Investigation of the Void Coefficient and Other Integral Parameters in the Proteus-Lwhcr 0 Phase Ii Programme

NEACRP-A- 246

INVESTIGATION OF THE VOID COEFFICIENT AND OTHER INTEGRAL PARAMETERS IN THE PROTEUS-LWHCR 0 PHASE II PROGRAMME

R. SEILER, R. CHAWLA, K. GMiiFl, H. HAGER Swiss Federal Institute for Reactor Research 5303 Wiirenlingen, Switzerland

H. D. BERGER Kraftwerk Union 8520 Erlangen, FRG

R. BijHME 0 Kernforschungszentrum Karlsruhe 7500 Karlsruhe, FRG 0 f

ABSTRACT

Comparisons of calculated and measured neutron balance components are reported for the 7.5% fissile-Pu reference test lattice of the PROTEUS-LWHCR Phase II programme, both wet (with HzO) and dry (1007 o void). Special experimental techniques have been developed and applied, particularly for k,, and the range of directly measured reaction rate ratios has been extended. For the two cell codes tested, viz. WIMS-D/l981 library and KARBUS/KEDAK-4, specific shortcomings have been identijied - the new measurements being found to be signijicantly more representative and accurate than the earlier Phase I experiments. The k, void coeficient for the Phase II reference lattice between 0 and 100% 0 void has been fomd to be qualitatively different from those assessed for the earlier Phase I test lattices. Consideration of the individual void coefficient components show this to be largely a consequence of the mope LWHCR-representative fuel rod diameter and plutonium isotopic composition of the fuel cwwntly being used. Results of control rod studies conducted for the Phase II reference lattice - both wet and dry - serve to illustrate the efforts being made towards the investigations of special power reactor features.

0 l

1 1 INTRODUCTION

The light water high converter reactor (LWHCR) concept aims at improved fuel utilization in LWRs through an increase in the conversion ratio to values he- tween 0.8 and 1.0 (cf. 0.5 to 0.6 in current-day plants). The high conver:;ion ratio is achieved with an undermoderated lattice, characterised by a volumet.- ric fuel-to-moderator ratio in the range 1.0 to 2.0 and the use of U/Pu mixed oxide fuel with a relatively high fissile content. The neutron energy spectum in such tight-pitch LWR lattices is shifted into the epithermal energy range and standard reactor codes, qualified for either thermal or fast reactor spectra, o:ften yield discrepant results when applied to the physics analysis of LWHCR cores 0 under normal and accident conditions. Thus, for example, a key question related to the technical feasability of a given LWHCR design is the sign of the moderator void reactivity coefficient. Larger fuel-to-moderator ratios and higher fissile-Pu enrichments result in less negative and, ultimately, positive void coefficients. Cal- culated accuracies for predicting such parameters and effects, however, are still not adequate. Integral neutron physics measurements relevant to homogenous-design LWHCRs were first carried out during 1981-82 at the PROTEUS zero-power reactor facility at Wiirenlingen, and a new, more comprehensive programme of experiments is currently under way. During the Phase I programme, two test lattice configurations with an average fissile-Pu enrichment of 6% (Cores 1-3) and 8% (Cores 4.6), respectively, were investigated for three different HzO-voidage states, viz. a 0%, 42.5% and IOO% void. The moderators used were HzO, Dowtherm (any or- l ganic liquid) and air (dry lattice), respectively. Each of the test lattices was constructed using two different fuel rod types, viz. 12% fissile-Pu mixed oxide and depleted UOs, the plut,onium in the mixed-oxide corresponding to that from Magnox reactors. In the Phase II programme, which is now in progress, the same three moderator states are being investigated in a single rod lattice of 7.5% fissile-Pu enrichment (Cores 7-9). The isotopic composition of the plutonium COT- responds to that discharged from LWRs, and the fuel rod diameter is also more LWHCR-representative.

2 In all the PROTEUS test lattices investigated to date, integral reaction rate ratios and other k,-related parameters were measured. Other types of experiments, such as control rod studies, have been conducted for the first time in the ongoing Phase II programme. Another important aspect of the current mneasure- xnents is the improvement of the experimental accuracies that were achieved for the neutron balance investigations in Phase I. In this context, new experimental techniques have been developed and applied - mainly for the determination of km, but also in extending the earlier range of measured reaction rates. The present paper describes the experimental procedures, particularly those which have been newly developed, and presents comparisons of calculational and exper- l imental results for the PROTEUS-LWHCR Phase II reference lattice with Hz0 (Core 7) and air (dry, Core 8). The consistency with respect to corresponding results for Phase I, where applicable, is also discussed. Two independent calculational routes have been applied in the various analyses, viz. the U.K. lattice code, WIMS-D, and the German code system, KARBUS.

