Annals of Nuclear Energy 120 (2018) 286–295

Contents lists available at ScienceDirect

Annals of Nuclear Energy

journal homepage: www.elsevier .com/locate /anucene

Neutronic experiments with fluorine rich compounds at LR-0 reactorq

https://doi.org/10.1016/j.anucene.2018.05.0560306-4549/� 2018 Published by Elsevier Ltd.

q This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US governmentretains and the publisher, by accepting the article for publication, acknowledgesthat the US government retains a nonexclusive, paid-up, irrevocable, worldwidelicense to publish or reproduce the published form of this manuscript, or allowothers to do so, for US government purposes. DOE will provide public access tothese results of federally sponsored research in accordance with the DOE PublicAccess Plan (http://energy.gov/downloads/doe-public-access-plan).⇑ Corresponding author.

E-mail address: evzen.losa@cvrez.cz (E. Losa).

E. Losa a,⇑, M. Koštál a, T. Czakoj a, B. Jánsky a, E. Novák a, V. Rypar a, J.J. Powers b, N.R. Brown b,c,D. Mueller b

aResearch Centre Rez, 250 68 Husinec-Rez 130, Czech RepublicbOak Ridge National Laboratory, PO Box 2008, Oak Ridge, TN 37831-6172, USAc The Pennsylvania State University, 229 Reber Building, University Park, PA 16802, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 7 December 2017Received in revised form 26 May 2018Accepted 31 May 2018

Keywords:LR-0 reactorFLIBETEFLONCritical experimentMSRFluorineSpectrum measurement

Research on molten salt reactor (MSR) neutronics continues in Research Centre Rez (Czech Republic) withexperimental work being conducted using fluoride salt that was originally used in the Molten Salt ReactorExperiment (MSRE). Previous results identified significant variations in the neutron spectrum measuredin LiF-NaF salt. These variations could originate from the fluorine description in current nuclear data sets.Subsequent experiments were performed to try to confirm this phenomenon. Therefore, another fluorine-rich compound, Teflon, was used for testing. Critical experiments showed slight discrepancies in C/E-1 forboth compounds, Teflon and FLIBE, and systematic overestimation of criticality was observed in calcula-tions. Different nuclear data libraries were used for data set testing. For Teflon, the overestimation ishigher when using JENDL-3.3, JENDL-4, and RUSFOND-2010 libraries, all three of which share the sameinelastic-to-elastic scattering cross section ratio. Calculations using other libraries (ENDF/B-VII.1, ENDF/B-VII.0, JEFF-3.2, JEFF-3.1, and CENDL-3.1) tend to be closer to the experimental value. Neutron spectrummeasurement in both substances revealed structure similar to that seen in previous measurements usingLiF-NaF salt, which indicates that the neutron spectrum seems to be strongly shaped by fluorine.Discrepancies between experimental and calculational results seem to be larger in the neutron energyrange of 100–1300 keV than in higher energies. In the case of neutron spectrum calculation, none ofthe tested libraries gives overall better results than the others.

� 2018 Published by Elsevier Ltd.

1. Introduction

Results presented in this work are a part of GEN IV reactorresearch carried out in Research Centre Rez. Since 1996, reactorphysics experiments performed at LR-0 reactor have supportedresearch of reactors with coolant and/or fuel in the form of fluoridesalts (fluoride high-temperature reactors [FHRs] and molten saltreactors [MSRs]). Discrepancies between criticality calculationsand experiments were discovered during re-evaluations of EROSexperiments (Losa et al., 2015) in cases with core configurationscontaining LiF-NaF salt (40 mol% LiF and 60 mol% NaF with natural

Li, referred to as FLINA hereinafter) and a combination of FLINA andgraphite. Therefore, some of these experiments were repeated toconfirm the results. The elemental parts of graphite and fluorinatedsalts were investigated using integral experiments during 2014and 2015. Thorough analysis showed that the discrepanciesbetween calculation and experiments in cases with graphite inser-tions were within 1 r uncertainty interval in terms of criticality(Koštál et al., 2016). Similar results were also obtained for experi-ments with FLINA insertion. Experimental uncertainties werereduced by repeating experiments many times and by having awell-characterized core. These assurances enabled acceptance ofthe experiments as a benchmark into the IRPhEP database in 2017.

Critical experiments provide an integral measure to assessparameters of interest, so even if the comparison of calculationand measurement of this parameter gives correct results, some dis-crepancies might eventually be present in neutron spectrum ratesor reaction rates. Effects of the fast neutron spectrum in the empty(void) experimental channel were studied and analyzed previouslyby Koštál (Koštál et al., 2015). For an experiment using a graphiteinsertion, it was found that discrepancies in calculation results

E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295 287

divided by experimental results (C/E) did not exceed 7% in allenergy groups. This value was better than the result obtained whenthe measurement was taken in the void channel with no materialinsertion, where the discrepancies were as high as 13%. More pro-nounced discrepancies are apparent from measurement in FLINAsalt, where the C/E – 1 comparison can differ by 40% for specificenergy regions. Influence of the FLINA salt on criticality was stud-ied by Losa (Losa et al., 2015), with the conclusion that the 6Li(n,t)reaction data can have a significant reactivity impact. Additionally,description of 19F(n,elastic) reaction could have some impact.

