Spallation neutron source moderator design

ELSBVIER

Nuclear Instruments and Methods in Physics Research A 411 (1998) 494-502

Spallation neutron source moderator L.A. Charlton*, OakRidge

NUCLEAR INSTRUMENTS 8 METNOOS IN PHYSICS RESEARCH SecttonA

J.M. Barnes, T.A. Gabriel,

design

J.O. Johnson

National Laboratory ‘, Computational Physics and Engineering Dirision. P.O. Box 2008. Oak Ridge, TN 378334363.

if.%4

Received 13 January 1998

Abstract This paper describes various aspects of the Spaliation Neutron Source (SNS) moderator design. Included are the effects of varying the moderator location, interaction effects between moderators, and the impact on neutron output when various reflector materials are used. Also included is a study of the neutron output from composite moderators, where it is found that a combination of liquid Hz0 and liquid Hz can produce a spectrum very similar to liquid methane (L-CHJ. CQ 1998 Elsevier Science B.V. All rights reserved. Keywords:

Spallation neutron source; Neutron; Neutronic; Moderator; Target station; Reflector

1. Introduction The proposed pulsed-neutron source Spallation Neutron Source (SNS) facility consists of a linear accelerator, an accumulator ring, and a target station. It will start operation at 1 MW; later upgrade to 5 MW is planned. The protons from the accumulator ring will be injected into the target station at 1 GeV. The subsequent spallation process will then produce low-energy thermal neutrons that may be used for a wide variety of experiments. In this paper we discuss neutronic calculations that address various aspects of the moderator design.

*Corresponding author. Tel.: + 1423 574 0628, fax: + 1 423 574 9619; e-mail: lac~ornl.gov. * Managed by Lockheed Martin Energy Research Corp. for the U.S. Department of Energy under Contract No. DE-ACOS960R22464.

The computer codes HETC [l] and MCNP [Z] were used for these calculations, with the former code performing the high-energy transport. Neutrons that fell in energy to 20 MeV or less were then passed to MCNP for further transport.

2. Moderator position study For the study of moderator position a model geometry (Fig. 1) was used in which two 12 cm x 5 cm x 15 cm cryogenic Hz moderators were placed above a 12 cm x 30 cm x 60 cm Hg target and two ambient Hz0 moderators were placed below. Following the current SNS plans, the upstream H2 moderator was decoupled and poisoned and the downstream one was coupled. Both Hz0 moderators were decoupled and poisoned. The protons (spatially distributed with an elliptic parabolic profile with major and minor radii of 6 and 30 cm)

016S-9002/98/$19.00 c:! 1998 Elsevier Science R.V. AI1 rights reserved. PII: SO168-9002(98)00312-X

g

0.008

-J

0.006

Bottomupstream moderator (Decoupled-poisoned H,O) Fig. I. Moderator location model.

entered the 12 x 30 cm2 Hg face. A Be reflector with dimensions 96 cm x 90 cm x 90 cm was used with the upstream face of the target 36 cm from the front of the 90 cm x 90 cm face of the Be reflector. For the basic model, the upstream edge of the upstream (downstr~am~ moderators were located 4 cm (25.5 cm) from the front of the Hg. Vacuum neutron channels were included to exactly accommodate the output moderator faces, which are the faces parallel to the proton beam and ~rpendi~ular to the top and bottom Hg surfaces. A 12 x 30 cm2 vacuum proton channel was excluded from the upstream part of the reflector. When the moderators were decoupled, the moderator and the neutron channels were surrounded by a 1 mm Cd sheet. The moderator was poisoned by a 50 ym Gd sheet placed in the center of the moderator, parallel to the viewed moderator face. All neutron currents presented in this section and the next were calculated by counting neutrons that had a direction of travel within 25” of normal to the output moderator face, In Fig. 2, the neutron current from the upstream moderators is shown as a function of the distance (L) from the moderator center to the front of the Hg target (see Fig. 1). The current peaks when the upstream edge of the moderator is at the upstream edge of the target (L = 6 cm). The current in the Hz0 moderator is greater because the average energy of the neutrons is higher, resulting in less Hz absorption. The current in the downstream moderators is shown in Fig. 3 as their location is varied. Current from the Hz moderator is larger because the increase due to the coupling is greater

Lb4 Fig. 3. Thermal neutron current from downstream moderators.

