NEACRP-A-1052 Topic 1.4 E

The cold moderator for the continuous SINQ

F. Atchison, G. Bauer, W. Fischer, K. Skala, H. Spitter

PAUL SCHERRER INSTITUTE CH-5232 Villigen-PSI,

Abstract

The paper describes the boundary conditions and the resulting design concept of the cold D+uoderator for the continuous high power spallatidn neutron source SINQ. Initial thermal-hydraulic calculations performed have shown that the system can be operated safely under the expected load conditions and that also accidental situations such as loss of insulating vacuum do not.cause serious safety hazards.

Paper presented at the International Workshop on Cold Neutron Sources March 5 - 8, 1990 Las Alamos NW, USA 17110001 *. .

I. INTRODUCTION

SINQ is a spallation neutron source under construction at the Paul-Scherrer- Institute in Switzerland. It will be driven by the proton beam of a high current isochronous with a radio-frequency of 51 MHz. This means that SIN&, like most research reactors, will be a cw-neutron source in contrast to most other spallation neutron sources in operation world wide. This has several consequences on the design of the source itself and of its cold .

II. THE SINQ GENERAL CONCEPT AND BOUNDARY CONDITIONS FOR THE COLD MODERATOR DESIGN

Given the fact that SINQ will be a cw-neutron source, the decision was made to make 0 it resemble as closely as possible a in terms of oppor- tunities. This means that the prime optimisation criterion was a high thermal and cold and the maximum useful number of horizontal beam ports. This lead to four significant decisions:

1. A large DsO-tank should be used as neutron moderator 2. A vertical beam injection from underneath into the target was selected to max- imise the useful space around the target shield 3. A target of low absorption cross section for thermal was favoured (liquid Pb-Bi eutectic mixture) 4. Much emphasis was placed on an efficient cold moderator and associated neutron guides.

A vertical cut through the SINQ target station along the direction of the incident proton 0 beam on the left hand side and along the direction of the main axis of the neutron guide system on the right is shown in Fig. la. Fig. lb shows a somewhat enlarged cut through the plane containing the cold moderator inserts.

1 17110002 Fig. 2 gives a horizontal section through thr target block .I5 Iowing XII bca~~n port~s ~vhich in fact are on three different levels as indicated by thr corresponding ~~umbcrs.

The neutron production target in the center of the target block will lx cooled I>y nn~tnr;rl convection of the Pb-Bi eutectic and hence has a height of almost 4 m to provide enough buoyancy force for the hot liquid. As a consequence, the target brcomes wry massive and gives good self-shielding for the high energy neutron component in the forward dirrction of the proton beam. This can be seen from Fig. 3 which gives a p&r plot of the nrutrou intensity in various energy regimes leaving the target. It can also bc seen from this plot, that it significant component of high energy neutrons is present which leads to a, biologicnl shield much thicker than e.g. with a reactor. This should be as compact as possible. This has two important consequences for the choice of cold moderator concept:

1. Shielding in this direction should remain a,s compact a.s possible

2. Significant heating rates and activation of structural cwuponcnts will pc~-sis~. out to quite large distances from thr center of the SOIIKC.

Design provisions have been made on the SINQ D,O- modrrntclr lank to 1lil~VC 110111, a II2 and a Dz-cold moderator availnblc as a long term opt.iou. Ilrsults of calculations performed for the spectral intensity available at the outer edgr of the rnn,i~l shirltl for thr three types of moderators at their projected positions (no effect of neutron guidrs incllldcd) are shown in Fig. 4. In view of these findings, the final decision on the 112-cold ~nodcrnto~ was delayed until operational experience on the source itself and instruments Iocn.~.ctl at its beam tubes are available.

2 Service

Target bulk neutron target hall

w

maintenance room

I \ Target catcher o 5m IO I L- - I I

Figure 1: a) Vertical section through the S’INQ target station. Section is along the proton I injection line on the left and lower half of the right hand side. Upper half of the right side is a cut along the main axis of the neutron guide system. ,

Figure 1: b) Section through part of the SlNQ target block along the plane of the coId moderator inserts. For detailed parts designation see Figs. 721) and i’b).