Fluorine-rich compounds were selected for use in further studyof the nuclear properties of these substances through neutronspectrum measurement and analysis. Teflon contains a substantialamount of fluorine bound in CF2 molecules, is readily available, andis compatible with the LR-0 experimental water-moderated reac-tor in terms of nuclear safety. Based on well-known carbon proper-ties in the LR-0 reactor in the form of graphite (Koštál et al., 2016;Koštál et al., 2015), it was thought that potential discrepanciesusing Teflon insertions could be due to the properties of fluorinebound in CF2 molecules. Based on the analysis of this simple com-pound, more complex materials with fluorine can be studied. Flu-orinated salt LiF-BeF2 (FLIBE) was originally used in the coolantcircuit of the Molten Salt Reactor Experiment (MSRE) operated atthe Oak Ridge National Laboratory (ORNL) in the United States.FLIBE is a sample of a real material intended for use in advancedreactors and was obtained (US DOE, 2013A) within the frameworkof a cooperation between Czech Republic and the United States,covered by a memorandum of understanding signed in 2011 (USDOE, 2013B).

2. Methodology of experiments and calculations

The influence of fluorinated materials on reactivity has beenassessed in a relative manner by comparing the experimentallydetermined critical states and corresponding calculations (C/E –1). Critical experiments were carried out in the benchmark coreof the LR-0 reactor (see Fig. 1) (Koštál et al., 2017), and calculations

Fig. 1. Overhead view inside the LR-0 reactor special core (left) and radial plot of texperimental position (large dry channel).

were performed in MCNP6.1 code (Goorley, 2012) using differentnuclear data libraries. The cross section sensitivity analysis wasperformed using TSUNAMI from the SCALE 6.2 software package(Rearden and Jessee, 2016).

2.1. Experimental setup – criticality

Neutronic experiments were performed in the zero power lightwater research reactor LR-0, with benchmark core composition(see Fig. 1). The benchmark core for insertion material experimentsconsists of six fuel assemblies, each with a nominal enrichment of3.3%. The fuel assemblies have similar construction to VVER-1000fuel except for a major difference in the fission column length(active length 126 cm in LR-0 vs. more than 350 cm in VVER-1000). They are arranged in a hexagonal lattice with a pitch of23.6 cm. A dry channel for material insertion experiments islocated in the core centre and has a hexagonal section fitting intothe lattice position.

In the shutdown state, the reactor does not contain any moder-ator. Criticality of the core containing the material insertion isreached by moderator addition, slowly pumping water into thereactor vessel. Differences in reactivity due to inserted materialsare thus reflected by differences in the critical moderator level indifferent experiments. Precise moderator level measurementenables criticality to be determined with an uncertainty on theorder of a few tens of pcm. All control rods are fully removed fromthe core during these benchmark experiments, so the control rodsare only used for reactor shutdown.

FLIBE insertion occurs using a solidified melt in a stainless steel(X5CrNi18) canister. Its shape is suited for neutron spectrum andfission rate measurement in the salt. The canister (Fig. 2) is filledwith 27.54 kg of salt with a density of 1.95 g/cm3. The salt containslithium that is highly depleted in 6Li. The height of the salt columnin the canister is 52 cm, so the salt occupies approximately 87% ofthe available volume in the canister.

The Teflon insertion has a cylindrical shape with a central cavityfor the detectors of the neutron spectrometric system. The cylinder

he core with specified enrichment for each assembly (right) and empty central

Fig. 2. Photographs of TEFLON (left) and FLIBE (right) insertions.

Table 1Dependence of keff on 6Li content (ENDF/B-VII.0) [13].

6Li content [ppm] keff std. dev.

0 1.00219 0.000025 1.00202 0.0000510 1.0021015 1.0020130 1.0020050 1.00184100 1.00158

288 E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295

is 64 cm high, with an outer diameter of 20.5 cm and an innerdiameter of 7.4 cm. The density of the Teflon is 2.19 g/cm3. Theentire cylinder is positioned at the aluminium base to fix thebeginning of the insertion at the same level as the beginning ofthe fuel fission column. No cladding is needed because Teflon isnontoxic and inert.

2.2. Experimental setup – neutron spectrometry

The fast (high energy) region of the neutron spectrumwas mea-sured by the method of recoiled protons in two energy ranges: 0.1–1.3 MeV with fine group structure (40 groups per decade), and 1.2–10 MeV with coarse structure (0.1 MeV–wide energy groups). Themeasurements in fine groups were taken with hydrogen propor-tional detectors, while a stilbene detector was used in coarsegroups.