L&m) Fig. 4. Thermal neutron current from I-I,0 moderators

than the decreases caused by the downstream Iocation and the moderator material. As may be seen in Fig. 4, the current is very close to the same (at a given L) from an upstream moderator as from a downstream moderator if both have the same moderating material and coup~~ng~poisoni~g

L.A. Char&on er al./Nucl.

Instr. and Metk

Fig. 5. Thermal neutron current from HZ moderators.

in P&s. Res. A 41 i (1998) 494-502

Fig. 7. Neutron moderator.

current

per proton

from upstream

Hz0

wavelengths and for a Hz0 moderator are shown; however, this situation is generally true for other wavelengths and the H2 moderators.

IO4

c 1000 % 2 100 5

3. Moderator interaction

10 1..

l oL1oo ?i?kl

ib

30

i

4oc

TimecpP

Fig. 6. Thermal neutron pulse from a decoupled poisoned H2 moderator and for a coupled Hz moderator (Be reflector).

and the above separation between moderators (21.5cm). With a smaller separation this is no longer true, as will be discussed in the next section. A comparison between the two hydrogen moderators is in Fig. 5. The large difference is caused by the decrease in neutron current when decoupling and poisoning is used. This effect is much larger than the decrease because of the downstream location. The large loss of thermal current with poisoning and decoupling is tolerated because there is a sharpening of the thermal pulse, as may be seen in Fig. 6, where an order of magnitude drop takes - 50 us with decoupling and poisoning but takes - 500 us with a fully coupled moderator. The current reduction caused by moving a moderator downstream is approximately uniform with wavelength (Fig. 7). The current at all wavelengths is reduced by about the same fractional amount as the full-the~al current when the moderator location is varied. Only results for a limited number of

The dependence of the neutron current on the moderator location can be modified by the interaction between upstream and downstream moderators. Figs. 8 and 9 show the effect on the current from the upstream (downstream) moderators when they are held fixed and the downstream (upstream) moderators are moved upstream (downstream). The upstream moderators lose neutron current ( 5 10%) when they are held fixed and the downstream moderators are moved upstream. If the downstream moderators are held fixed and the

12

14

vi

18

20

22

AL(cm) Fig. 8. Interaction between liquid hydrogen moderators coupled--poisoned upstream and coupled downstream).

(de-

L.A. Charlton et cd.INucl. lnstr Upstream moderatordecoupled puiwned H,O ar I. = &xl

OL1’ 12

14

i

16

I,._.

Iis

20

22

AL(cm) Wm)

Fig. 9. Interaction between water moderators (decoupled-poisoned).

upstream moderators are moved downstream, the current in the downstream moderators increases if they are decoupled and poisoned and decreases if they are coupled. Although it is not shown, when the interaction calculations were repeated with a decoupled-poisoned downstream Hz moderator, the current in the downstream moderator increased when it was fixed and the upstream moderator was moved downstream. Thus the loss of current in the downstream HZ moderator in the SNS moderator confi~urat~on is caused by the coupling, and not the moderator material. A possible explanation for the above variation is as follows: the loss of current in the upstream moderators is caused by the loss of reflection. The current in the downstream moderators increases due to interaction effects if they are decoupled and poisoned, because the neutrons that were reflected into the upstream moderators are now available to the downstream moderators. If the downstream moderator is coupled (as is the case for the SNS Hz moderators here), the current decreases, because the upstream moderator acts as a partial forward decoupler. Interaction effects between the upstream and downstream moderators are shown in Fig. 10. The upstream moderator is decoupled/poisoned Hz0 at L = 6 cm. The current from the downstream coupled Hz moderator is shown as it is moved upstream with the upstream moderator present and with it removed. Interaction effects reduce the current in the downstream moderator but are not large

Fig. 10. Effect of moderator coupled Hz moderator output.

interaction

on

downstream

enough to offset the increase caused by moving the moderator upstream. Thus, an increase in downstream current occurs if the downstream moderator is moved upstream with or without interaction effects.

