International benchmark tests of the FENDL—1 nuclear data library

E L S E V I E R Fusion Engineering and Design 37 (1997) 9 21

Fusion Engineeq'ng and Design

International benchmark tests of the FENDL-1 Nuclear Data Library

Ulrich Fischer *, International Working Group on Experimental and Calculational Benchmarks on Fusion Neutronics for FENDL Validation ~

Forsehungszentrum Karlsruhe, Institut fiir Neutronenphysik und Reaktorteehnik, PO Box 3640, D-76021 Karlsruhe, Germany

Abstract

An international benchmark validation task has been conducted to validate the Fusion Evaluated Nuclear Data Library F E N D L - 1 through data tests against integral 14 MeV neutron experiments. The main objective of this task was to qualify the F E N D L - 1 working libraries for fusion applications and to elaborate recommendations for further data improvements. Several laboratories and institutions from the European Union, Japan, the Russian Federation and US have contributed to the benchmark task. A large variety of existing integral 14 MeV benchmark experiments was analysed with the F E N D L - 1 working libraries for continuous energy Monte Carlo and multigroup discrete ordinates calculations. Results of the benchmark analyses have been collected, discussed and evaluated. The major findings, conclusions and recommendations are presented in this paper. With regard to the data quality, it is summarised that fusion nuclear data have reached a high confidence level with the available F E N D L - 1 Data Library. With few exceptions this holds for the materials of highest importance for fusion reactor applications. As a result of the performed benchmark analyses, some existing deficiencies and discrepancies have been identified that are recommended for removal in the forthcoming F E N D L - 2 data file. © 1997 Elsevier Science S.A.

Keywords: Benchmark validation task; Fusion Evaluated Nuclear Data Library; Integral 14 MeV

1. Introduction

* Corresponding author. Tel.: + 49 7247 822437; fax: + 49 7247 823824; e-mail: [email protected]

Y. Oyama, F. Maekawa, C. Konno, M. Wada, JAERI, C. Ichihara, Kyoto University; Y. Makita, A. Takahashi, Osaka University; K. Ueki, Ship Research Institute; K. Kosako, Sumitomo Industries; K. Hayashi, Hitachi Engineering; M. Youssef, UCLA; H. Hunter, C. Slater, ORNL; U. Fischer, F. Kappler, E. Stein, H. Tsige-Tamirat, E. Wiegner, FZK Karlsruhe; P. Batistoni, L. Petrizzi, V. Rado, ENEA Frascati; L. Benmansour, A. Santamarina, CEA Cadarache; K. Seidel, TUD Dresden; A. Blokhin, S.P. Simakov, V. Sinitsa, IPPE Obninsk: D. Markovskij, RRC-KI Moscow.

The Fus ion Eva lua ted Nuc lea r D a t a L ib ra ry ( F E N D L ) is a compi l a t ion o f fus ion-or ien ted d a t a evalua t ions selected f rom the nuclear da t a files E N D F / B - V I [1] (USA) , B R O N D [2] (Russ ian Federa t ion) , J E N D L [3] ( Japan) and E F F [4] (European Union) in an in te rna t iona l effort ini- t ia ted and co -o rd ina t ed by the I A E A Nuc lea r D a t a Sect ion [5,6]. The F E N D L 1 d a t a file has been r e c o m m e n d e d as reference l ibrary for design ca lcula t ions in the Engineer ing Des ign Act iv i ty

0920-3796/97/$17.00 © 1997 Elsevier Science S.A. All rights reserved. Pll S0920-3796(97)00046-X

10 U. Fischer et al./Fusion Engineering and Design 37 (1997) 9-21

Table 1 List of FENDL/E-I.0 data evaluations

Nuclide Source Nuclide Source Nuclide Source

H 1 ENDF/B-VI H 2 BROND-2 H 3 ENDF/B-VI Li--6 ENDF/B VI Li 7 ENDF/B VI Be-9 ENDF/B VI B-10 ENDF/B VI B-I1 ENDF/B VI C--nat ENDF/B VI N-14 BROND 2 N 15 BROND 2 0 16 ENDF/B-V1 F 19 ENDF/B-VI Na-23 JENDL-3 Mg nat JENDL 3 AI-27 JENDL-3 Si-nat BROND 2 P-31 ENDF/B-V1 S nat ENDF/B-VI C1 nat ENDF/B-VI K-nat ENDF/B-VI Ca-nat JENDL-3 Ti-nat JENDL-3 V nat ENDF/B VI Cr-50 ENDF/B VI Cr 52 ENDF/B VI Cr 53 ENDF/B-VI Cr-54 ENDF/B VI Mn-55 ENDF/B-VI Fe 54 ENDF/B-VI Fe 56 ENDF/B VI Fe-57 ENDF/B VI Fe 58 ENDF/B VI Co 59 ENDF/B-VI Ni 58 ENDF/B VI Ni 60 ENDF/B VI Ni-61 ENDF/B VI Ni-62 ENDF/B-VI Ni-64 ENDF/B--VI Cu-63 ENDF/B-VI Cu 65 ENDF/B VI Zr 90 BROND 2 Zr 91 BROND 2 Zr-92 BROND-2 Zr-94 BROND 2 Zr-96 BROND-2 Nb 93 BROND-2 Mo nat JENDL-3 Sn nat BROND 2 Ba-134 ENDF/B-VI Ba 135 ENDF/B V! Ba 136 ENDF/B VI Ba 137 ENDF/B VI Ba 138 ENDF/B-VI Ta 181 JENDL-3 W 182 ENDF/B VI W 183 ENDF/B VI W 184 ENDF/B VI W-186 ENDF/B-V1 Pb-206 ENDF/B VI Pb-207 ENDF/B-VI Pb 208 ENDF/B-VI Bi 209 JENDL-3

(EDA) phase of the International Thermonuclear Experimental Reactor ( ITER) Project.

F E N D L - 1 working libraries have been processed for use in continuous energy Monte Carlo transport calculations ( F E N D L / M C - 1 . 0 [7] for applications with the M C N P code [9]) and discrete ordinates transport calculations (multigroup library F E N D L / M G 1.0 [8]). Prior to their use in design calculations there is a need to validate the F E N D L 1 working libraries through integral data tests.