2.2.1. Hydrogen proportional detectors (HPDs)Neutron spectra in the energy interval of 0.1–1.3 MeV were

measured using an analogue spectrometer (multichannel pulseheight analyser) and nuclear electronics for detector signal pro-cessing. The systemwith two spherical proportional detectors withdifferent hydrogen gas filling 400 kPa (energy interval 0.1–0.8MeV) and 1000 kPa (energy interval 0.2–1.3 MeV) was used. Thespherical, angle-independent detector manufactured in IBJ Swierk(Otwock, Poland) was used. Deconvolution of measured recoil pro-ton spectra was performed using the differentiation method withtheoretical and experimental corrections of response functions.The shape of the input neutron spectrum above the upper energythreshold of the proportional detector, measured using the Stil-bene detector system, was used to correct the higher energy neu-trons effect. More information about the methodology can befound in literature from Jansky and Novak (Jansky and Novak,2014).

2.2.2. Stilbene detectorNeutron spectra in the 0.8–10 MeV energy range were mea-

sured with a Stilbene scintillator (10 � 10 mm) with neutron andgamma pulse shape discrimination. The two-parameter spectro-metric system FD-111 (Veškrna et al., 2014) is fully digitized andcapable of processing up to 100,000 impulses per second in fullenergy range. The input signal from the photomultiplier wasdivided into two branches in the amplifier. Each branch was indi-

vidually amplified and digitized by separate analog-to-digital con-verters. This different amplification increased the dynamic range ofparticle energies so that the spectrometer could process andincrease the signal-to-noise ratio. Two fast analog-to-digital con-verters working on the sampling frequency of 1 GHz were used,and the digital signal processing was implemented in a field-programmable gate array. Therefore, all data flow could be pro-cessed from both analog-to-digital converters without any deadtime. Pulse shape discrimination was realized by the integrationmethod, which compared the area limited by a trailing edge ofthe measured response with the area limited by the wholeresponse. Deconvolution of recoiled proton spectra was performedusing maximum likelihood estimation (Cvachovec et al., 2008).

2.3. Calculations – MCNP

For criticality and neutron spectrum calculations, an LR-0 modelhas been analyzed using MCNP6.1 with data from various nuclearlibraries (ENDF/B-VII.1, ENDF/B-VII.0, JEFF-3.2, JEFF-3.1, JENDL-3.3,JENDL-4, RUSFOND-2010, CENDL-3.1). The older versions oflibraries (ENDF/B-VII.0 and JEFF-3.1) are used for comparison witholder data and data from benchmarks. Different data libraries wereused only for definition of the material insertion; the definition offuel, moderator, and structural materials is fixed in ENDF/B-VII.0 tosuppress the other possible effects to criticality (e.g. from fuel) thatare not being investigated in this study. ENDF/B-VII.0 is approvedby the national regulator for use in performing licensing calcula-tions at LR-0. The free gas model was used for thermal neutronscattering treatment in case of FLIBE, TEFLON, and stainless steelcanister description, and the photo-neutron production is switchedoff in the physical model.

The isotopic composition of Li used in FLIBE salt was not veri-fied yet, so a parametric study investigating the influence of 6Licontent in the salt on system reactivity was performed as shownin Table 1. Literature about salts form in MSRE (Shaffer, 1971)states that the 6Li concentration should not be higher than 50ppm. For the purposes of parametric study, 100 ppm was takenas the limit. Otherwise, the 6Li content is neglected in the MCNPmodels, as it was shown that the possible reactivity influence inthis concrete experiment is small (Losa et al., 2017) (see Table 1).The possible reactivity impact is reflected in broadened uncer-tainty interval. This study gives also insight into the possible influ-ence of impurities in the salt, which can be effectively expressed interms of 6Li equivalent. Also the uncertainty in the LiF and BeF2ratio in the salt was studied and the related uncertainty appearsto be in order of 10 pcm per 1% change in this ratio.

The standard benchmark LR-0 model (Losa et al., 2016) wasused for calculations. Criticality calculations with 20,000 neutronsper generation and 100,050 (100,000 active and 50 inactive) gener-ations yielded results with a standard deviation of 0.00002 in keff.These settings were chosen to obtain a statistical uncertainty of10% or less in the energy group 9.9–10 MeV in order to aid the fastneutron spectrum evaluation. Experimental uncertainty estimationwas performed by the standard ICSBEP methodology (Dean, 2008)

Table 2Uncertainties in parameters used to express uncertainty in criticality.