2 The PROTEUS-LWHCR CORES

The PROTEUS reactor (Fig.1) is a coupled system, in which a central test zone is driven critical by thermal driver zones. A natural uranium metal buffer, located between the test and driver regions, largely reduces the effects of the mismatch between the thermal driver spectrum and the intermediate or fast spectrum in the 0 test zone. Details of the reactor, as well as the LWHCR test zone configurations l and experimental results of the Phase I programme, we given in Refs. 1 and 2. The test zone in the Phase II programme has a diameter of 50 cm and contains 1900 fuel rods (d=8.46 nun), canned in steel tubes of 9.57 mm ad. The rods are arranged hexagonally with a pitch of 10.7 mm (Fig.2), resulting in a fuel- to-moderator volumetric ratio of 2.07 for the reference lattice. LWR-discharged plutonium was used in the fuel rod fabrication, the plutonium isotopic composition being ~1% ‘=Pu, 64% ‘=‘Pu, 23% =“Pu, 8% 241Pu and 4% 242Pu. The 7.5% fissile-Pu enrichment of the mixed oxide thus corresponded to about 10.5% total-Pu. As in Phase I, the plutonium containing part of the test zone extends

3 axially over 84 cm which, together with the top and bottom blanket regions of depleted UOz, results in a total test-zone height of 140 cm. The LWHCR test lattice is located in a cylindrical steel tank, containing grid plates at appropriate levels in the axial direction. Segmented steel rods are used for filling the ou.ter- most radial positions of the grid plates, so that the quantity of ‘extra’ moderator at the outer edge of the test zone is minim&d.

3 EXPERIMENTAL TECHNIQUES AND ANAL- YSIS

3.1 Reaction Rates

The measurement of core-cater reaction rate ratios provides a valuable check on individual components of the neutron balance in the test lattice. Experimental results, corresponding to two different HZO-voidage states, can be combined to yield diagnostic information regarding the reliability of void coefficient predictions. The reaction rates measured in PROTEUS-LWHCR Cores 7 and 8 (H[zO- moderated and dry, respectively) were captures in 238U (Cg) and 242Pu (C,), and fissions in 235U (Fs), 238U (Fa), 239Pu (Fg) and z4’Pu (Fl). Toget,her these accounted for 67% and 86% of calculated neutron absorptions in the wet and dry cases, respectively. Experimental evidence for the non-measurable reaction rates was obtained indirectly through the k, measurements. Central fission rates were measured using demountahle parallel-plate fission chambers, containing absolutely calibrated fissile deposits and intercalibrated fissile foils. The chamber was located in a cavity near the core center. By compaaing the fission-product r-activity of foils irradiated in the fission chamber and in fuel rods at, the core center, count rates obtained from the calibrated chamber could be converted to absolute fission rates in the fuel. Due to the problematic manufacture and handling of foils containing z’nPu in sufllcent quantity, the catcher foil technique [3] was applied for relative F’~ measurements. The foil arrangement, placed between fuel pellets, consisted of a 0.015 mm-thick Al catcher foil of high purity placed in contact with a 1 &an’ thick 241Pu layer deposited on a 0.1 mm-thick Al backing. The activity of fi:ision

4 products that have emerged from the fissile deposits and embedded themselves~ in the catcher foil was measured. In parallel to this activation technique, three intercalibrated miniature fission chambers containing ‘ssPu, 241Puand 235U ,respectively, were utilised. The chambers, having an outer diameter of 4 mm, were slipped into empty fuel cans which replaced fuel rods near the reactor center. Calculated correction factors served to convert the measured reaction rate ratio into the fuel-averaged fission rate ratio. To test this measurement technique and the applied correction factors, Fs/Fs was determined both by the standard foil- activation/demountable-fission-chamber method, as well as by using miniature fission chambers. The results agreed well within the experimental errors. In the 0 dry Core 8 lattice, the catcher foil method for Fi was not efficient due to the (n,a) activation of the aluminhun and t.he miniature fission chambers yielded the lncre accurate result. Cs was measured by activating r3sIJ foils in the fuel rods and counting the 278 keV y-activity of the activation product essNp. The detector system was calibrated with 243Am sourcesfor absolute capture measurements, and a thermal comparison technique was also applied in parallel. r4rPu deposited as a layer of 0.2 mg/cmr thickness on 0.1 mm-thick Al was used for Ce determinations. The deposits were simultaneously irradiated between fuel pellets in the core center and in a demountable fission chamber in the PROTEUS thermal column. The 84 keV y-ray peak of r4sPu was assayed, 0 and the core-center Ce value was obtained via the known r4ePu thermal capture cross-section. a Experimental errors on the standard types of reaction rate ratios measured in Cores 7 and 8 were between k1.5 and &2.0% (la), as compared to &2.0 to f3.0% in the Phase I experiments. For Fi and Ca, as measured with the specially developed techniques in Phase II, the errors were larger than for the standard reaction rates, viz. typically 53%. The calculated correction factors for converting core-center reaction rate ratios in PROTEUS to the corresponding fundamental mode values [1],[2] were significantly smaller in Phase II, due to the larger test sons diameter. The effects of the outer reactor regions were typically less than 1% in both Cores 7 and 8.