An international benchmark validation task has been conducted to validate the F E N D L 1 working libraries through data tests against available integral experiments. The main objective was to qualify the F E N D L 1 working libraries for fusion applications and to elaborate recommendations for potential data improvements.

Several laboratories and institutions from the European Union, Japan, the Russian Federation and the US have contributed to this benchmark task by analysing a large variety of existing integral 14 MeV benchmark experiments with the F E N D L 1 working libraries, i.e. the F E N D L /

M C - 1 . 0 and the F E N D L / M G - 1 . 0 data libraries. Results of the F E N D L - 1 data testing are summarised and evaluated in this report.

2. F E N D L - 1 data fibraries

Following the recommendations of several IAEA consultants meetings, a F E N D L - 1 general purpose evaluation data file, designated as F E N D L / E - 1 . 0 , has been compiled including neutron interaction and photon production cross section data from the B R O N D - 2 , E N D F / B - V I , and J E N D L - 3 data files [5,6] (Table 1).

Working libraries for use in design calculations have been derived by R.E. MacFarlane of LANL, using the NJOY code system [10]. For use in continuous-energy Monte Carlo calculations with the M C N P code, a F E N D L ACE data library was derived, denoted as F E N D L / M C - 1 . 0 [7]. For use in discrete ordinates calculations the multigroup library F E N D L / M G - 1 . 0 [8] has been cre- ated in the V I T A M I N - J group structure with 175 neutron and 42 photon groups and the related V I T A M I N J weighting spectrum.

U. Fischer et al./Fusion Engineering and Design 37 (1997) 9-21 11

The FENDL/MG-1.0 and FENDL/MC-1.0 are the reference working libraries for ITER-EDA design calculations. It was mandatory to use these working libraries in the benchmark analyses for the purpose of the FENDL validation. The appli- cation of other data sources, e.g. processed libraries derived from BROND-2, EFF-2 , ENDF/B-VI and JENDL-3, was recommended for comparison and cross-checking purposes.

3. The FENDL-1 benchmark validation task

Following the recommendations of the IAEA Advisory Group Meeting on 'Improved Evalua- tions and Integral Data Testing for FENDL' held at Garching, Germany, September 12 16, 1994, an international FENDL benchmark task was launched with the objective of performing integral data tests to qualify the F E N D L - l working libraries for fusion applications.

Several laboratories and institutions from the European Union, Japan, the Russian Federation and the US have contributed to this benchmark task by analysing a variety of existing integral 14 MeV neutron benchmark experiments, see the overview given in Table 2. The majority of the analysed experiments are included in the IAEA compilation of fusion neutronics benchmark experiments [11] which is available on-line at IAEA/ NDS.

4. Benchmark contributions

Benchmark contributions were provided by JAERI, Tokai-mura [12], the Universities of Os- aka and Kyoto [13], Hitachi Engineering [14,15], UCLA [16,17], ORNL [18], ENEA Frascati [19], the Technical University of Dresden (TUD) [20], Forschungszentrum Karlsruhe (FZK) [21], CEA Cadarache [22], IPPE Obninsk [23,24], and the Kurchatov Institute (KIAE), Moscow [25].

The JAERI contributions comprise benchmark analyses of FNS (Fusion Neutron Source of JAERI) time-of-flight measurements on angular neutron spectra for Li20, Be, C, O, N, Fe and Pb cylindrical slabs, in-system measurements on

Li20, Be, C, Fe, Cu and W cylindrical slabs, analyses of the FNS bulk shield experiments on large SS-316 assemblies, and analyses of the OK- TAVIAN spherical shell measurements on gamma ray leakage spectra for LiF, CF2, A1, Si, Ti, Cr, Mn, Co, Cu, Nb, Mo, W and Pb. A comprehensive documentation of the analysed experiments is included in the JAERI collection of experimental fusion benchmark data [26]. The FNS bulk shield experiments are documented in Ref. [27,28]. The benchmark calculations have been performed with the MCNP code and the FENDL/MC-1.0 data library. In addition, comparisons with the JENDL-3 data were provided.

The FNS in-system experiments on Li20, Be and C cylindrical slabs were also analysed by Hitachi, using the DOT 3.5 code [29] and the FENDL/MG-1.0 data library, thus providing the data base for comparing continuous energy MCNP-calculations with deterministic SN-transport calculations using multigroup data in the VITAMIN-J group structure.

The universities of Kyoto and Osaka contributed benchmark analyses of the OKTAVIAN spherical shell transmission experiments on neutron leakage spectra for the materials Be, Li, LiF, CF2, A1, Si, Ti, Cr, Mn, Co, Ni, Cu, Zr, Nb, Mo and W [13]. The documentation of the OKTA- VIAN experiments is again included in the JAERI collection of experimental fusion benchmark data [26]. The benchmark calculations were performed with lhe MCNP code and the FENDL/MC 1.0 data library including comparisons with JENDL 3 data.

The FNS bulk shield experiments on large SS- 316 assemblies were analysed by UCLA using the discrete ordinates code DORT [30] and the FENDL/MG 1.0 data [16,17]. The FENDL 1 multigroup data were collapsed to 80 neutron and 24 gamma groups for use in the two-dimensional DORT-calculations. Comparisons with ENDF/ B-VI based calculations were included.

Two experiments included in the SINBAD compilation of shielding benchmark experiments [31] were analysed by RSIC/ORNL applying two- dimensional DORT calculations and FENDL/ MG-1.0 multigroup data: a bulk shield experiment with stainless steel with and without a