4 Subthermal neutron 1 _

Cooled iron shielding,

Cold neutron beam ports

Outer concrete shielding layer

,. Massive iron shieldin

Thermal neutron --=-!f+Y +btherq.a; neutron beam ports Y oeam po Plug position for 0, cold source

Figure 2: Horizontal section through the SIiVQ target block showing aJJ horizontal pene- 0 trations, although on different levels (levels above Aoor indicated). 0 0 or Figure 3: Angular distribution ofneutrons in different energy bins emerging from the SINQ Pb-Bi liquid metal target. Note the forward self shielding effect!

III. THE GENERAL DESIGN OF THE SINQ COLD D,-MODERATOR

Given the high energy neutron distribution shown in Fig. 3, it was decided not to place any components of the cold moderator system other than the moderator vessel itself into areas of high radiation fields of fast neutrons. This means that a horizontal insertion port for 0 the cold moderator was foreseen and the heat exchanger of the natural convection system was moved back into a well shielded region. This results in fairly long tubes between the moderator vessel (main heat source) and the heat exchanger (heat sink) and hence requires a large vertical separation of the two to overcome the frictional forces in the tube system by the buoyancy force of the liquid-gas mixture flowing back from the moderator vessel.

Experiments carried out by Hoffmann 111 in conjunction with the development of the second cold moderator at the ILL-reactor 121 showed that such a system could be operated over a large range of heat input. In our case we decided to place the vertical leg partway to the outside of the shielding block for three reasons:

1. The need for shielding behind the horizontal leg would have blocked space valu- able for neutron scattering experiments

2. Placing the heat exchanger in the shielding block automatically provides good mechanical protection

6 17170007 . .

log- 1 I I I I 1 I I 1 _ .a Beam Tubes 6m long Cross Section 15x8cm2 -

Neutron Wavelength A CA1

Figure 4: Anticipated neutron current at the nominal monochromator positions (6m from the start of the beam tube) for the different moderators of SINQ. The effect of neutron guides is not included.

3. The length of the tube and hence the amount of liquid Ds as well as the frictional forces could be reduced.

This design implies that the horizontal and vertical leg of the cold source cannot Abe installed as one unit, but have to be put in place separately with a connection made before the 0 shielding plug behind the horizontal leg can be mounted. Cryogenic equipment (cold box) and the control system for the source will be placed next to the main shield at a level of 9.5m above ground on top of the neutron guides with a connecting channel to the service areas and to the Ds-storage tank and parts of the gas handling system, located outside the neutron target hall on the roof of the equipment access building.

A block diagram of the various sub-systems is shown in Fig. 5.

IV. THE COLD MODERATOR SYSTEM

The design of the cold moderator part was governed by the neutronics considerations published elsewhere 131. In order to avoid rethermalizationof cold neutrons upon extraction from the moderator, the cold moderator insertion port will be connected to the neutron extraction port resulting in a T-shaped structure inside the moderator tank, Fig. 6.

19110008 Chimney Chimney /////I 4

2Om3. 2.8 bar

tleuterium supply system vscuum supply system Nz-bsrrler' cryogenic---+a--- plant 3OOOW. 19K --i>ot

-! Nfll l4!

DC compressor

OB Burst disc

Y".DM.HM Gas detector

Moderatorts

--.-._____

Helium atmosphere Nitrogen atmosphere

2 C 0 Figure 6: Layout of the SINQ-DZO-Tank with T-shaped structures for cold moderators 3 insertion and neutron extraction to avoid rethermalising D20 layers. c? The mechanical stability of this structure under thermal loads, buoyancy forces and various differential pressure conditions is presently under investigation by computational methods.