5 3.2 k, k, is defined here as the ratio of neutron productions to neutron absorptions in the fundamental mode spectrum. In the wet 6% fissile-Pu lattices of Phase I, k, was determined by the buckling method. This method is not well suited for dry test zonea in PROTEUS and, accordingly, an experimental k, for the dry lattice was deduced from null- reactivity k, measurements in a similar core of the PROTEUS gas-cooled fast reactor programme. In the case of the 8% fissile-Pu lattices, k, was not measured due to the small test zone size. In the test lattices of the Phase II programme, alternative reactivity--based methods have been developed and applied to assess k,. Thus, the reactivity worth of a unit lattice cell was determined and independent norm&z&ion procedures applied using a calibrated “‘Cf source, as well as reactivity/reaction rate measurements with a series of reference absorbers. In the dry core, for which the experimental analysis has been completed, the two normslization routes yielded satisfactory agreement within the respective la-errors of about 51% in k,. The buckling method was also applied in both the HzO-moderated and dry Phase II test lattices, with la-errors of about &0.5% and +2%, respectively, on the deduced k, values.

3.2.1 Buckling Method

Applying the approach developed for the wet lattices in the PROTEUS-LWHCR Phase I programme [l], k, was deduced in Cores 7 and 8 from measurements of material buckling B& and calculated values for the migration area M’:

k,=l+B,f,M2 (1)

By measuring axial and radial reaction rate traverses across the test zone, the corresponding components of the material buckling can be obtained. For the axial traverses, miniature fission chambers were used, whereas the radial t.raverses were measured by foil activation. In both cases, the traverses were measured through the central part of the test zone, in which the fundamental mode spectrum is

6 .

expected to be nearly established. The measured reaction rate distributions were fitted to cosine and Bessel functions, so that the axial (a&) and radial (/?k) components of the material buckling could be determined, giving

In moderated cores in PROTEUS, the radial reaction rate traverses fall into two groups namely eV-reactions (e.g. Fs, Ce) and MeV-reactions (e.g. Fe, Rh(n,n’)), each with a characteristic curvature, Pf, where the subscript i denotes a particular reaction rate type. This results from effects of the outer reactor re- 0 gions in PROTEUS which, of course, are energy dependent. From whole-reactor calculations for PROTEUS, an appropriate correction A@ for each reaction rate type can be estimated and experimental values for the radial material buckling component deduced as:

&=&AD; (3)

In the dry test zone, the equilibrium spectrum at the cater is hard, and the effects of low-energy neutrons from the outer reactor regions on radial traverses are more complex, resulting in calculated corrections (A@) which are specific to each reaction rate type measured. Moreover, because of the larger mean free path l of the neutrons in the dry test lattice,the central region in which an equilibrium spectrum can be assumed established is significantly smaller. These effects cause a the sensitivity of the calculated correction to errors in the whole-reactor modeling to be considerably greater Table I shows the results for radial material buckling values deduced in Core 7 (Hz0 moderated) and Core 8 (dry). Th e experimental errors indicated represent statistical errors in measuring the reaction rate traverses. Uncertainties in the calculated corrections, based on a 1-D transport-theory model of the PROTEUS reactor, were estimated as about f0.5 me2 in Core 7 and &l rnT2 in Core 8. It is reassuring, of course, that for both test lattices consistent p& values could be deduced from the different types of reaction rate traverses measured

7 Whereas the radial traverses had to be corrected significantly for the influence of the outer reactor regions, such effects in the axial dir&ion were negligibly small. Small corrections to the experimental axial traverses, however, were necessary to take into account the influence of the grid plates. Further, slight a;-variations radially across the central part of the test zone had to be explicitly taken into account. With the material buckling B,!,, thus experimentally deduced (Equation 2), calculated values of the migration area in the two lattices were obtained !from appropriate cell calculations. An error of less than f5% was estimated for the calculated @-value in each case. The migration area was about a factor of 4 greater in the case of the dry core. A given absolute error in Bi thus results in a correspondingly greater k, error for the dry lattice. This, of course, is an important consideration which weakens the applicability of the buckling method to dry cores.

3.2.2 Cell Reactivity Method

Alternative methods for the assessment of Ic, have been developed and ap:plied in the Phase II pmgramme, particularly for application in dry test lattices. These methods consider the reactivity worth of a unit lattice cell at the center of the test zone and the adjoint-weighted ratio of neutron productions in the cell to those in the whole reactor. The quantity that can be experimentally deduced in this manner is kf , viz. the adjoint-weighted ratio of neutron productions to neutron removals at the center of the test zone. Calculated correction factors are used to convert k+ to k,, these factors having been estimated as 1.002 in Core 7 and I.036 in Core 8. The basic measurement was a PCTR type determination of the central cell reactivity worth -141. A cent.ral void was created by removal of fuel and cladding in the dry test lattice (Core 8), and by the replacement of appropriate amounts of fuel, clad and moderator by empty hexagonal boxes in the wet lattice (Core 7). The ‘void-boxes’ (Fig.3) were fabric&d from epoxy resin reinforced by boron- free glass fiber tissue. Reactivity changes due to such replacement of unit cells by

8 void were infered from the movement of a compensating autorod (a sphenoidal copper blade with an almost linear reactivity characteristic, located outside the thermal driver zones), which kept the reactor power constant during the course of the experiment. In Core 7, the true reactivity worth of an ideal central void pc at the center of the test zone was deduced from the reactivity changes measured with void boxes having different volume and surface area, and from supplementary measurements involving the reactivity effects of water gaps and the epoxy resin used for manu- facturing the boxes. Additional corrections, in both Core 7 and 8, were required to take mto account changes of geometry which occured above the test zone in 0 the process of creating the central voided volumes.