U,

Tab

le 2

In

tegr

al e

xper

imen

ts a

nd a

naly

ses

cont

ribu

ted

to t

he F

EN

DL

-I

benc

hmar

k va

lida

tion

tas

k

Typ

e of

int

egra

l ex

peri

men

t M

ater

ial

conf

igur

atio

n M

ater

ial

Exp

erim

ent

Ben

chm

ark

anal

ysis

TO

F-m

easu

rem

ents

of

angu

lar

Cyl

indr

ical

sla

bs

Li2

0, B

e, C

, O

, N

, Fe

, P

b F

NS

/JA

ER

I F

NS

/JA

ER

I le

akag

e sp

ectr

a In

-sys

tem

mea

sure

men

ts o

f ne

u-

Cyl

indr

ical

sla

bs

Liz

O,

Be,

C,

Fe,

Cu,

W

tron

spe

ctra

and

rea

ctio

n ra

tes

TO

F-m

easu

rem

ents

of

neut

ron

Sphe

rica

l sh

ells

B

e, B

e-L

i, L

i, L

i20,

LiF

, C

, le

akag

e sp

ectr

a C

F2,

A1,

Si,

Ti,

Cr,

Mn,

Co,

Ni,

C

u, Z

r, N

b, M

o, W

L

iF,

CF2

, A

1, S

i, T

i, C

r, M

n,

Co,

Cu,

Nb,

Mo,

W,

Pb

SS-3

16

SS-3

16

SS-3

16,

SS-3

16 a

nd p

ersp

ex

Be,

AI,

Fe,

Pb-

17L

i

Be

Fe

Be

Fe

TO

F-m

easu

rem

ents

of

gam

ma

ray

leak

age

spec

tra

Bul

k sh

ield

exp

erim

ent

Bul

k sh

ield

and

str

eam

ing

expe

ri-

men

t B

ulk

shie

ld a

nd n

ucle

ar h

eati

ng

expe

rim

ent

TO

F-m

easu

rem

ents

of

neut

ron

leak

age

spec

tra

TO

F-m

easu

rem

ents

of

neut

ron

leak

age

spec

tra

Neu

tron

and

pho

ton

leak

age

spec

- tr

a m

easu

rem

ents

M

ulti

plic

atio

n ex

peri

men

t M

ock-

up s

hiel

d ex

peri

men

t

Sph

eric

al s

hell

s

Cyl

indr

ical

blo

ck

Rec

tang

ular

blo

ck,

wit

h an

d w

itho

ut d

uct

Rec

tang

ular

blo

ck

Sphe

rica

l sh

ells

Sph

eric

al s

hell

s

Rec

tang

ular

sla

b, w

ith

and

wit

h-

out

stra

ight

gap

S

pher

ical

she

lls

Rec

tang

ular

sla

b, w

ith

an w

itho

ut

stra

ight

gap

FN

S/J

AE

RI

Uni

vers

itie

s of

O

saka

and

Ky-

o

to

Uni

vers

itie

s of

O

saka

and

Ky-

o

to

FN

S/J

AE

RI

OR

NL

EN

EA

F

rasc

ati

1PPE

Obn

insk

FZ

K K

arls

ruhe

TU

D D

resd

en

KIA

E M

osco

w

MeP

hl-K

IAE

M

osco

w

FN

S/J

AE

RI

Hit

achi

Ltd

., K

IAE

M

osco

w

Uni

vers

itie

s of

Osa

ka a

nd K

yoto

, F

ZK

Kar

lsru

he,

IPP

E O

bnin

sk,

KIA

E M

osco

w

FN

S/J

AE

RI

FN

S/J

AE

RI,

U

CL

A

RS

IC/O

RN

L

EN

EA

Fra

scat

i

IPP

E O

bnin

sk,

KIA

E M

osco

w,

FZ

K K

arls

ruhe

, C

EA

Cad

arac

he

FZ

K

Kar

lsru

he,

Hit

achi

Ltd

.

TU

D D

resd

en,

FZ

K K

arls

ruhe

KIA

E M

osco

w

KIA

E M

osco

w

e~

2"

t~

',a,a

"-

-I

,g

I


polyethylene shield, and an iron duct experiment for both a plugged and unplugged configuration.

ENEA Frascati contributed analyses of the FNG integral experiments on a SS-316 bulk shield and a SS-316/water shield assembly [32]. The benchmark calculations were performed with MCNP and FENDL/MC-1.0 data including comparisons with E F F - I and -2 data [19].

The TUD iron slab experiment [33] on neutron and photon spectra leaking from a 30-cm thick iron slab with and without a straight gap was analysed by TUD using the MCNP code and the FENDL/MC-1.0 data library. Comparisons were included with MCNP-calculations using E F F - I and -2 data [20].

The benchmark contributions of Forschungs- zentrum Karlsruhe [21] comprise analyses of spherical shell neutron transmission experiments for Be, Be-Li, A1, Fe, Cu, Si, Zr, Pb and Pb-17Li conducted at OKTAVIAN, IPPE Obninsk, FZK and TUD. Calculations were performed with both the ONEDANT code and FENDL/MG 1.0 data and the MCNP code with FENDL/MC-1.0 data. Comparisons were performed with EFF-1 and

2 based calculations. CEA Cadarache contributed benchmark analy-

ses of spherical iron shell assemblies performed with the two-dimensional discrete ordinates code BISTRO [34] using the FENDL/MG-1.0 iron data. Comparisons with EFF-1 and EFF 2 multigroup data and sensitivity/uncertainty analyses were included [22].

IPPE Obninsk contributed analyses of the IPPE transmission experiments on Be and Fe spherical shells [24,35] and the OKTAVIAN experiments on Nb, Si and Zr spherical shell assemblies. Cal- culations were performed with both discrete ordinates codes (ONEDANT [36], ANISIN [37]) and FENDL/MG-1.0 and the MCNP code with FENDL/MC--1.0 data. Comparisons with ENDF/B VI and BROND-2 based calculations were provided.

The Kurchatov Institute Moscow performed benchmark analyses on spherical shell and slab experiments tbr beryllium and iron using the MCNP4A code and the FENDL/MC-1.0 data library [25]. Comparisons were included with Monte Carlo calculations using the BLANK code

[38] and data processed from the ENDF/B-VI data file.

5. Main results of FENDL-1 data test analyses

The main results of the integral data tests are summarised in the following for the three material groups: (1) multiplying and breeding materials, comprising the neutron multipliers Be and Pb, the breeding material Li and the breeding material constituents A1, Si, Zr, (2) the major shielding/ structural material Fe, Cr, Mn, Ni, Cu, W; and (3) the other materials C, O, N, F, Co, Nb, Mo, Ti. The focus is on the comparison of measured and calculated neutron spectra, as the neutron spectrum is the primary quantity of interest in neutron transport calculations. This includes analyses of direct neutron spectrum measurements and of activation and fission rate measurements, which are sensitive to different parts of the neutron spectrum. Note that only summary results are given here whereas detailed numerical results including tables and figures are given in the final report on the FENDL benchmarking [39]. In the following, for evaluating the differences between calculational and experimental results on a qualitative basis, the term 'good agreement' is used when the calculation clearly agrees with the mea- surement, 'satisfactory' when the agreement is still within the experimental error band apart from some insignificant deviations, 'unsatisfactory' when the calculation is clearly outside the experimental error band.