Parameter Identification Mean Measured or Design Value Reported Uncertainty in Parameter 1r Uncertainty Used in Calculation

Clad outer diameter [cm] 0.915 0.005 0.0016Fuel assembly pitch [cm] 23.6 0.15 0.15Moderator height H [cm] Exp. value 0.05–0.08 0.05UO2 density [g/cm3] 10.371 0.0077 0.0093235U enrichment [wt.%] 3.3 0.01 0.01U impurities [wt.% of U] 0.223% 0.1115% 0.223%Moderator density [g/cm3] 0.9982 0.0004 0.0004234U content [wt.% of U] NA 0.0309% 0.0309%

E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295 289

using parameter perturbations. Parameters used to express uncer-tainty in criticality can be found in Table 2. More details aboutthese parameters can be found in LR(0)-VVER-RESR-003 CRITbenchmark evaluation. The largest factors contributing to the totaluncertainty are directly connected with fuel definition: 234U con-tent, lattice pitch and fuel cladding thickness. 234U was found asthe largest contributor to the total uncertainty. The exact contentof 234U is not known for the LR-0 fuel and thus it was estimatedfrom the knowledge of composition of other fuel type from thesame producer. The effects of assumed uncertainties in materialinsertions are mostly marginal. It is assumed that the uncertaintieshave uniform distribution and are independent. Total uncertaintyis thus expressed by the ‘‘square root of sum of squares” rule(see Table 3).

2.4. Calculations – SCALE

Sensitivity and uncertainty (S/U) calculations were performedusing SCALE/TSUNAMI-3D/TSUNAMI-IP and TSAR (Rearden andJessee, 2016; Perfetti et al., 2016) with ENDF/B-VII.1 continuousenergy (CE) cross sections (Chadwick et al., 2011). Simplifiedtwo-dimensional (2D) and detailed three-dimensional (3D) modelsof the LR-0 reactor were developed in SCALE. The lower plate andgrid spacers were homogenized, but all other key core features (e.g.fuel pin geometry and water moderator height) were modeledexplicitly. The detailed 3D models agree well with draft bench-marks intended for the International Reactor Physics ExperimentEvaluation (IRPhE) Project handbook (Koštál et al., 2017) and otherLR-0 references (Losa et al., 2015; Koštál et al., 2016). S/U analysesperformed for a configuration that models September 2016 LR-0experiments using MSRE FLIBE salt identified the top contributorsto eigenvalue bias due to nuclear data, with a specific focus on salt-specific contributions, including fluorine.

Table 3Estimation of experimental uncertainties in pcm.

Parameter Configuration

Teflon FLIBE Canister

Fuel cladding 99 32 82density 10 25 8enrichment 44 57 38234U content 128 65 110pitch 102 68 98

Moderator level 12 24 7contamination 37 13 39

Insertion density 12 221 22

outer diameter 16 x xinner diameter 17 x xsalt level x 4 xins. contamination x 38 xLiF-BeF2 ratio x 10 x

Total 202 128 177

1 Density of the FLIBE salt.2 Density of the stainless steel canister.

3. Results

3.1. Comparison of calculations and experiments: criticality

Critical levels of experiments with the Teflon cylinder and theFLIBE canister are summarized herein, and other relevant bench-mark results are presented in Table 4. Different critical levels indi-cate the influence of inserted materials on the neutron balance inthe reactor core. Even though the FLIBE salt with Li depleted in6Li isotope acts as a moderating material, the critical level remainsrather high. The stainless steel casing of the canister is responsiblefor the high level of moderator in the case with FLIBE salt. The crit-ical level of the core with an empty canister of the same construc-tion was measured, as well. In Case 15 (Koštál et al., 2017), thebenchmark experiment with FLINA salt using natural Li requireda critical moderator level that was significantly higher than inother experiments due to increased neutron absorption in 6Li.

Calculations of criticality with different nuclear data librariesare shown in Table 5 and are graphically interpreted with theresults of benchmark values of the following cases (shown inFig. 3) to compare the agreement with similar moderatingmaterials:

� empty channel (Case 1)� graphite insertion without the central part (Case 7)� graphite insertion including the central graphite part (Case 8)� FLINA insertion (Case 15)

Results of the experiment with Teflon insertion show accept-able agreement with calculations in the benchmark model for allnuclear libraries used in this work. The calculation using ENDF/B-VII.0 reproduces the result of void channel (Case 1) extremely welland is therefore considered as a reference. Slight systematic dis-crepancies (overestimation) can be seen when using JENDL-3.3,JENDL-4, and RUSFOND-2010 libraries. These libraries use differentratios of inelastic and elastic scattering than ENDF/B-VII.0, JEFF-3.1,and CENDL-3.1 (Losa et al., 2015). Values in criticality are graphi-cally shown only for older versions of ENDF/B-VII and JEFF libraries(ENDF/B-VII.0 and JEFF-3.1), because differences between resultsfor newer versions (ENDF/B-VII.1 and JEFF-3.2) are only statistical(stochastic) in nature. The complete results can be found in Table 5.

Table 4Critical moderator levels in different LR-0 experiments.

Case Crit. level [cm] Unc. [cm]

Case 1 55.60 0.05Case 7 44.38Case 8 43.22Case 15 80.26EROS 3 49.36Teflon 46.42Canister for FLIBE (empty) 58.96FLIBE 54.40

Table 5Comparison of calculations (C/E – 1) in different data libraries with experiment [pcm].