Normalization using a calibrated z5zCf neutron source

For the normalisation of the central cell worth, the greater emphasis was placed on measuring the reactivity worth of fission neutrons emitted by a calibrated ‘-“Cf source at the core center. This measurement was carried out at different neutron flux levels, the absolute fission rate in the surrounding lattice cells being determined by foil activation and fission chamber measurements for each flux level. Analysis of the experiments followed the route described by Redman and

Bretscher -[5] and by Chaudat et al. -[6]. Accordingly, one may express the ratio 0 of the reactivity worth of a central unit cell to that of the neutron source as:

(4)

where $+, 4 are the adjoint and direct neutron fluxes at the center of the test region, P is the production operator, H is the removal operator corresponding to capture, fission and scattering, and S is the source strength operator for the e6%f fission- neutron source. If the central fission rate Rf and the absolute somce strength S are written as

9 Rf = I,;$ (5) ( 1

the combination of the measured quantities yields

where lz+ = (r$+, P$)/(b+, H4) is the quantity of interest. The factors F - an appropriately weighted value for the nunlher of neutrons produced per fission - and the worth ratio v/F’ of fission neutrons produced in the cell to “‘Cf fission neutrons, have to he deduced from calculations, and the correspond:ing uncertainties have to he taken into account. This holds also for the non-measured contributions to the total fission rate in the cell. Thus, with the fission rates of 239Pu, z41Pu, 235U and 238U having been measured, calculated corrections of about 3% and 6% for the Core 7 and Core 8 lattices, respectively, had to he applied to the measured fission rates. However, the more important contribut,ion to the error in the cell worth normalization - at least in the case of the dry Core 8 lattice for which the analysis has heen completed - stemmed from t.he uncertainty in the absolute calibration of the “‘Cf neutron source.

Normalization by the Reactivity-Reaction Rate Method

The second approach taken for normalizing the cell worth is based on the measurement of the reactivity and reaction rates for a reference material- usually one in which the capture reaction dominates. This method, described by Redman and Bretscher [5] and also reviewed by Davey and Redman r3], in connection with the determination of the capture-to- fission ratio a, has to date been applied and analyzed in the dry Core 8 lattice. The derived value for k, is relatively sensitive to the calculated correction factors. In order to overcome this shortcoming, a variety of different samples were employed, viz. manganese powder, pellets of 0.25% depleted UOz, nlicrospheres

10 r

of ThOz, and foils of nickel plated plutonium (isotopic composition: 91.3/7.9/0.8 at% 239/240/241) containing 1.7 wt% aluminium. The material worths of un- wanted sample constituents were either determined experimentally (oxygen, aluminium, steel) or deduced from earlier measurements in fast lattices (nickel, 23sU, 2’0Pu), appropriate corrections then being applied to the total sample reactivity signals. AU the reference materials were housed in thin-walled stainless steel tubes, the inner diameter of which was approximately equal to the fuel pellet diameter. The samples were oscillated vertically between the core center and a position outside the biological shield of the reactor. Each sample was of 20 cm length, and the l reactivity signals had to be corrected for the axial flux decrease and for axial leakage changes. These corrections were only 1% or less. For a central reference sample (of type denoted by index r) instead of the 2szCf source, Equation 6 reads

where

is a calculated factor for the conversion of the absorption term to the total reactiv- l ity worth of the sample, Ar being the absorption (capture plus fission) operator, 0

is a calculated weighting factor for the conversion of the integral absorption rate to the corresponding component of the reactivity worth, and R and p, are the measured absorption rate and the reactivity signal of the sample, respectively. In the plutonium sample, only the fission rate was measured and the production term derived. Accordingly , the calculated correction factors for this case

11 where F is the fission operator for the plutonium sample. The correction factors IV, I p and s$ were calculated by applying first-order 0 perturbation theory to a single-zone-reactor model corresponding to the test lattice. Macroscopic cross-sections for the samples were extracted from NIMS-D 17] multicell caiculations. Since certain cross-sect.ions necessary for this analysis (e.g. manganese scattering) were lacking in the WIMS IQ81 library 181, a new library based on the JEFF-1 data file was applied -[Q]. F or these calculated correction factors, as for v?,, appropriate uncertainties have to be taken into account. As a numerical check, an additional calculation of the correction for the ‘Yl sample was performed using the WIMS 1981 library. This was found to change the value of (~1- l/k+) only by about 1%.