The horizontal cold moderator insert is completely independent of this T-structure and has the following main sub-units: (see Fig. ?‘a)

l a double-walled vacuum jacket with a helium barrier between the outer and the inner tube. The part of the vacuum jacket intruding into the DzO-tank consists of an outer 2 mm thick AlMg,-tube and an inner zircalloy tube, 3 to 4 mm thick which is the pressure safety tube and will withstand an internal pressure of 30 bar.

. The source plug, comprising the moderator vessel, the concentric transfer tubes for liquid deuterium (inner tube) and liquid-gas mixture (outer tube) of the heat removal system and a protective support tube which changes to a massive shielding sleeve outside the DzO tank. 0

l The shielding and reflector plug comprising a D,O reflector volume to be located behind the Dz moderator to minimize losses, its associated DzO circulation tubes and the cylindrical central shielding plug of the insert. Fig. 6 also shows the Dz moderator vessel in the T-shaped tube structure and the D,O reflector volume which makes up for most of the reflector displacement by the cold source insertion tube.

The moderator vessel itself will be made of 99.5 % pure aluminium, semi-hard with one welding seam around its cylindrical surface at the far end with respect to the target. Material softening by the weld requires that the wall thickness of the vessel be increased from 3 to 5 mm in this region to obtain the desired strength to resist an internal pressure of 30 bar. Fig. 8 shows a vertical cut through the moderator vessel and its concentric Dz-flow tubes. Some relevant data are given in table 1. The vertical leg of the cold moderator system is shown in Fig. 7b. It will be inserted from the top and connections to the horizontal leg will be made inside the connector box where the two inserts meet. This box will afterwards be sealed to connect the insulating vacua of the two inserts. The second flange on the connector box shown in Figs. 7a and 7b completes the protective helium barrier (see below).

The main components of the vertical insert are:

l the vacuum jacket surrounding all cold parts for thermal insulation

l the D2/He heat exchanger (plate heat exchanger with interconnectors) to remove the total of 2.65 kW of heat load from the Dz (capacity 3 to 3.5 kW).

l the phase separator located beneath the heat exchanger to remove the liquid from the mixture coming up from the moderator vessel.

l the concentric Dz-transfer tubes ending in the phase separator in such a way that liquid Dz is flowing downwards in the inner tube and the liquid-gas mixture comes up in the outer tube to a higher level for phase separation

IO 7711uo11 vacuum jacket D2-transfer pipes \ helium barrier D20-pipe \ - !A-

.-

thimble' (double walled with He-barrier) ! ! I H20-reflector !

inner , radiation mode;ator shielding and shield insert shielding reflector plug

Figure 7: a) Vertical section through the horizontal insert of the SINQ cold Dz moderator [pipe system, phase separator and heat exchanger) mounting flange helium barrier .- ~_ for feedthroughs cold helium out

cold helium in heat exchanger

level indicator

020 for reflector

D,-transfer pipes (concentric)

vacuum jacket

expansion bellow

Figure 7: b) Vertical section through the vertical insert of the SINQ cold Dz moderator (pipe system, phase separator and heat exchanger) Figure 8: Vertical section through the Dz moderator vessel

. the vacuum dome above the heat exchanger where the cold He-transfer tubes and the Dz (;as tube penetrate the cold moderator insulating vacuum space.

V. THE PROTECTIVE BARRIER SYSTEM

The protective barrier system is designed to avoid any possible contact between deuterium and oxygen from the air which could ultimately form an explosive gas mixture. Although the SINQ beam ports are filled with helium which also surrounds the horizontal insert of the cold moderator, it was decided to have a separate additional helium system around the cold part of the Dz circuit for the following reasons:

. the pressure in the beam-tube gas system is below atmospheric pressure to make sure that in case of a leak the flow always is to the inside to avoid spreading a possible radioactive contamination into the atmosphere. By contrast, the pressure in the protective He-barrier of the cold source should be above atmospheric pressure (1.1 bar) to avoid contamination with oxygen in case of a leak to the outer atmosphere

13 79770074 . the cold moderator helium barrier will be continuously monitored for oxygen con- tamination which suggests having as small a volume as possible

9 in case of the need for access to the structural parts of the cold moderator system, the beam port helium system needs not be opened.