Case 1 Case 7 Case 8 Case 15 EROS3 Teflon FLIBE Canister

Unc. 192 180 178 153 155 202 128 177ENDF-VII.1 46 23 �132 �128 63 217ENDF-VII.0 41 13 �13 �161 �181 65 219 134JEFF-3-2 48 18 �139 �136 65 214JEFF-3.1 34 �6 �133 �187 65 217JENDL- 3.3 22 3 �153 �191 108 245JENDL- 4 �59 �22 �153 �190 105 239RUSFOND-2010 17 �2 �107 �171 108 241CENDL-3.1 56 34 �133 �129 63 146

Fig. 3. Scheme of benchmark cores and EROS 3 experiment.

Fig. 4. Comparison of the benchmark results with different benchmark insertions, TEFLON, and FLIBE (the blue line shows critical water level, and red lines represent 1runcertainty intervals).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

290 E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295

Larger variations of C/E-1 comparison can be observed whenusing the FLIBE salt provided by ORNL; systematic overestimationcan be observed using all data libraries. This overestimation is notexplained by the fact that the residual concentration of 6Li wasneglected in the MCNP model. A parametric study on 6Li content(Losa et al., 2017) shows that the configuration is not sensitive to6Li in ranges expected for the residual concentration in the salt.This insensitivity is caused by the stainless steel canister itself,which is already causing a significant amount of thermal absorp-tions. The same statement is valid for possible salt contaminants

causing absorptions in thermal spectrum. However, the real con-tent of 6Li in the salt is an important factor in the future assess-ment because this isotope largely influences the criticality ofthermal reactors through (n,t) reaction, which has a very high crosssection at thermal neutron energies. Precise isotopic definition ofFLIBE salt by means of mass spectroscopy methods is currentlyunder investigation.

Experiments with an empty canister show that the structure inthe reactor core introduces a discrepancy reactivity defect of 134pcm. Fluoride salt addition reduces reactivity by an additional

E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295 291

90 pcm. Assessment of libraries concerning the inelastic to elasticscattering ratio on fluorine shows that the JENDL and RUSFONDlibraries tend to overestimate reactivity for LR-0 experiments, asalso observed in the case of Teflon (see Fig. 4).

3.2. Comparison of calculations and experiments: neutron spectrum

Neutron spectrum measurements supplemented the criticalexperiments. The fast part of the neutron spectrum is not influ-enced by uncertain concentration of highly absorbing isotopesin thermal spectrum (6Li content). The neutron flux densities inFLIBE salt and Teflon are presented for energies between 0.1and 1.3 MeV measured by hydrogen proportional counters(Fig. 5) and for energies over 0.8 MeV measured by Stilbene(Fig. 7). The direct comparison between experimental and calcu-

0

0.5

1

1.5

2

2.5

3

3.5

4

0 0.2 0.4 0.6

Flu

x d

en

sity

[a

.u]

Energy [Me

Fig. 5. Neutron spectrum in fluoride salts FLIBE, FLINA and in TEF

-50%

-40%

-30%

-20%

-10%

0%

10%

20%

30%

40%

0 0.2 0.4 0.6

C/

E-1

Energy

Fig. 6. C/E-1 comparison for FLIBE and TEFLON attainab

lational values is realized by means of C/E-1 comparison; Fig. 6shows flux 0.1–1.3 MeV and Fig. 8 presents flux above 0.8 MeV.The most significant discrepancies can be found in region 0.1–0.5 MeV. Similar discrepancies were reported by previous workwhere the neutron spectrum in LiF-NaF was measured (Koštálet al., 2015) and the discrepancies were assumed to reflect theincorrect description of fluorine.

The current neutron spectrum measurement appears to verifythe assumptions made in previous work. Comparison of the shapesof C/E-1 in different energy ranges (Fig. 6 and Fig. 8) show that theyare very similar for Teflon and FLIBE, the observed discrepanciescould be attributed to inaccuracies in fluorine description bynuclear data libraries since fluorine is present in both of thesematerials and other experiments without fluorine do not exhibitdiscrepancies.

0.8 1 1.2 1.4

V]

Measured spectrum FLIBE

Measured spectrum Te�lon

Measured spectrum FLINA

LON in the energy range of hydrogen proportional detectors.

0.8 1 1.2 1.4 [MeV]

ENDF/B-VII.1 FLIBE

ENDF/B-VII.1 Te�lon

ENDF/B-VII.0 FLINA

le by HPD, grey lines show 1r uncertainty interval.

1.E-04

1.E-03

1.E-02

1.E-01

1 2 3 4 5 6 7 8 9 10

Flu

x d

en

sity

(a

.u.)

Energy [MeV]

Meas. spectrum FLIBE

Meas. spectrum Te�lon

Meas. spectrum FLINA

Fig. 7. Neutron spectrum in fluoride salts (FLIBE and FLINA) and TEFLON in the energy range of a Stilbene scintillator.

-30%

-20%

-10%

0%

10%

20%

30%

0 1 2 3 4 5 6 7 8 9 10

C/

E-1

Energy [MeV]

ENDF/B-VII.1 FLIBE

ENDF/B-VII.1 Te�lon

ENDF/B-VII.0 FLINA

Fig. 8. C/E-1 comparison for FLIBE and TEFLON in the energy range attainable by a Stilbene scintillator, grey lines show 1r uncertainty interval.