Table II displays the calculated reactivity terms which det,ermine t.he correc- Con factor LV?, the calculated ratio of the weight factors @“/‘@, and (1- l/rC+) l as derived from the reactivity and the reaction rate measurement for each refer- l ence sample. The latter results have been expressed relative to the same quantity as determined by the Z5zC!f-source method. The relative errors in the reactivity and the reaction rate measurements were typically 3~4% and +2%, respectively. For (l-l/k+), anexperimentalstandard deviationof 15% may thus be expected. Differences between the individual results we seen to be somewhat larger, proba- bly refl&ing errors in calculated down-scattering terms for the absorbers. Com- paring the weighted mean result from the various reference samples with that obtained by the 252Cf-source method, however, the difference is not significant. I

3.3 Control Rod Worths

Reactivity control in standard PWRs is based on the use of two independent control nuxhanisms, viz. soluble boron in the moderator for compensation of slow reactivity changes (e.g. bunlup), and control rod clusters for fast adjustments (e.g. power control) and shutdown. In LWHCRs, the reactivity worth of soluble boron (due to the low moderator content as well as the resulting intermediate neutron spectrum) is lower by a factor of typically 35 relative to standard PWFls

-[lo]. Thus, the use of soluble boron appears to be inappropriate in LWHCRs, especially when considering that the moderator void coefficient would be ad- versely affected. For the above reasons, it is likely that reactivity control in LWHCRs will have to be based exclusively on the use of solid absorbers. Since there are currently considerable uncertainties in the physics design of the come- sponding control systems, related studies have formed an important part of the PROTEUSLWHCR Phase II programme.

Measurements of control rod worths for alternative absorber materials were performed at the center of the test aone in both Cores 7 and 8 by remotely controlled, step-wise insertion of the absorber under study in an empty guide tube during normal reactor operation. The negative reactivity effect was infered from the movement of the compensating autorod. Clearly, the reactivity response of a &riven system like PROTEUS, to the insertion of an absorber rod in the renter of the test sane, cannot be directly interpreted in terms of the absolute l worth of the control rod in a single-zone LWHCFL. However, the measurements l are representative in a relative sense, since the ratio of two individual rod worths thus measured is an LWHCR-specific characteristic, i.e. largely independent of the outer reactor regions.

This is illustrated in Table III which lists the various control rod types investigated and gives the so-called PROTEUS/single-zone factors for the relative reactivity worths (a natural B4C-pellet rod serving as standard in each case). These factors quantify the calculated deviations of PROTEUS core-center values for the considered reactivity worth ratios from values that would occur in a critical single-zone reactor. It is seen that in the wet Core 7 lattice, the necessary

13 corrections are almost negligibly small for ail control rods investigated. For the dry Core 8 lat.tice, the corrections are only slightly larger (upto about 1.5%), except for the relative reactivity worth of the Smz03 rod which is seen to require a 5 % correction. Calculations of the controlrod worth ratios were carried out by applying exact perturbation theory to appropriate 1-D transport-theory models for hoth l?IIO- TEUS and the corresponding single-zone LWHCFL. Homogtized 28.group cross- sections for the control rod cell consisting of three regions (absorber, cladding and moderator), and also for the 6 fuel cells adjacent to the control rod, were generated from WIMS-D/1981 calculations using an appropriate multicell model. Sensitivity studies using simplified 2-D models, as well as subsidiary experimental 0 checks, confirmed the adequacy of a 1-D analytical treatment of the control rod experiments.

4 EXPERIMENTAL RESULTS AND COMPARISONS WITH CALCULATIONS

4.1 Cell Codes and Nuclear Data Tested

Fbr comparisons of the experimental results with calculated values, two calculational procedures have basically been applied, viz. WIMS-D with its ‘1981’ data library $3J and the German code system KARBUS with its associated 69.group library based on KEDAK-4 data 1121. l The WIMS-D code, though in principle applicable to a wide range of reactor types and neutron energy spectra, has been applied and validated mainly l for uranium-fueled thermal reactors. While the cell calculationai procedures in WIMS-D have remained largely unchanged over the years, successive data library adjustments have been made for the principal nuclides - the 1981 library repre- senting a recent set of changes. WIMS-D calculations for the PROTEUS-LWHCR Phase I experiments were originally reported based on the older ‘standard data library [II], and Ref.(l3j d’muws the effects on the results of changing over to the 1981 library. It should be ment.ioned that data for 238Pu, 242Pu, 241Am and Mo currently used in the WIMS-D calculations were generated separately from