4. Reflector materials The model used for the reflector study (Fig. I 1) is similar to that described above. except only the upstream set of moderators is included and the reflector is replaced with an inner reflector (a cylinder of radius 32.5 cm and height of 90 cm) and an outer reflector (also a cylinder, but with a radius of 100 cm and a height of 200 cm). The center of the cylinders is at a point 10 cm downstream from the front of the Hg target but at the center in the other two dimensions. The volume occupied by the target moderator assembly is excluded from the inner reflector, and the volume occupied by the targetmoderator assembly and the inner reflector is excluded from the outer reflector. The composition of the reflectors is varied. All neutron currents presented in this section were calculated by counting neutrons with a direction of travel within 90” of normal to the output moderator face. In Table 1 the neutron current from decoupledpoisoned moderators is shown for various inner and outer reflector compositions. The thermal neutron currents depend very weakly on reflector

498

Hg: 10x40~60 cm3

P

Moderators: 10x12~15 cm3 Hg target and moderators

h = 200cm r = 1OOcm

h = 90cm r = 32.5cm

I

Hg target, moderators and inner reflector

Hg target, moderators, inner and outer reflector

Fig. 11. ReRector model.

Table 1 Compa~s~n: thermal neutron output for diffe~ng reflector materials and geometry for decoupled-poisoned moderators Inner Ref.

Be

Outer Ref. J(n/p)-Hz Mod. &n/p)-H,O Mod.

BE 0.0615 0.116

Inner

Be Be Pb Pb

Pb Ni 0.0615 0.117

Pb 0.0644 0.125

Ni 0.0643 0.124

Outer Be Ni Pb Ni

E(meV)

I 9.0

material or on the use of an outer reflector where the outer section of the moderator is replaced by an outer reflector. The spectra are shown in Figs. 12 and 13. They are virtually identical except at very low energy, where the Monte Carlo statistics are poorer (1 standard deviation is - 20%) than at higher neutron energies. The thermal pulses are shown in Figs. 14 and 15. The pulse peak is independent of the reflector material, and the pulses down to 10e3 of the peak are very similar. Including an outer reflector, however, causes a decrease in the pulse width for currents < low3 of the peak. In Table 2, the time-averaged neutron output from coupled moderators is shown for various reflector materials. It is less for a Pb inner reflector than for Be and less for both inner reflector materials when the outer section of the reflector is replaced with an outer reflector. The neutron output is larger when Pb is used as an outer reflector and smaller when Ni is used. Spectra for neutrons from

I

I

I 29

IL9

%A1 .90

Fig. 12. Spectra from a decoupled-poisoned differing reflector materials and geometry.

HZ moderator for

the moderator face are shown for H2 and Hz0 in Figs. 16 and 17. The spectra are about the same at the highest energies shown but are very different at lower energies. The neutron intensity at - 3 meV ( - 5 A) differs by almost a factor of 3 between a Pb-Ni (inner-outer) and a Be-Be reflector. The largest number of low-energy (rl > 1-2 A) neutrons result from a simple Be-Be reflector. The pulses that result when the various reflector materials are used are shown in Figs. 18 and 19. The peaks are the same. The reduction in thermal neutron output when Pb is used instead of Be or when a Ni or Pb outer reflector is used, comes primarily from the pulse tail, resulting in a sharper thermal pulse. As may be seen in Fig. 20, this difference in the pulse

LA. Char&on et al. /Nucl. Instr, and ,Meth. in Phy.

Res. ‘4 411 (1998) 494

499

Sft-7

Table -! Comparison: thermal neutron output for differing reflector materials and geometry for coupled moderators

/ 9.0

!

Pb

Inner Ref.

Re

Outer Ref. ~(n!t)-H~ Mod.

Re 0.414

Ni 0.267

Pb Cl“34

J(n!p)-H,O

0.466

0.344

ox

Mud.

E(meV)

L-..----

I

Ni 0.182

~-

0.272

-I 29

Fig. 13. Spectra from a decoupled-poisoned H2Q moderator for differing reRector materials and geometry.

Fig. 16. Spectra from 3 coupled Hz moderator for ditkring reflector materials and geometry (similar result for H20J.

Pb

Ni

Refkctor inner Outer Be Be -Be Pb Be Ni Pb Pb Pb Ni

Fig. 14. Thermal neutron pulse from a decoupled poisoned Hz moderator for differing reflector materials and geometry.