The helium gas in the protective barrier will be continuously circulated, carefully monitored for pressure loss and contamination by other gases and refrigerated to remove the heat it picks up from the walls in the region of high neutron flux. Outside the main shield all system parts containing warm D,-gas will be surrounded by a Nz-gas barrier rather than helium for the following reasons:

. Nz is much cheaper than helium 0 . Helium is used for leak detection on experimental facilities and thus should be present in the smallest possible amounts in the air

. monitoring for deuterium is easier in nitrogen than in helium

The pressure in the Nz protective barrier will also be kept at 1.1 bar to avoid oxygen-intake in case of a leak.

VI. THE COLD HELIUM SYSTEM

Refrigeration of the D, in the heat exchanger is accomplished by cold helium gas from an existing cold box whose capacity is being upgraded to 3.1 kW. With a total of 2.65 kW to be removed from the Dz-source, this offers some reserve. The helium will be supplied to the heat exchanger at 19 K and 7.5 bar and will return at 25 K, 7 bar. The cold box will operate at constant cooling power; at reduced source power heat will be provided to the system by auxilliary heaters in the helium circuit. The cold helium transfer line between 0 the cold box and the heat exchanger will be 20 m long each way and will be in a channel inside the target shielding block. Its outer diameter will be 90 100 mm.

VII. THERMAL-HYDRAULICS AND ACCIDENT CONSIDERATIONS

In order to be able to predict the behaviour of the cold moderator system under different operating conditions and in case of abnormal situations, a computer model of the Dz circuit was established. The model treats the three vessels of the system as “localized capacitors” where phase changes can take place and the mass content can vary. Connecting pipes are treated as “continua” which are subdivided into individual elements. The corresponding coupled partial differential equations are replaced by difference equations and applied to the finite elements. Coexistence of vapor and liquid is allowed in all volumes of the cold part of the system, whereas only vapour exists in the storage tank and the pipe connecting it to the heat exchanger. Dynamical equilibria in mass, energy and forces are established

14 17110015 , .

for all elements and the resulting differential equations are integrated numerically. The compressibility of the vapour as well as of the liquid are taken into account in the whole system.

To date, heat input has been allowed for in the moderator vessel and in the phase separator- heat exchanger volume. An extension to the case where heat input is possible also along the connecting pipes is under preparation.

In order to benchmark the model and to obtain realistic values for some poorly known parameters, the experimental data given by Hoffmann 111 for the ILL-test experiments were verified in a first step.

Based on these results a preliminary series of calculational runs was performed to test the behaviour of our planned system. It could be shown that, when starting from an arbitrary 0 non-consistant set of operating parameters (flow velocity in the pipes, heat input at the moderator vessel, liquid-to-vapor ratio etc.), the system moves towards a stable set of values within a few seconds. The operating parameters obtained for the stationary flow conditions are listed in Table 2. These calculations will be continued to obtain a complete picture of the way the system reacts under changing operating conditions.

As a worst case assumption, complete instantaneous loss of the insulating vacuum was considered with a resulting step function like heat input of 30 kW to the moderator vessel and 10 kW to the heat exchanger. Heat input to the pipes has so far been neglected but will be taken into account in the next round, together with a more realistic rise function of the heat input. However, even with these very pessimistic assumptions, the results obtained for the maximum pressure in various parts of the system were very reconciling.