14 ENDFIB-IV files at EIR. Whole-reactor considerations necessary for deducing correction factors for the various types of PROTEUS-LWHCR measurements described in Section 3 were made on the basis of 2%group cross-sections for individual material zones generated from WIMS-D. The final evaluation of control rod worth ratios measured in the test zone was also carried out in the same 2%group structure. For the comparison of calculated and measured values of k, and reaction rate ratios, however, the full 69-group structure of the WIMS library was utilized in the corresponding fundamental mode calculations for the test lattices. KARBUS contains special cakulational procedures to deal with some of the LWHCR-specific features which cannot be satisfactorly treated by standard thermal and fast reactor codes. Thus, while containing methods for adequate treatment of an operating LWHCR with emphasis on the resolved resonance region, KARBUS also permits a more accurate calculation for the fully voided core for Rhich fast reactor methods would normally be required. The data base chosen for the current KARBUS analysis of the PROTEUS- LWHCR experiments was a 69-group cross-section set with energy boundaries corresponding to that of the WIMS library. Microscopic cross-sections and shield- ing factors contained in this data set, G69COLD, were directly derived from the Karlsruhe data file KEDAK-4. The KARBUS results for the two-rod Lattices of the Phase I programme were obtained from single-rod calculations and corrected applying two-rod/single-rod factors deduced from WIMS-D.

4.2 k, and Reaction Rate Ratios

Table IV gives calculated/experimental (C/E) values for reaction rates and k-, measured in PROTEUS-LWHCR Cores l-8. Bearing in mind the experimental errors in the Phase I and Phase II programmes (see Sections 3.1 and 3.2), the following observations niay be made:

. k, values for the wet Phase I test lattices seem to be satisfactorily pre- dicted by WIIMS-D using the 1981 library, while KARBUS results appear somewhat low. For the HzO-moderated Phase II reference lattice, however,

15 .

the WIMS result is too high, the trend for KAR.BUS remaining t.he s,ame as earlier.

For both the Phase I and Phase II dry cores, WIMS k, values are markedly overpredicted, while KARBUS results are low by amounts similar to t:hose for the moderated lattices.

a For the standard types of reaction rate ratios measured in both Phase I and Phase II, viz. C~!F~, Fa/Fg and Fs/F9, results are seen to be broadly consistent. Thus. WIMS/1981 predictions appear quite satisfactory in all cores, whereas KARBUS overestimates the measured values significantly in several cases. The latter observation applies particularly to C~/FQ; the l most important single reaction rate ratio measured. Several contradictory trends between the C/E values for Cs/FQ and those for k, can be observed witch both WIMS and KARBUS, particularly in the dry cores.

. For the additional reaction rates measured in Cores 7 and 8, viz. FI /Fb and C2/Fg, t.he C/E values are generally much poorer - especially for C,/FSI in the HzO-moderated lattice where both WIMS and KARBUS overestimate the experimental result by >50%. Clearly, this is an indication that the data for the higher actinides require closer scrutiny in LWHCR applications.

4.3 k, Void Coefficient

The k, void coefficient between voidage states ~1 and ~2 % can he considered as l 111: l k - km, a, = 6;;zy2 - c1) (12)

While 01, was confirmed to be strongly negat,ire in the 6% fissile-Pu Phase I lattice, a slightly negative value was estimated for the 8% case (on the basis of an extrapolation made for the non-measurable reaction rates u). The positive experimental result for a, in the 7.5% fissile-Pu Phase II reference lattice would, at first sight, appear to contradict the above. It is, however, important to hear in mind that the Phase II fuel difkrs from that used in Phase I in several v~.ys

16 .

other than just the effective enrichment, via. its single rod nature, the more representative plutonium isotopic composition and the larger (LWHCR-specific) fuel rod diameter -[14]. The most important single effect which causes the k, void coeflkient to be positive in the Phase II referencelattice is the significantly smaller absolute value of the important negative component due to &/Fe. Also important is the increased posit.ive contribution of the non-measurable reaction rate ratios (involving mainly captures in 23”Pu and 240Pu). The significant positive contribution of 242Pu capture, which was almost negligibly small in the Phase I lattices, is another factor -resulting from the more representative plutonium isotopic corn- l position in Phase II. As ~regardscomparisons of the experimental results with calculations, KAR- BUS is seen to yield satisfactory agreement for the net k, void coefficient, while WIMS/1981 gives too positive a value. Compensating errors in the individual components, however, are clearly indicated for both cell codes. This is, of course, consistent with the partly contradictory trends noted earlier for the C/E values for k, and reaction rate ratios (Section 4.2)

4.4 Control Rod Worths

The experimental control rod investigations (seeSection 3.3) for the HaO-moderated Phase II reference lattice, Core 7, showed that boron carbide - even of natural isotopic composition - is the most promising of the control materials investigated 0 1151.~The rod made of AgInCd-alloy, i.e. the material commonly used in stan- a -dard PWRs. was found to yield a reactivity worth about 25% lower than that of the natural BdC-pellet rod. The reactivity worths of the hafnium and tantalum rods - two control materials which due to their large resonance absorption cross-sectionshave been considered as possible alternatives for LWHCRs - were also lower, viz. by about,25% and 35%, respectively. The rods with the strong thermal absorbers, Gdz03 and Sm20, - included in the experimental programme for the sake of completeness- yielded reactivity worths only about 45 and 35%, respectively, of that of the natural B,C-pellet rod. The relative worths of the vario& control rods in the dry Core 8 test lattice,