109

Reflector Materi& Inner Outer

8----b

Ni

Fig. 15. Thermal neutron pulse from a decoupled poisoned Hz0 moderator for differing reflector materials and geometry.

tail is caused by a selective reduction in low-energy neutrons. This finding poses an interesting question: why is the peak neutron current the same for all the differing reflector con~~urations (Figs. 18 and 19), but yet

L....--.. 9.0

1

E(meV)

‘7.9 MA,

.%I

J 29

Fig. 17. Spectra from a coupled Hz0 moderator for d;~~ri~g regector materials and geometry.

the spectra indicate that the intensity of bob-energy neutrons can differ by large factors (Fig. XI)? This question is addressed in Fig 21. where a sample of pulses for Be-Be and PbPb reflectors are shown for various wavelengths (if the pulses at all thermal wavelengths were added they would give the thermal pulses of Fig. 18, which are also plotted). At the pulse peak, the neutrons are almost completeIy high energy (such as the i. = 0.8 A

LA. Charlton et al.~‘Nuel. Inst~ and Met& in Phrx Res. A 411 (19981 494-502

500

Reflector Material. Inner Outer Be Pb Ni

1

P dI -Z 5

Pb Ni 200 300 Time(p)

400

500

Fig. 18. Thermal neutron pulse from a coupled Hz moderator for differing reflector materials and geometry.

Fig. 21. Thermal neutron pulse from a coupled Hz moderator for a Be (inner and outer) and a Pb (inner and outer) reflector.

Be reflector not shown

10

0

100

300 200 Time&s)

400

500

Fig. 19. Thermal neutron pulse from a coupled H20 moderator for differing reflector materials and geometry.

Reflector Inner Outer _L -Be Be - I - ‘Be Pb Ni -Be ---Pb Pb . .. . . . Pb Ni 0t

I

1

10

100

.,.

Fig. 22. Composite moderator model. The left hand side of both moderators is H20, and the right hand side is Hz. The location of the division between them is varied.

for a neutron experiment is an intense A = 5.0 A (for instance) current, a narrower thermal pulse is undesirable.

5. Composite moderators

1000

I

, E(meV) 1

I

9.0

2.9 k(A) 0.90

0.29

Fig. 20. Ratio of the spectra found with various reflector materials to that found when Be was used as both an inner refiector and an outer reflector, This comparison is for a coupled Hz moderator (similar result for H,O).

shown). The reduction in the tail of the thermal pulse when Pb is used as a reflector instead of Be is almost uniform with wavelength. The larger decrease in the totals for the low-energy neutrons is because a larger proportion of the low-energy neutrons are “tail” neutrons. Thus, if the top priority

The composite moderator study reported here is similar to and was suggested by the earlier Los Alamos National Laboratory (LANL) work [3]. A moderator with output peaks at an energy corresponding to the peak produced by Liquid methane (L-CH4) is frequently desirable [4]. The behavior of L-CH4, however, would be expected to be a problem at the higher anticipated SNS powers ( > 1 MW) [S]. It would thus be of use to find a material that could produce the same spectra as L-CH, but without the problems expected from L-CH4. The model geometry used (Fig. 22) is identical to that described in Section 2 except: (1) the front moderator is fixed at L = 10 cm, (2) the rear

L.A. Chariton et al. /Nucl. Instr. and Meth. in Phys. Res. A 4/l

501

(199X) 494- 502

1

-5.0 7.

0.1

T T x

/../~ __._ ,..’ _:’

_..-*

. . . .

-. -%., ‘Y.,, “...(

-5.0

cm 5.0 cm

H, H,O

cm L-CH,

0.01 i ‘.... 0.001 ~ 1

100

10

1000

-

I .2S cm Hz -3.75 cm H,O Hz side

E(meV) .29

9.0 2.9

14)

.90

Fig. 13. Spectra from composite moderators of total width 5 cm mixtures of HZ0 and H2 ns riewed,from H,O side.

for different

1

‘..

0.1 --,/,

5

-g 0.01

I

_ ____.~._“_.--&.-‘.

.=z 1

1

10

‘<

1

:

~

0.001 I

\

[.G,_T

,..

I

\

. . . . . 5.0cm

Hz

_ -

HZ0

_

--\..,

- 5.0 cm

2.5 cm Hz 2.5 cm H,O

I

100

1000

.90

29

E(meV) .A

9.0

2.9

h(A)

Fig. 15. Spectra from composite decoupled moderator> width 5 cm) fnr different mixtures of Hz0 and HL.