292 E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295

The effects of using various libraries on calculation results forFLIBE salt are presented below in Fig. 9 and Fig. 10. Notable dis-crepancies between calculations using all libraries and experi-ments are observable in the region 0.1–1.3 MeV. Two trendsrelated to the nuclear data description of elastic and inelastic scat-tering on 19F are observable. These trends were also described inearlier measurements (Koštál et al., 2015). Influence in library dif-ferences is dampened in measurement in higher energies, and thediscrepancies are not as dramatic as in lower energies.

3.3. Results from S/U analyses

S/U analyses of LR-0 with MSRE FLIBE salt in the insertion zoneaimed at determination of similarity of the cores containing differ-

ent fluorinated compounds and investigated the energy-dependentsensitivity of keff to several nuclides. Table 6 shows the correlationcoefficient indices (ck) showing that from the point of view of crosssections, the cores Case 1, Case 15, Teflon and FLIBE are almostidentical. The similarity index of empty core (Case 1) and the corewith FLIBE insertion is in the same order as the similarity indexbetween the case with FLIBE insertion and Teflon insertion. Thecore with empty channel is almost identical to the core with Tefloninsertion. This implies that these small cores are rather identicalfrom the point of view of inserted material.

Fig. 11 shows the sensitivity per unit lethargy for 7Li and 19Ftotal cross sections over a full neutron energy range. Fig. 12 focuseson neutrons in the energy range of 0–1.4 MeV neutrons based onthe C/E results presented in Fig. 6. Assessments of whether discrep-

-40%

-30%

-20%

-10%

0%

10%

20%

30%

40%

0 0.2 0.4 0.6 0.8 1 1.2 1.4

C/

E-1

Energy [MeV]

ENDF/B-VII.1 FLIBE

JEFF-3.2 FLIBE

CENDL-3.1 FLIBE

RUBAFOND-2010 FLIBE

JENDL-4 FLIBE

Fig. 9. C/E-1 comparison for FLIBE using various libraries in the energy range attainable by an HPD.

-40%

-30%

-20%

-10%

0%

10%

20%

30%

0 1 2 3 4 5 6 7 8 9 10

C/

E-1

Energy [MeV]

ENDF/B-VII.1 FLIBE

JEFF-3.2FLIBE

CENDL-3.1 FLIBE

RUBAFOND-2010 FLIBE

JENDL-4 FLIBE

Fig. 10. C/E-1 comparison for FLIBE and various libraries in the energy range attainable by a Stilbene detector.

Table 6The similarity indices of the core with different fluorinated compounds.

Experiment Application Similarity index

Case 1 FLIBE 0.9970 ± 0.0002Case 1 Teflon 0.9991 ± 0.0002FLIBE Teflon 0.9963 ± 0.0003

E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295 293

ancies between experimental measurement and flux calculation(Fig. 6) line up with nuclear data sensitivities (Fig. 12) are compli-cated by the fact that the sensitivities are to keff of the systemwhileflux measurements are in the local salt insertion zone. Though it isdifficult to directly determine a single most important source of

error by comparing discrepancies in the C/E results to those inthe S/U results, the S/U results appear to show that 19F has rela-tively high sensitivity in the energy regions in which C/E resultsdiffer the most; however, 19F also has high sensitivity in otherenergy regions in which the C/E results show good agreementbetween experiments and calculations.

The SCALE TSAR tool was used to obtain more pronounced sen-sitivity profiles in two ways. In the first calculation, Case 1 (emptycentral channel) was taken as a reference core for replacementanalysis and subsequently this was filled by FLIBE and Teflon inser-tions to observe their influence. Table of top 10 reaction sensitivitycoefficients integrated over the energies and model volume isshown in Table 7. The largest sensitivity of the keff in both cases

Fig. 11. Sensitivity per unit lethargy over full neutron energy range for 19F and 7Liin LR-0 with MSRE FLIBE salt target.

Fig. 12. Sensitivity per unit lethargy over a neutron energy range of 0–1.4 MeV for19F and 7Li in LR-0 with MSRE FLIBE in the insertion zone.

Table 7Chart of the 10 most sensitive reactions obtained by application of the TSARmethodology with reference empty channel.

FLIBE insertion Teflon insertion

Isotope Reaction Sensitivity Isotope Reaction Sensitivity

1H elastic 1526 1H elastic 299656Fe n,gamma 591 235U nu-fission 2264235U fission 448 235U fission 95119F elastic 374 19F elastic 73916O elastic 331 1H n,gamma 5431H n,gamma 285 C elastic 521235U nu-fission 208 16O elastic 511238U nu-fission 192 238U nu-fission 404238U n,n’ 172 238U n,gamma 393238U fission 150 238U fission 324

Table 8Chart of the 10 most sensitive reactions obtained by application of the TSARmethodology with reference channel filled by FLIBE insertion and replaced by Teflon.