17 with its fast spectrum, were quite different. The most significant change was that for the 93% “B-enriched B*C-pellet rod. In the wet Core 7 lattice, this rod. had a worth about 2.3 times that of the natural BdC-pellet rod. In the dry lattice, the reactivity worth ratio was as high as about 5 - it value almost reflecting proportionality to the 1°B content. While the reactivity worths of the Ag:LnCd and tantalum rods, relative to that of the natural B4C-pellet rod increased in the dry lattice to values slightly above unity, that for hafnium decreased somewhat. For Gdz03 and Smz03, decreases in the relative rod worths were more significant, viz. to values of about 30% and lo%, repectively, of the natural B4C-pellet rod worth. Calculations for the relative reactivity worths of the various control rods in- 0 vestigated in Cores 7 and 8 (see Sections 3.3 and 4.1) were carried out on the basis of macroscopic cross-sections generated using WIMS-D with the 1981 library. Ta- ble VI gives calculated/experinlental (C/E) va 1ues for the different worth ratios. The experimental accuracies which need to be borne in mind, were between fl.O to 11.5% in both cores. In general, the C/E values in Table VI are not as close to unity as the results for k, and the standard types of reaction rate ratios measured in the test lattices (cf. Table IV). This is partly due to the non-availability of appropriate data sets in the WIMS 1981 library for some of the investigated control materials. The C/E values in the case of the tantalum and hafnium rods in Core 7, for example, are as high as 2.1 and 1.5, respectively. This may be largely explained by the fact a that infinite dilute resonance cross-sections had to be used for these materials. The analytical procedure for the prediction of a control rod worth ratio is, of l course, considerably more complex than the single, fundamental mode calculation required for obtaining the neutron balance component.s (k,, etc.). Thus, taking into account that three different codes wit.h corresponding numerical approxima- tions were employed in the generation of each calculational result in Table VI, t.he agreement with experiment - for control materials other than hafnium and tantalum in Core 7 - may be considered acceptable. The most important discrepancy is that for the 93% “B-enriched BdC-pellet rod, for which the reactivity worth, relative to natural B4C, is overpredicted by 14Y o in Core 7. This clearly requires

18 futher investigation, considering that enriched B4C is the control material with the most promising features for applications in LWHCRs.

5 CONCLUSIONS

Experimental and calculational results have been presented and discussed for the PROTEUS-LWHCR Phase II reference lattice with Hz0 (Core 7) and air (dry, Core 8). It has been found that comparisons for k, and reaction rate ratios are broadly consistent with those made for the Phase I experiments. The new measurements, however, are mire representative and accurate and, as such, of greater value for identifying specific shortcomings in the tested codes/data libraries. The k, void coefficient for the Phase II reference lattice between 0 and 100% void has been found to be positive - as against the negative values assessed for the Phase I test lattices. This results largely from the more representative plutonium isotopic composition of the fuel, as well as the larger (LWHCR-specific) fuel rod diameter. For both cell codes tested, viz. WIMS-D/1981 library and KARBUS/KEDAK-4, compensating errors in individual void coefficient components have been clearly indicated. There is a need, therefore, of an even broader experimental data base, and this will be made available during the remaining course of the PROTEUS-LWHCR Ph ase II programme through the simulation of an intermediate HzO-voidage state, as well as investigations for somewhat a wider-spaced test lattices. i It should be mentioned that the k, void coefficient, as discussed in the paper, is the basic parameter determining the physics behaviour of a given type of LWHCR upon moderator voidage. Extrapolation to an operating power reactor, however, involves taking into account the effects of leakage, temperature, the presence of control rods, ,as well as fuel composition changes with burnup. Thus, for example, enhanced leakage would provide an important negative contribution to the effective void coefficient. The application of a zero-power facility for the investigation of such power reactor features is naturally limited in certain respects. The wider range of exper-

19 itnents being conducted in the PROTEUS-LWHCR Phase II progranune, however, does aim at clarifying some of these aspects. The currently described investigations for control rod worths, under both normal and voided conditions, is illustrative of these aims.

NOMENCLATURE k, = neutron productions to neutron absorptions in the fundamental mode spectrum Ri = Reaction Rate, type i (macroscopic) c* = *3*u capture rate per atom 0 cz = ==P” capture rate per atom F5 = 9J fission rate per atom Fs = 238U fission rate per atom Fg = 239P” fission rate per atom Fl = 241Pu fission rate per atom a” = k, void coefficient (10F4/%) a,; = a, component due to Ri expressed relative to 230Pu fission (lo@/%) M2 = migration area Bk = material buckling 2 % = axial component of material buckling 0; = radial component of material buckling 0 fif = radial curvature of experimental reaction rate traverse of type i A/3,? = correction term APf = @ - 0: l k+ x adjoint-weighted ratio of neutron productions to neutron removals at the center of the test zone 4 = neutron flux at the enter of the test zone $4+ = adjoint neutron .flux at the enter of the test Anne PC = reactivity worth of central unit cell PF = reactivity worth of reference sample r Pa = reactivity worth of “‘Cf neutron source i7 = average number of neutrons produced per fission