5.1. Multiplying and breeding materials

5.1.1. Beryllium The neutron multiplication power of beryllium

can be reproduced fairly well by calculations with cross-sections from different data files, including FENDL 1, EFF-1, E F F - 2 and ENDF/B VI. In any case the deviation is within the experimental uncertainty band of 3-7%.

However, this agreement is caused by compen- sating over- and underestimations in the lower and upper part of the spectra measured in the spherical shell transmission experiments. In par-

14 U. Fischer et al./Fusion Engineering and Design 37 (1997) 9--21

ticular, the spectrum is underestimated with FENDL-1 data by 10-20% in the energy range around l MeV. Better agreement is obtained with other beryllium data, e.g. the Young and Stewart evaluation [40] which is contained in the EFF-1 data file.

In the FNS TOF-experiments on 5.08- and 15.24-cm thick beryllium slabs an overall good agreement is obtained for the angular spectra. This is different from the observations found in the spherical shell experiments. It may be due to the fact that in the latter experiments all scattering angles--including backward directions--con- tribute to the measured leakage spectra, whereas in the TOF slab experiments the leakage spectra are measured in forward directions with no significant contributions from backward scattering events. This is in agreement with the observation that measured double-differential beryllium data are strongly underestimated by FENDL-1 (ENDF/B-VI) at backward angles [41].

In conclusion, a revision is needed for the secondary energy and angle distributions of the FENDL-1 Be data evaluation.

5.1.2. Lead The neutron multiplication of lead can be pre-

dicted within he experimental uncertainty band of 3-4%. In addition, the neutron leakage spectra measured both in spherical shell transmission experiments and the FNS slab experiments--angular spectra for thin (5.08 cm) and thick (20.3, 40.6 cm) lead slabs have been analysed--can be well reproduced by FENDL 1 based calculations. It is concluded that no major revisions or improvements are needed for the FENDL-1 lead data evaluation.

5.1.3. Lithium Several integral experiments have been analysed

involving the breeding material lithium: the FNS experiments on Li20 slabs with both TOF and in-system measurements, the OKTAVIAN experiments on pure lithium and LiF spherical shells as well as on combined Be/Li shell configurations, and the IPPE Obninsk experiment on a spherical Pb-17Li shell. Good agreement was obtained for the angular spectra measured in the TOF-experi-

ments for the Li20 slabs, although the forward neutron transmission above 14 MeV tends to be overestimated with increasing thickness. Further investigations are required to trace back the source of this discrepancy to one of the Li20-constituents, i.e. lithium or oxygen.

An overall good agreement for the neutron leakage spectrum is also observed in the OKTA- VIAN experiment on a lithium metal spherical shell. There is, however, a underestimation of the high energy spectrum 5-10 MeV which is not seen in the FNS slab experiments. Although this discrepancy is not too serious, it is indicated to check again the secondary energy and angle distributions of the 6"7L FENDL-1 data evaluations.

5.1.4. Breeding material constituents: Al, Si, Zr Results are available for measured neutron

leakage spectra of spherical shell experiments performed at OKTAVIAN (A1, Si and Zr) and 1PPE Obninsk (AI). For these materials the agreement between experiment and FENDL 1 calculations in general is very unsatisfactory. In the Oktavian experiment for aluminium, e.g. there is an overestimation of the low energy part of the spectrum (E< 1 MeV) by about 25% and an underestimation by about 40% of the high energy region 5 10 MeV. The latter underestimation is also observed in the secondary energy distributions of the Al(n, xn)-reaction when compared to measured neutron emission cross-section data. JENDL-3.2 and EFF-2 data give similar results as FENDL l, with a better reproduction of the high energy range, however.

For silicon there is an overestimation of the low energy part of the measured leakage spectrum ( E < I MeV) by as much as 50-100%,, but a systematic underestimation by about 30% of the high energy region 1-10 MeV. Again the latter underestimation is observed in the differential secondary energy distribution data. The more recent JENDL-FF (Fusion File) and EFF-2 data give a better reproduction of the leakage spectrum in the energy range below 1 MeV.

The leakage spectrum of zirconium can be reproduced more satisfactorily with FENDL-1 data, although there is a systematic overestimation by about 20% of the high energy part be-

u. Fischer et al./Fusion Engineering and Design 37 (1997) 9 21 15

tween 1 and 10 MeV. This does not correspond to the good agreement found for the differential secondary energy distribution data in that energy range.

5.2. Structural and/or shielding materials"

below 5 MeV. The spectrum in the energy range 5 - l 0 MeV is overestimated by about 36%. Better agreement is obtained with JENDL-data. A similar trend is observed in the FNS in-system slab experiment for the neutron spectrum above 1 MeV.

5.2.1. Iron For the analysed iron transmission experiments

an underestimation of the measured total neutron leakages was obtained which is mainly caused by the underestimated low energy part (E< 1.5 MeV). This holds for all iron data evaluations used, including FENDL-1 and EFF-1, and EFF 2. In the TUD iron slab experiment with a thickness of 30 cm, e.g. the low energy part (E < 1.0 MeV) of the spectrum is underestimated by about 15% and the total measured neutron flux by 10%. In the high energy range (2 10 MeV) good agreement is obtained with both FENDL-1 and EFF 2 data.

5.2.2. Nickel No appropriate integral data are available for

nickel. For the existing OKTAVIAL transmission experiment there appears to be a normalisation problem as the transmitted neutron source peak is overestimated by about a factor 2.5. The qualitative shape of the leakage spectrum can be reproduced fairly well by the F E N D L - I based calculation. Additional integral experiments are required to qualify the FENDL-1 nickel data.

5.2.3. Chromium There is quite good agreement with the neutron

leakage spectrum measured in an OKTAVIAN spherical shell experiment for the energy range 15 to about 2 MeV. Below 1 MeV neutron energy FENDL-1 gives, however, an overestimation by as much as 50%. In the energy range 5-10 MeV FENDL-1 overestimates the measured spectrum by about 10% whereas good agreement is obtained with JENDL-3.2 and J E N D L - F F data.