Table 3 Comparison: thermal neutron currents erator for various moderator materials Moderator

J(n:p)

L-H20 L-HZ L-CH, Composite”

0.0’5 O.O’? 0.0’3 0.022

“15% L-H, Fig. 24. Spectra from composite width 5 cm) for different mixtures H2 side.

from a decoupled

(total

mod-

and 75% L-HZO.

decoupled moderators (total of H20 and Hz as viercjed.fiom

moderator is not used, and (3) the front moderator is divided into two sections by a plane parallel to the moderator-neutron output face. The section on one side of the plane is filled with Hz0 (and neutrons from this side are “from the Hz0 face”). The section on the other side is filled with H2 (and neutrons from this side are “from the Hz face”). All neutron currents presented in this section were calculated by counting neutrons with a direction of travel within 25’ of normal to the moderator output face. The spectra for neutrons coming from the Hz0 face were very similar to those from a H20 moderator of varying widths. For 50% : 50% Hz0 : Hz the spectra of neutrons from the Hz0 side is shown in Fig. 23 and compared with 100% Hz0 and 100% Hz. Except at low energy, the spectra for the moderator with 50% H20: 50% Hz is uniformly less than when 100% Hz0 was used. The reduced-moderator width would be expected to preferentially affect low-energy neutrons

because of the greater moderation needed at the low energies. Those from the H, side (Fig. 24) show an admixture behavior where the spectra look like a superposition of Hz0 and H,. It is thus possible to obtain spectra that look very different from Hz0 or H2 from the H2 side but not from the Hz0 side. In Fig. 25 a spectrum from the H, face resulting from a 25% Hz : 75% Hz0 configuration is compared with spectra for H20, L-H2 and to L-CH4. As may be seen, a spectrum closely resembling L-CH4 is achieved. A comparison of the total thermal-neutron output for the same mixture of materials is shown in Table 3. The output for the composite is virtually identical to that of L-CH,. Thus. by combining materials, both of which are commonly used for moderators. one can produce a spectrum that closely resembles the spectrum for L-CH,. In the Fig. 26, the thermal pulse from a L-CH, moderator is compared with that from the 75% Hz0 : 25% Hz admixture shown in Fig. 25. The combination material gave a narrower pulse width than did the pure L-CHS moderator.

502

L.A. Charlian et al. /Nucl. Insir. and Meth. in Phys. Res. A 4i1 (1998) 494-502

Fig. 26. Thermal pulse comparison: viewed from the HZ side.

6.

L-CH,

and

HZ/HZ0

For a coupled moderator, Pb as an inner reflector material makes a substantial reduction in the neutron tail relative to Be and the use of an outer reflector causes a reduction in the tail for both inner-reflector materials. Reducing the neutron tail, however, causes a selective decrease in low-energy neutrons. A composite moderator in which 25% of the volume contains Hz and 75% of the volume contains H,O can give a spectrum very similar to L-CH4. However, the “L-CH4-like” moderator has a narrower thermal pulse than does the L-CHJ moderator.

Summary

The neutron output from each moderator is maximized if the upstream edge is placed as far toward the upstream edge of the mercury target as possible but not past the front of the mercury target. The interaction between moderators causes a small decrease in the current from the upstream moderators and a decrease in the current from the downstream moderators if these moderators are coupled, but an increase if they are decoupled and poisoned. The decrease when the downstream moderator is coupled is not su~cient however, to offset the increase from moving the moderator forward. The effect of various reflector materials depends on whether the neutron output from a decoupled or a coupled moderator is being considered. For a decoupled-poisoned moderator, the use of Pb or Be as an inner reflector material and Ni as an outer reflector makes little difference, except for a reduction in the neutron tail starting at + IO-” of the peak value.

The authors acknowledge the many useful conversations with Jack Carpenter, Kent Crawford and Erik Iverson of the Argonne National Laboratory.

References 1’1 T.A. Gabriel et al.. CALOR87: HETC87, MICAP, EGS4. and SPECT, A Code System for Analyzing Detectors for Use in High-Energy Physics ~x~riments, Proc. Workshop on Detector Simulation for the SSC, ANL, August 24-28, 1987. VI J.F. Briesmeister (Ed.), MCNP ~ A general purpose Monte Carlo code for neutron and photon transport, LASL Report LA-7396-M, Rev. 2. September 1986. c31G.D. Baker, A, Bridge. T.O. Brun. L.L. Daeman, E.J. Pitcher, G.J. Russell. Moderator neutronics design for a next-generation (1 MW) pulsed spallation neutron source, LANL, private communication. t-41SNS Users Workshop, Oak Ridge, TN, October 31-November 1 1996. c51A.T. Lucas, Private communication. Oak Ridge National Laboratory. 1997.