It should be noted that, as a starting situation, both pipes connecting the moderator vessel to the heat exchanger were assumed to be completely filled with liquid at rest (i.e. the spallation source was not operating)., Hence, as a consequence of the heat input to the l moderator vessel and the resulting vaporization the liquid had to be expelled from the pipes into the heat exchanger where it vaporizes. The resulting pressure as a function of time in the three “capacitor” volumes is shown in Fig. 9. Under the very pessimistic assumptions made here, the pressure in the moderator vessel reaches the design pressure of 3.6 bar for a very short period of time. This shows that, under slightly more realistic assumptions the design pressure of the moderator vessel will not be reached even in case of a very rapid loss of the insulating vacuum. The peak is caused by inertia of the liquid in the connecting pipe and has practically disappeared as soon as the flow velocity of the medium reaches its maximum value at the pipe exit into the phase separator. Fig. 10 shows the flow velocities of the medium (liquid resp. vapour) at the exit point of the moderator vessel and at the entrance point of the storage tank. The delay caused by compression of the medium inside the system (sound velocity) is clearly seen. The diameters of the two pipes and the temperature of the medium are significantly different. Therefore the absolute values of the velocities are not comparable.

15 17110016 - moderator cell ---- heat exchanger/phase separator 1.6 --- entrance to storage tank

1.58

1.56

1.52

1.5

0 0.2 0.4 0.6 0.8 1 Time (s) Figure 9: Time dependence of the internal pressure at three different positions in the Dz-system as a result of a sudden heat influx of 30 kW at the moderator vessel and 1~0 kW at the phase separator. The enlarged scale at the right refers to curves with arrows pointing to the right.

VIII. CONCLUSION

Although still of a somewhat preliminary nature and not yet complete, the calculations performed so far for the design of the SINQ cold moderator system show that the system can easily cope with the expected heat load and is also safe against accidential loss of the insulating vacuum. Nevertheless, in a next step, rupture of the moderator vessel will be considered and the resulting pressure buildup in the vacuum space will be investigated.

IX. ACKNOWLEDGEMENTS

We would like to thank Dr. K. Gobrecht, Dr. H. Hoffmann, C. Doose and E. Krahling for valuable advice during the conceptual design phase and A. Hijchli and J. Ulrich for help with the design and supporting interest in the calculation.

16 ~7110017 ” c

exit of moderator vessel (D,tt=39md entrance to storage tank (D =90mm)

Y t t I I 0.2 0. 4 0.6 0.8 1 Time (s) Figure 10: Time dependence of mass ffow velocity at the exit of the moderator vessel and at the entrance to the storage tank as a result of a sudden heat influx of 30 kW at the moderator vessel and IO kW at the phase separator. The time delay is clearly visible. The connecting pipes were assumed full of liquid at t = 0.

X. REFERENCES

1. H. Hoffmann: “Natural convection of a cold neutron scurce with vaporieing deu- terium at temperatures of 25 K”, NATO, Advanced Study Institute: Natural Convention, Fundamentals und Applications, July 16-27, 1984 Izmir Turkey 2. K. Gobrecht, this conference 3. F. Atchison, W. Bucher, A. Hiichli, I. Horwarth and L. NordstrGm “The Dz cold- neutron source for SINQ” in “Advanced Neutron Sources 1988”, Institute of Phy- sics Conference Series No. 97, D. K. Hyer, ed. Bristol and New York 1989, pp. 473 - 482

17 17110018 Total volume 22.3 1 Volume filled with liquid Dz - 20 1 Total mass of Al 4120 Total weight of filled vessel -9 ki Design internal pressure 3.6 bar Max. operating pressure 2.8 bar

Heat deposition at 1 mA proton current: front cover 173 W cylindrical part 67 w back cover 15 W deuterium 1250 W Total 1505 W Max. surface heat flow 0.2 w cm*

Table 1: Some relevant data for the Dz-vessel of the SINQ cold moderator

Heat load on the moderator + vessel (1.6 mA) 2 kW Total mass of Dz in the moderator vessel 2.9 kg Mass flow in and out of the moderator vessel 40 id” Flow velocity of the liquid gas mixture (forward) 0.2 m/s Flow velocity of the liquid Dz (return) 0.9 m/s Mass fraction of vapon in the moderator volume 0.2 % Mass fraction of vapour in the forward flow tube 17.1 % Pressure in the moderator vessel 1.6 bar 0 Pressure in the heat exchanger 1.5 bar

Table 2: Stationary-state operating conditions of the SINQ cold source

18 17110019