5.2.4. Copper For the OKTAVIAN copper experiment on a

27.5-cm thick spherical shell there is good agreement with the measured neutron leakage spectrum

5.2.5. Manganese For manganese there is an OKTAVIAN experi-

ment on a 27.5-cm thick spherical shell. As for chromium there is rather good agreement with the measured neutron leakage current over the energy range 15 to about 0.5 MeV with a slight overestimation in the order of 10%.

5.2.6. Tungsten For the OKTAVIAN experiment on a 9.8-cm

thick tungsten spherical shell there is an underestimation of the measured neutron leakage spectrum by about 10% on the average. For the high energy range 5-10 MeV the underestimation amounts up to 30%.

In the FNS in-system slab experiment a similar trend is observed. The neutron spectrum is underestimated in the energy range from l to about 12 MeV at most locations in the tungsten slab. The high energy spectrum part (E> 12 MeV), however, is considerably overestimated. The low energy part of the spectrum again is underestimated.

5.2.7. SS-316 Fox" the SS-316 bulk shield experiments per-

formed at FNG and FNS, a unique trend was observed in underestimating the high energy tail (E > 10 MeV) of the neutron spectrum. This cor- responds to the results obtained for the integral experiments on iron, which is the main con- stituent of SS-316. However, inconsistent results were found for the low energy (E < 0.1 MeV) tail of the neutron spectrum as for the energy range between 1 and 10 MeV.

In the two FNS experiments the underestimation of the high energy neutron flux (E > 10 MeV) amounts to 25-35% at deep locations when using discrete ordinate calculations with multigroup data and to about 10% when using Monte Carlo calculations. In the FNG-experiment the underestimation of the fast responses which are sensitive

16 U. Fischer et al./Fusion Engineering and Design 37 (1997) 9 21

to this energy range, e.g. the 27AL(n, 7)-reaction, amounts to I0-20% at maximum.

For the neutron spectrum in the energy range 2 -10 MeV an overestimation is observed at the front locations of the FNS steel block. With increasing depth this overestimation turns into an underestimation of ~ 10% both with Monte Carlo and discrete ordinates calculations. This is consis- tent with the observations found in iron spherical shell experiments. For the Monte Carlo calculations, the deviations from the measured values are in general much smaller and there is a tendency to agree with the measurements within the experimental error. In the F N G experiment, there is a trend for underestimating the reaction rates sensitive to neutrons in the energy range 1 10 MeV, e.g. the ~lSIn (n, n') reaction by about 10 20% and the 237Np fission rate by up to 25%.

In the energy range 0.1-1.0 MeV, there is again an underestimation observed in the neutron spectra of the FNS SS-316 experiments that amounts up to 40% both with Monte Carlo and discrete ordinates. Below 0.1 MeV there is an underestimation at deep locations inside the block, but again an overestimation at the front part. For the Monte Carlo calculations this trend is not so clear than it is for the discrete ordinates calculations. Again the Monte Carlo calculations tend to agree with the measurements within the experimental error.

In the FNG-experiment, the 235U fission and the 55Mn(n, ),) rate are satisfactorily predicted with F E N D L - 1 data while the t97Au(n, ~,)--rate is underestimated by about 10%. In the FNS experiment, on the other hand, the 235U fission rate is underestimated whereas the 197Au(n, ) , ) --rate is well predicted by Monte Carlo calculations with a tendency to overestimations for the locations deep inside the steel block. For the discrete ordinates calculations, however, there is a different trend: underestimations at the front locations, overestimations in the middle part of the block and strong underestimations at the back positions for the 197Au(n, y) - - ra te .

In summary, there is a trend of underestimating the integrated spectrum in the energy ranges E, > 10 MeV and 2 < E, < 10 MeV at deep locations which is also reflected in high-threshold reactions at these locations. No unique trend can be de-

duced from the analyses of the SS-316 experiments for the neutron energy below 0.1 MeV. It can be stated, however, that, except for locations that are deep inside the SS-316 assembly, the agreement with the experimental results is not too bad taking into account the experimental errors as well as the uncertainties in the dosimetry cross-section data. In particular this holds for Monte Carlo calculations that agree much better with the measurements than do the discrete ordinates calculations with multigroup data.

5.3. Other materials: C, O, N, F, Co, Nb, Mo, 77

5.3.1. Graphite In the FNS TOF-experiments an underestima-

tion of the low energy leakage spectra by 10-20% is observed while JENDL-3 data agree better with the measured spectra. For the high energy part of the spectra the agreement with F E N D L - 1 data is in general satisfactory.

5.3.2. Oxygen The analyses of the FNS TOF-experiments

show a systematic underestimation of the measured leakage spectrum, except for the forward direction. The underestimation amounts to about 20% for the low energy part (E < 1 MeV) and up to 40% for the high energy parts of the spectrum. Better agreement is obtained with J E N D L - 3 data.

5. 3.3. Nitrogen As for oxygen, the analyses of the FNS TOF-ex-

periments show a clear trend of underestimating the measured leakage spectra with F E N D L - 1 data. The underestimation in general is as much as 40%. J E N D L - 3 data can better reproduce the measured spectra.

5,3.4. Fluorine There are two OKTAVIAN experiments in-

volving fluorine: one on a LiF and one on a CF2 spherical shell assembly. For both experiments serious deficiencies are observed that may be largely addressed to the F E N D L - 1 data for fluorine: there is an underestimation of the leakage spectra by as much as 40% over the whole

u. Fischer et al./ Fusion Engineering and Design 37 (1997) 9-21 17

measured energy range in the case of the CF 2 spherical shell and a strong underestimation of the energy range 3-1 MeV in the case of the LiF sphere. For the latter one there is also an underestimation of the 5-10 MeV range by about 20%, which was also found in the OKTA- VIAN Li sphere experiment, see Section 5.1 above.

5.3.5. Cobalt There is an underestimation by about 40% of

the leakage spectrum below 10 MeV measured in the OKTAVIAN experiment on a 9.8-cm thick cobalt spherical shell. Only the high energy part of the spectrum (E > 10 MeV) is well predicted.

5.3.6. Niobium For niobium an overestimation by 20-30% is

observed for the leakage spectrum below 10 MeV measured in an OKTAVIAN experiment. The high energy part of the spectrum (E > 10 MeV) is well predicted. JENDL 3.2 and JENDL-FF data reproduce the measured spectrum above 1 MeV within 10%.