Isotope Reaction Sensitivity

1H elastic 108956Fe n,gamma 654C elastic 51319F elastic 41616O elastic 384238U elastic 184238U n,n’ 155235U nu-fission 14819F n,n’ 13253Cr n,gamma 123

294 E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295

is carried by reactions on hydrogen and 235U, in case of FLIBE inser-tion in stainless steel casing, capture on 56Fe is very important aswell. Elastic scattering on fluorine is the fourth with major sensi-tivity and larger importance than oxygen, even in the LR-0 (LWR)reactor type. Reactions on Be and 7Li are not present in the top10, elastic scattering on Be and 7Li are on the 16th and 17th placewith sensitivities 104 and 93 respectively.

In the second type of comparison, the TSAR methodology wasapplied to the core with FLIBE insertion, which was replaced byTeflon insertion. Results (see Table 8) show again large sensitivityon hydrogen and 56Fe; elastic scattering on fluorine occupies againthe fourth position before oxygen.

4. Conclusions

Experimental work with fluorine-rich compounds, includingoriginal MSRE salt, is a logical step toward validation of calcula-tional tools (computer codes) and methodologies for analysingand evaluating fluoride salt systems. Critical experiments withTeflon provided results with discrepancies similar to the bench-mark reference configuration with void channel (Case 1). However,systematic differences are clearly observable between two groupsof nuclear data libraries, differing by description of inelastic andelastic scattering. The suggested conclusion is that the ENDF/B-VII.1, ENDF/B-VII.0, JEFF-3-2, JEFF-3.1, and CENDL-3.1 libraries giveresults closer to the benchmark reference than the other threelibraries investigated (JENDL-3.3, JENDL-4, and RUSFOND-2010),which systematically overestimate reactivity by approximately90 pcm.

This systematic deviation between library sets is slightly sup-pressed in critical experiments with FLIBE salt, where all nucleardata libraries are 220–240 pcm off compared to the experiment,except the CENDL-3.1 library, which is substantially closer to thereference. A critical experiment with an empty canister shows thatthe canister material (stainless steel) used for FLIBE experimentsintroduces substantial discrepancy. If this is taken into account,discrepancies in the results of the critical experiment with FLIBEsalt would be very close to those of Teflon.

Neutron spectrum measurement confirmed notable discrepan-cies between experiment and calculation in the case of fluorinatedcompounds, as observed in previous experiments. There is a veryhigh probability that these discrepancies are caused by inaccura-cies in 19F cross section description. This conclusion is in agree-ment with previous results. Nuclear data libraries used forcalculations can be divided into two groups according to theinelastic to elastic scattering ratio description (Losa et al., 2015;Koštál et al., 2015), but based on the neutron spectrum measure-ment (conversely to the criticality), none of these libraries providesbetter results than the others.

Sensitivity analysis made by SCALE 6.2 software utilizingTSUNAMI-3D/TSUNAMI-IP and TSAR methodology shows thatelastic scattering on fluorine plays very important role in case ofMSR reactor studies. The neutron spectrum measurements in dif-ferent fluorinated insertions showed that the fast part of the spec-trum (over 100 keV) is shaped by interactions on fluorine.

Acknowledgments

This work was supported by the Project CZ.02.1.01/0.0/0.0/15_008/0000293: Sustainable energy (SUSEN) – 2nd phase 0108 realizedin the framework of the European Structural and Investment FundsPresented, and with use of infrastructures Reactors LVR-15 and LR-0, which was financially supported by the Ministry of Education,

E. Losa et al. / Annals of Nuclear Energy 120 (2018) 286–295 295

Youth and Sports – project LM2015074. Portions of this materialare also based upon work supported by the US Department ofEnergy Office of Nuclear Energy through the Advanced ReactorTechnologies program.

The authors also wish to acknowledge several people at ORNLfor their contributions: William J. Marshall for his important con-tributions and feedback on S/U analysis as well as his input on thismanuscript; Matt Jessee and Christopher Perfetti for their sugges-tions and guidance on the underlying S/U methods in SCALE andfeedback on input files; and David Holcomb for his helpful reviewand insights during this work.

References

Chadwick, M.B., Herman, M., Oblozinsky, P., Dunn, M.E., Danon, Y., Kahler, A.C.,Smith, D.L., Pritychenko, B., Arbanas, G., Arcilla, R., Brewer, R., Brown, D.A.,Capote, R., Carlson, A.D., Cho, Y.S., Derrien, H., Guber, K., Hale, G.M., Hoblit, S.,Holloway, S., Johnson, T.D., Kawano, T., Kiedrowski, B.C., Kim, H., Kunieda, S.,Larson, N.M., Leal, L., Lestone, J.P., Little, R.C., McCutchan, E.A., MacFarlane, R.E.,Macinnes, M., Mattoon, C.M., McKnight, R.D., Mughabghab, S.F., Nobre, G.P.A.,Palmiotti, G., Palumbo, A., Pigni, M.T., Pronyaev, V.G., Sayer, R.O., Sonzogni, A.A.,Summers, N.C., Talou, P., Thompson, I.J., Trkov, A., Vogt, R.L., Van der Marck, S.C.,Wallner, A., White, M.C., Wiarda, D., Young, P.G., 2011. ENDF/B-VII.1 nucleardata for science and technology: cross sections, covariances, fission productyields and decay data. Nucl. Data Sheets 112 (12). Special Issue on ENDF/B-VII.1Library, 2887–2996. ISSN 0090-3752.

