicatedthat in the transient the coolant mixing in the upperplenum was weak. It could also be demonstrated withthe codes that included a relevant mixing model and adetailed enough core channel description.

y more than 5% (KI result 3.4%). In the final sthe results are better. In all the calculations the larifferences are in the middle of the core.

The calculated core outlet temperatures are systically higher than the measured ones. The differetween the calculations and the measurements s

o depend on bundle power (Fig. 16). The phenomenos even more pronounced at the end of the tranith smaller mass flow and larger temperature

n the core. This could possibly be explained byocation of the thermocouple in the mid line of tundle below the conical part of the bundle head.easurement could be at least partly disturbed

he colder water coming from the central tube.ifference to the measurements could be artificompensated with higher core mass flow, as may biced in the INRNE points, but this does not removerend.


The features that make the Kozloduy transient inter-esting, such as lowered power and flow reversals in theloops, also proved to be difficult both for data collec-tion and for modelling. As an example, it was very hardto find detailed enough data about pump characteristicsand control logics. Anyway, the general behaviour ofthe Kozloduy second pump trip was calculated satisfac-torily with all the codes. In the comparison of the coreoutlet temperatures, a linear dependency was found be-tween the assembly power and the difference betweenmeasured and the calculated temperatures. The depen-dency could possibly be explained by temperature ef-fect of the bundle central tube flow.

Also in the Kozloduy case the initial fuel tempera-tures and the temperature changes during the transientvary remarkably. This supports the conclusion of theprevious SRR1/95 project that more accurate fuel mod-els are needed in the codes.

The comparison between the codes and validationagainst the measurements was successful and the re-sults were reasonably accurate. The task emphasizedcareful plant data interpretation and balanced plantmodelling, where transient specific asymmetric phe-nomena were in key role.

Acknowledgements

The Commission of the European Community(CEC) is acknowledged for funding the work in theV

R

G ther-del

DYN3D. In: Proceedings of International Conference on Math-ematics and Computations, Reactor Physics, and EnvironmentalAnalyses, Portland, USA, p. 257.

Hamalainen, A., Kyrki-Rajamaki, R., Mittag, S., Kliem, S.,Weiss, F.P., Langenbuch, S., Danilin, S., Hadek, J., Hegyi,G., 2002. Validation of coupled neutron kinetic/thermal-hydraulic codes. Part 2. Analysis of a VVER-440 transient(Loviisa-1). Ann. Nucl. Energy 29, 215–321, ISSN: 0306-4549.

Hegyi, Gy., Kereszturi, A., Trosztel, I., Langenbuch, S., Horche, W.,Velkov, K., 1998. Improvement of plant transient analysis forVVER by coupling KIKO3D with ATHLET. In: Proceedings ofICONE-6, San Diego, CA, USA.

Kyrki-Rajamaki, R., 1995. Three-dimensional reactor dynamicscode for VVER type nuclear reactors. Dr. Tech. Thesis, Espoo:Technical Research Centre of Finland, 51 pp. + app. 80 pp. VTTPublications 246.

Lizorkin, M.P., Semenov, V.N., Ionov, V.S., Lebedev, V.I., 1992.Time dependent spatial neutron kinetic algorithm for BIPR8 andits verification. In: Proceedings of the Second Symposium ofAER, Paks, Hungary, pp. 389–407.

Miettinen, J., Hamalainen, A., 2000. Development and valida-tion of the fast running thermohydraulic model SMABREfor simulator purposes. In: Proceedings of ICONE-8: EighthInternational Conference on Nuclear Engineering, Baltimore,USA, 2–6 April 2000. New York American Society ofMechanical Engineers, p. 12, Paper ICONE8-8188, ISBN0791819922.

Mittag, R., Kliem, S., Weiss, F.P., Kyrki-Rajamaki, R., Hamalainen,A., Langenbuch, S., Danilin, S., Hadek, J., Hegyi, G., Kuchin,A., Panayotov, D., 2001. Validation of coupled neutronkinetic/thermal-hydraulic codes. Part 1. Analysis of a VVER-1000 transient (Balakovo-4). Ann. Nucl. Energy 28, 857–873,ISSN: 0306-4549.

Strmensky, C., 2001. The continuing progress of RELAP5/DYN3DtingMay

T dol-of thehopuire-

ALCO project within the EU FP5 programme.

eferences

rundmann, U., Lucas, D., Rohde, U., 1995. Coupling of themohydraulic code ATHLET with the neutron kinetic core mo

coupled code development. In: AER Working Group D Meeon VVER reactor safety analysis, Moscow, Russia, 21–232001.

eschendorff, V., Austregesilo, H., Lerchl, G., 1996. Methoogy status and plans for development and assessmentcode ATHLET. In: Proceedings of the OECD/CSNI Workson Transient Thermal-Hydraulic and Neutronic Codes Reqments, Annapolis, USA, pp. 112–128.

validation of coupled codes using vver plant measurements

Documents