5.3.7. Molybdenum The leakage spectrum for a 27.5-cm thick

molybdenum spherical shell again has been measured in an OKTAVIAN experiment. The agreement for the calculated FENDL-1 leakage spectrum is rather satisfactory, although there is an underestimation of the spectrum in the energy range 5-10 MeV by as much as 30%. There is an overall better agreement for JENDL- 3.2 data.

5.3.8. Titanium As for cobalt, a large discrepancy is found for

titanium in an OKTAVIAN experiment on a 9.8 cm thick spherical shell. FENDL-1 gives a overestimation by about 40% for the leakage spectrum below 5 MeV. The high energy part of the spectrum (E > 10 MeV) is overestimated by 20%. For the energy range 1-10 MeV there is a better agreement with JENDL-3.2 and JENDL-FF data.

5.4. Gamma-ray spectra and heating rates

Gamma-ray spectra have been measured in OKTAVIAN experiments for the materials LiF, CF2, A1, Si, Ti, Cr, Mn, Co, Cu, Nb, Mo, W and Pb. In the energy range 0.5 to about 5-6 MeV there is an overall good agreement between measured and calculated gamma-spectra for most of the analysed materials, e.g. LiF, CF2, A1, Si, Cu, Mo, W and Pb. Significant discrepancies are observed in this part of the spectrum for the materials Si, Cr, Mn, Nb, Co and Ti. This results in underestimations of the integrated spectra by 20% and more for Mn, Nb and Si, and an overestimation by about 30% for Cr. Systematic discrepancies- in general underestimations--are observed for the high energy part (E> 5 MeV) of the gamma-spectra. In particular this holds for titanium, manganese, cobalt, niobium, tungsten, and lead. With regard to FENDL 1 data, this is serious for lead, for which JENDL-3.2 data give much better agreement. When comparing C/E- data for the integrated gamma spectra and energy release rates, it is observed that there is much more disagreement for the latter ones. For Co, e.g. FENDL-1 gives the correct total gamma flux but an underestimation of the related heating rate by about 20%.

In the FNG nuclear heating experiment the nuclear heating rate has been measured for SS- 316 inside the layered shielding block up to 70 cm depth. An underestimation by about 10% has been found for MCNP-calculations with FENDL/ MC 1.0 data, both at shallow and deep positions inside the block. This underestimation, however, is within the experimental error. As the nuclear heating in SS 316 is mainly due to gamma ray interactions, this underestimation has to be addressed to the gamma heating of the SS-316 constituents which is mainly iron.

In the FNS bulk shielding experiment on large SS-316 assemblies the gamma heating rate has been measured along a central channel up to 91 cm depth. A similar trend of underestimating the nuclear heating in SS-316 has been obtained as in the FNG-experiment, although the results of the MCNP-calculations with FENDL-1 data largely lie within the band of experimental uncertainties.

18 U. Fischer et al./Fusion Engineering and Design 37 (1997) 9-21

5.5. Monte Carlo vs. discrete ordinates calculations

Monte Carlo calculations with continuous energy FENDL/MC-1.0 cross-section data and discrete ordinates calculations with FENDL/ MG-I .0 multigroup data have been performed for the same benchmark experiments to allow a direct comparison of the two different computa- tional approaches and data libraries. In general, the two approaches give the same results as has been shown both for analyses of spherical shell and slab experiments. There are, however, two exceptions to this rule encountered in the course of the benchmark analyses:

(1) Transport problems involving neutron thermalisation cannot be properly accounted for in discrete ordinates calculations with multigroup data in the VITAMIN-J group structure. This is due to missing up-scattering capabilities in the VITAMIN-J group data while splitting the thermal energy range into two energy groups.

(2) Deep penetration problems can be better described by Monte Carlo calculations with continuous energy cross-section data than by discrete ordinates calculations with multigroup data. Problems arise by the a-priori selection of the energy group structure, the spatially varying weighting function, the resonance shielding and the associated multigroup data processing. In principle, a problem dependent group data set would be required for analysing deep penetration problems with multigroup data. In addition to this, deficiencies are introduced by the discrete ordinates technique itself in describing the strongly forward peaked neutron transport in a deep penetration problem.

6. Conclusions and recommendations

Comprehensive data test analyses have been performed with the working libraries FENDL/ MG-1.0 for discrete ordinates calculations and FENDL/MC 1.0 for Monte Carlo calculations with the MCNP code. A variety of available integral fusion benchmark experiments has been

analysed to allow the qualification of the FENDL-1 working libraries for fusion applications.

With regard to the data quality, it can be stated that fusion nuclear data have reached a high confidence level with the available FENDL-1 data library. With few exceptions this holds for the materials of highest importance for fusion reactor applications. As a result of the performed benchmark analyses, existing deficiencies and discrepancies have been identified that are recommended to be removed in the forthcoming FE NDL -2 data file.

The major findings and conclusions of the FENDL-1 benchmark testing are summarised in a qualitative evaluation scheme in Table 3. For each material it is indicated by the term 'further improvements needed', if there is a need to improve the data evaluation for FENDL-2, or 'satisfactory' if the data quality of the existing FENDL-1 evaluations seems to be sufficient according to the results of the performed integral data tests. Detailed numerical results of the data tests that formed the basis for the qualitative evaluation are extensively given in the FENDL-1 benchmark report [39], both in tabu- lated and graphical form.

In comparing Monte Carlo calculations using continuous energy cross-section data and discrete ordinates calculations applying multigroup data in the VITAMIN J group structure it was found that the Monte Carlo approach in general gives better agreement with integral experiments. This was observed for thennalized systems, e.g. beryllium and graphite material assemblies with no absorbers present, and deep penetration problems, e.g. the FNS SS-316 bulk shield experiment. It is concluded that the Monte Carlo technique with continuously represented cross- section data allows to handle the encountered fusion neutronics problems with confidence while discrete ordinate calculations using the FENDL/MG-1.0 multigroup data library do not necessarily. Care has to be taken in applying FENDL/MG 1.0 to fusion neutronics problems with well thermalised neutron spectra and deep penetration problems.