Cvachovec, J., Cvachovec, F., 2008. Maximum likelihood estimation of a neutronspectrum and associated uncertainties. Adv. Mil. Technol. [online] 3 (2), 67–79.ISSN 1802-2308.

Dean, V.F. ICSBEP Guide to the Expression of Uncertainties. [Accessed 2017-11-27].https://www.oecd-nea.org/science/wpncs/icsbep/documents/UncGuide.pdf(September 30, 2008).

Goorley, T., 2012. Initial MCNP6 Release Overview. Nucl. Technol. 180 (3), 298–315.Jansky, B., Novak, E., 2014. Neutron spectrometry with spherical hydrogen

proportional detectors. Nucl. Instrum. Meth. Phys. Research Section A:Accelerators, Spectrometers, Detectors and Associated Equipment 735, 390–398. ISSN 0168-9002..

Koštál, M., Rypar, V., Losa, E., Schulc, M., Novak, A.E. 2017. VVER-1000 PhysicsExperiments Hexagonal Lattices (1.275 cm Pitch) of Low Enriched U(3.3 wt.%U235)O2 Fuel Assemblies in Light Water with Graphite and Fluoride SaltInsertions in Central Assembly. B.m.: OECD/NEA.

Koštál, M., Rypar, V., Milcák, J., Jurícek, V., Losa, E., Forget, B., Harper, S., 2016. Studyof graphite reactivity worth on well-defined cores assembled on LR-0 reactor.Ann. Nucl. Energy 87 (2), 601–611. ISSN 0306-4549.

Koštál, M., Veškrna, M., Cvachovec, F., Jánsky, B., Novák, E., Rypar, V., Milcák, J., Losa,E., Mravec, F., Matej, Z., Rejchrt, J., Forget, B., Harper, S., 2015. Comparison of fastneutron spectra in graphite and FLINA salt inserted in well-defined coreassembled in LR-0 reactor. Ann. Nucl. Energy 83, 216–225. ISSN 0306-4549.

Losa, E., Koštál, M., Rypar, V., Schulc, M., Novák, E., Veškrna, M., Mravec, F., Matej, Z.,Cvachovec, F. Benchmarking of the Graphite and Fluoride Salt Insertions in LR-0Reactor. In: Physics of Reactors 2016 (PHYSOR 2016): Unifying Theory andExperiments in the 21st Century. 2016, s. 1259–1270. ISBN 978-1-5108-2573-4.

Losa, E., Koštál, M., Powers, J.J., Brown, N.R., Mueller, D.E. Integral Experiments withTeflon and 7LiF-BeF2 Salt at LR-0 Reactor. In: M&C 2017 – InternationalConference on Mathematics & Computational Methods Applied to NuclearScience & Engineering. 2017.

Losa, E., Koštál, M., Rypar, V., Novák, E., Jurícek, V., 2015. Effect of inserted fluoridesalts on criticality in the LR-0 reactor. Ann. Nucl. Energy 81, 18–25. ISSN 0306-4549.

Perfetti, C.M., Rearden, B.T., Martin, W.R., 2016. SCALE continuous-energyeigenvalue sensitivity coefficient calculations. Nucl. Sci. Eng. 182 (3), 332–353. ISSN 0029-5639.

Rearden, B.T., Jessee, M.A., Eds. SCALE Code System, ORNL/TM-2005/39, Version 6.2,Oak Ridge National Laboratory, (2016).

Shaffer, J.H., 1971. Preparation and Handling of Salt Mixtures for the Molten SaltReactor Experiment. ORNL-4616, Oak Ridge National Laboratory, Oak Ridge, TN,USA.

US DOE, 2013A. Energy Department Completes Salt Coolant Material Transfer toCzech Republic for Advanced Reactor Research. [accessed 2014-12-17]. http://www.energy.gov/articles/energy-department-completes-salt-coolant-material-transfer-czech-republic-advanced-reactor (May 20, 2013).

US DOE, 2013B. United States and Czech Republic Establish a Joint Civil NuclearCooperation Center in Prague. [accessed 2014-12-17]. http://www.energy.gov/articles/united-states-and-czech-republic-establish-joint-civil-nuclear-cooperation-center-prague (June 12, 2013).

Veškrna, M., Matej, Z., Mravec, F., Prenosil, V., Cvachovec, F., Koštál, M. Digitalizedtwo parametric system for gamma/neutron spectrometry. In: 18th TopicalMeetings of the Radiation Protection & Shielding Division of the ANS (RPSD2014). 2014. ISBN 978-0-89448-714-9.