U. Fischer et al . / Fusion Engineering and Design 37 (1997) 9-21

Table 3 Major findings of the FENDL-I integral data tests

19

Element Data quality Comments

Neutron multiplication and breeding materials Be Further improvements SED/SAD a to be improved; neutron multiplication well predicted; discrepant

needed integral experiments need to be clarified Pb Satisfactory Li Satisfactory AI, Si, Zr Further improvements SED to be improved; 7-production to be improved for Si

needed

Structural and/or shielding materials Fe Further improvements

needed Cr Further improvements

needed Cu Further improvements

needed Mn Satisfactory Ni Unclear W Further improvements

needed SS-316 Further improvements

needed V Unclear

Fluctuation factors to be included in partial neutron cross-sections; need for including anisotropic 7-emission data indicated Need for additional integral experiments; 7-production to be improved

SED in 5-10 MeV range and below 1 MeV to be improved

Need for additional integral experiments Urgent need for new integral experiment Improvements for SED and y-production needed

Disagreement for high and intermediate energy range of neutron spectrum

Urgent need for integral experiment; currently no data available

Other materials C

N

O

F

Ti

Co

Nb

Mo

Further improvements needed Further ~mprovements needed Further ~mprovements needed Further ~mprovements needed Further amprovements needed Further ~mprovements needed Further ~mprovements needed Satisfactory

SED needs improvement

SED/SAD needs improvement

SED/SAD needs improvement

Large discrepancies observed in integral data tests; SED possibly to be improved although differential data in agreement with experimental data SED needs improvement

SED needs improvement; need for more integral experiments

SED needs improvement; 7-production to be improved

Minor discrepancy in neutron spectrum

SED/SAD means secondary energy distribution/secondary angular distribution.

Acknowledgements

It is a pleasure to thank the members of the In te rna t iona l Work ing G r o u p on 'Exper imenta l and Calcula t ional Benchmarks on Fus ion Neu- tronics for F E N D L Val ida t ion ' for their fruitful

co-operat ion and their valuable benchmark con-

tr ibutions. Special thanks are due to S. Ganesan , BARC, formerly at I A E A / N D S , and A.B. Pashchenko, I A E A / N D S , for their engagement and active suppor t of the F E N D L benchmark activity.

20 U. Fischer et al./Fusion Engineering and Design 37 (1997) 9 21

References

[11 P.F. Rose (Ed.), ENDF-6 Summary Documentation, BNL-NCS-17541 (ENDF-201), 4th edn. US National Nu- clear Data Center, Brookhaven National Laboratory, Upton, Newton, New York, USA, October 1991.

[2] V.N. Manokin et al., in: H.D. Lemmel, P.K. McLaughlin (Eds.), BROND-2.2 Russian Evaluated Neutron Reac- tion Data Library, Summary Documentation, IAEA- NDS-90, Rev. 8, January 1994.

[3] K. Shibata et al., Japanese Evaluated Nuclear data Li- brary, Version 3, JENDL-3, JAERI-1319, 1990.

[4] J. Kopecky, H. Gruppelaar, H.A.J. van der Kamp, D. Nierop, European Fusion File Version 2, EFF-2 Final Report on Basic Data File, ECN-C-92-036, June 1992.

[5] D.W. Muir, S. Ganesan, A.B. Pashchenko, FENDL A reference nuclear data library for fusion applications, Int. Conf. Nuclear Data for Science and Technology, Juelich, Germany, May 13 17, 1991.

[6] S. Ganesan, P.K. McLaughlin, FENDL/E--Evaluated nuclear data library of neutron interaction cross-sections and photon production cross-sections and photon-atom interaction cross-sections for fusion applications, Version 1.0, IAEA-NDS-128, May 1994.

[7] FENDL/MC-1.0, Library of Continuous Energy Cross- Sections in ACE Format for Neutron-Photon Transport Calculations with the Monte Carlo N-Particle Transport Code System MCNP4A, generated by R.E. MacFarlane by processing FENDL/E-1.0, Summary Documentation by A.B. Pashchenko, H. Wienke and S. Ganesan, IAEA- NDS-169, Rev. 3, November 1995.

[8] FENDL/MG-1.0, Library of Multigroup Cross-Sections in GENDF and MATXS Format for Neutron-Photon Transport Calculations, generated by R.E. MacFarlane by processing FENDL/E-I.0, Summary Documentation by A.B. Pashchenko, H. Wienke and S. Ganesan, IAEA- NDS-129, Rev. 3, November 1995.

[9] J.F. Briesmeister (Ed.), MCNP A General Monte Carlo N-particle Transport Code, Version 4A, LA-12625-M, November 1993

[10] R.E. MacFarlane, The NJOY Nuclear Data Processing System, Version 91, Los Alamos National Laboratory, Report LA-12740-M, 1994.

[11] S. Ganesan, Preparation of Fusion Benchmarks in Elec- tronic Format for Nuclear Data Validation Studies, Sum- mary Report of the IAEA Consultants' Meeting at IAEA Headquarters Vienna, Austria, December 13 .16, 1993, Report INDC(NDS)-298, March 1994.

[12] Y. Oyama, JAERl-contributions to the FENDL benchmark validation task, Report presented to IAEA consultants' meeting on Benchmark Validation of FNEDL-1, Karlsruhe, Germany, October 17-19, 1995.

[13] C. lchihara, Contributions of the Kyoto and Osaka Uni- versities to the FENDL benchmark validation task, Re- port presented to IAEA consultants' meeting on Benchmark Validation of FENDL-I, Karlsruhe, Ger- many, October 17~-19, 1995.

[14] K. Hayashi, Y. Oyama, Test of FENDL/MG-I.0 and JENDL-3.2 by analysis of FNS Benchmark Experiment, Report presented to IAEA consultants' meeting on Benchmark Validation of FENDL-1, Karlsruhe, Ger- many, October 17-19, 1995.

[15] R. Tayama, T. Tsukiyama, K. Hayashi, H. Giese, F. Kappler, U.V. Mrllendorff, Test of Various Evaluated Beryllium Nuclear Data against the Karlsruhe Neutron Transmission Experiment, Report presented to IAEA consultants' meeting on Benchmark Validation of FENDL-1, Karlsruhe, Germany, October 17-19, 1995.

[16] M.Z. Youssef, UCLA-contribution to the FENDL benchmark validation task, Report presented to IAEA consultants; meeting on Benchmark Validation of FENDL-1, Karlsruhe, Germany, October 17 19, 1995.

[17] M.Z. Youssef, A. Kumar, M.A. Abdou, C. Konno, F. Maekawa, H. Maekawa, Benchmarking FENDL library through analysis of existing benchmark experiments Part (1): Analysis of bulk shielding experiments on large SS-316 assemblies bombarded by d-t neutrons, Report UCLA-FNT-90, ITER/US/95/IV-BL-14A, University of California, Los Angeles, December 1994.

[18] C.O. Slater, Preliminary Results for the Analysis of the SB5 and SB6 Shielding Benchmark Experiments, Oak Ridge National Laboratory, unpublished, February 1995.

[19] V. Rado, L. Petrizzi, P. Batistoni, Validation of the FENDL library using the FNG integral experiments, report presented at 1AEA consultants' meeting on Bench- mark Validation of FENDL-1, Karlsruhe, Germany, Oc- tober 17- 19, 1995.

[20] K. Seidel, TUD-contribution to the FENDL benchmark validation task, Report presented to IAEA consultants' meeting on Benchmark Validation of FENDL-1, Karl- sruhe, Germany, October 17-19, 1995.

[21] U. Fischer, FZK-contribution to the FENDL benchmark validation task, Report presented to IAEA consultants' meeting on Benchmark Validation of FENDL-I, Karl- sruhe, Germany, October 17 19, 1995.

[22] L. Benmansour, A. Santamarina, Benchmark Analyses of the EFF-I, EFF-2 and FENDL Fe Evaluations for Neu- tron Transport and Shielding Calculations, Report presented to the JEF/EFF Meeting, Paris, December 8 9, 1994.

[23] A. Blokhin, IPPE-contribution to the FENDL benchmark validation task, Report presented to IAEA consultants' meeting on Benchmark Validation of FENDL-1, Karksruhe, Germany, October 17 19, 1995.

[24] S.P. Simakov, New results of IPPE experiments on spherical iron shells, presented to 1AEA consultants' meeting on Benchmark Validation of FENDL-I, Karlsruhe, Ger- many, October 17 19, 1995.

[25] D. Markovskij, Contribution to FENDL-1 integral data test, report presented to IAEA consultants' meeting on Benchmark Validation of FENDL-I, Karlsruhe, Ger- many, October 17 19, 1995.

[26] Sub working group of fusion reactor physics subcommit- tee (Ed.), Collection of experimental data for fusion neu-


tronics benchmarks, Report JAERI-M 94-014, February 1994.

[27] C. Konno, F. Maekawa, Y. Oyama, Y. Ikeda, K. Kosako, H. Maekawa, Bulk shielding experiments on large SS-316 assemblies bombarded by d t neutrons, Vol. 1, Experiment, JAERI-Research 94~043, December 1994.

[28] C. Konno, F. Maekawa, Y. Oyama, Y. lkeda, K. Kosako, H. Maekawa, Bulk shielding experiments on large SS-316/water assembly bombarded by d t neutrons, Vol. 1, Experiment, JAERI-Research 95-017, January 1995.

[29] W.A. Rhoades, F.R. Mynett, The DOT-II! Two-Dimen- sional Discrete Ordinates Transport Code, ORNL-TM- 4280, September 1973.

[30] W.A. Rhoades, R. Childs, The DORT two-dimensional discrete ordinates transport code, Nucl. Sci. Eng. 99 (1989) 88 89,

[31] H.T. Hunter, D.T. Ingersoll, R.W, Roussin, E. Sartori, C.O. Slater, SINBAD a shielding integral benchmark archive and database for PC's, 8th Int. Conf. on Radia- tion Shielding, Arlington, USA, April 1994.

[32] P. Batistoni, M. Angelone, M. Martone et al, The bulk shield benchmark experiment at the Frascati Neutron Generator, Fus. Eng. Des. 28 (1995) 504 514.

[33] H. Freiesleben, W. Hansen, H. Klein, T. Novotny, D. Richter, R. Schwierz, K. Seidel, M. Tichy, S. Unholzer, Experimental Results of an Iron Slab Benchmark, Report TUD-PHY-94/2, TU Dresden, 1995.

[34] G. Palmiotti, M. Salvatores, Optimised Two-Dimensional S~,-Transport, Nucl. Sci. Eng. 104 (1990) 23-33.

[35] S.P. Simakov et al., Neutron Leakage Spectra from Be, AI, Fe, Ni, Pb, LiPb, Bi, U and Th Spheres with T(d, n) and 252Cf Neutron Sources, Proc. 17th Syrup. on Fusion Technology, Rome, Italy, 14 18 September 1992, pp. 1489 -- 1493.

[36] R.D. O'Dell et al., User's manual for ONEDANT: A Code Package for One-dimensional Diffusion-accelerated Neutral Particle Transport, LA-9184-M, 1982.

[37] W.W. Engle, A User's Manual for ANISN, a One-Di- mensional Discreet Ordinates Transport Code with An- isotropic Scattering, Report K-1693, Union Carbide, 1967.

[38] S.B. Marin et al, The BLANK Code for one-dimensional neutron field calculation in the model with fast neutron source, Voprosy atomnoi nauki i tekhniki. Set.: Fizika i tekhnika yadernykh reaktorov vyp. 9 (22) (1981) 26-31.

[39] U. Fischer (Ed.), Integral data Tests of the FENDL-1 Nuclear Data Library for Fusion Applications, Summary Report of the International Working Group on Experi- mental and Calculational Benchmarks on Fusion Neu- tronics for FENDL Validation, FZKA-5785, Forschungszentrum Karlsruhe, 1996, INDC (Ger) 41.

[40] P.G. young, L. Stewart, Evaluated data for N + ~Be reactions, Report LA-7932-MS, July 1979.

[41] U. Fischer, E. Wiegner, Benchmark Analyses of the ENDF/B-VI, EFF-1 and EFF-2 Beryllium Data Evalua- tions for Neutron Transport Calculations, Proc. 3rd lnt. Symp. on Fusion Nuclear Technology, Los Angeles, CA, Part B, June 26 1 July, 1994, pp. 624 635.

International benchmark tests of the FENDL—1 nuclear data library

Documents

Transcript of International benchmark tests of the FENDL—1 nuclear data library