20 v, = number of neutrons produced per fission in the plutonium sample W7 = factor for converting the measured reaction term to the total reactivity signal for reference sample r S = absolute source strength of the “‘Cf source F = worth of fission neutrons produced in the central cell p = worth of fission neutrons produced by the ?2f source -T 4. = worth of reaction neutrons of the reference sample r A = absorption (fission plus capture) operator P = production operator H = Ternoval operator S = source strength operator for the 252Cf source F = fission operator

21 ACKNOWLEDGEMENTS

The authors are grateful to FL Brogli for his support of the PROTEUS- LWHCR activities, to G. Pi&r for his active participation in the early part of the Phase IE programme, to EL Graf for valuable contrihmionr fo t~he technical plan- ning and execution of some of the experiments, and to P. Bourqtin, T. Steiner and P. Thomi who, together with H.Graf, ensured the smooth and efficiem operation of the reactor. Thanks are also due to C.H.M. Breeders, E. Kieiha’ber, S. Pelloti and J. Stepmek for useful discussions and for making available some of the calculationai tools. References

PI R. Chawla, K. Gmiir, H. Hager and R. S&r, “Reactivity and Reaction Rate Ratio Changes with Moderator Voidage in R Light Water High Converter Reactor Lattice”,Nucl. Technol., fj7, 360 (1984)

PI R. Chawla, R. Seiler and K. Gmiir, “Effects of Fuel Enrichment on the Physics Characteristics of Plutonium-fueled Light Water High Converter Re- actors”, Nucl. Technol. 13, 296 (1986)

131 W.G. Davey and W.C. Redman, “Techniques in Fast Reactor Critical Ex- periments”, Gordon and Breach, 1970 0

[41 D.J. Donahue, D.D. Laming, R.A. Bennet and R.E. Heinemann “Determi- nation of km from Critical Experiments with PCTR”. Nucl. Sci. Eng., 4,297 (1958)

[51 W.C. Redman and M.M. Bretscher, “Direct Determination of Uranium-235 Capture-to-Fission Ratio in B Zero-Power Reactor”, Nucl. Sci. Eng., a, 34 (1967)

PI J.P. Chaudat, M. Darrouzet and E.A. Fischer, “Experiments in Pure Ura- nium Lattices with Unit k, Assemblies Sneak-8/8Z; UK 1 and UK 5 in ERMINE and Harmonic”, CEA-R-4552, Commissariat Energie-Atomique, 0 Cadarache (1974) 0 [7] J.R. Askew, F.J. Fayers and P.B. Kernshell, “A general description of the lattice code WIMS”, J. Brit. Nucl. Energy Sm., 5, 564 (1966)

[8] M.J. H&all and C.J. Taubman, “The ‘198l’WIMS Nuclear DataLibrary”“, AEEW-R 1442, U.K. Atomic Energy Authority, Winfrith (1983)

[9] S. Pelloni and J.Stepanek, “Testing of a JEF-1 Based WIMS-D Cross-section Library for Migratioti Area and k, Predictions for LWHCR Lattices”, EIR- Bericht Nr. 610, Swiss Federal Institute for Reactor Research (1987)

23 [lo] H.D. Berger, “Neutronenphysikalische Untersuchungen zu einem fort- geschrit~tenen Druckwasserreaktor nlit hoher Konversion”, GKSS 85/E/15, GKSS-Research Centre Geest~hacht (1985)

[II] C.J. Taubman, “The WIMS 69.Group Library Tape 166259”,AEEW-M1324, U.K. Atomic Energy Authority, Winfrith (1975)

1121 C.H.M. Breeders, “Neutron Physics Investigations for Advanced Pressurized Water Reactors”, Nucl. Technol. & 96 (1985)

1131 R. Chawln, H.-M. Hsieh and M.J. Walsall, ‘LEffects of Recent WIMS Data Library Changes on Calculational Results for LWHCR Lattices”. Ann. Nucl. Energy, 13, 523 (1986) l

[14] R. Chawla, K. Gmiir, H. Hager, G. Pi&r, R. SeileT and H.D. Berger, “Bench- mark Measurements of the Neutron Balance in Light Water High Converter Reactors”, Proc. European Nuclear Conference’86, Geneva, Vo1.2,567 (1986)

[15] H.D. Berger, R. Chawla and H.Hager, “Reactivity Control Investigations for Light Water High Converter Reactors”, Proc. European Nuclear Confer- ence’86, Geneva, Vo1.2, 561 (1986)

24 Table I: Deduction of the radial component of the material buckling, 0:. from reaction rate traverses measured in Cores 7 and 8 (units: m-‘).

Core 7 (HzO) Core 8 (Air)

Reaction Rate (Ri) F5,Cs F,,Rh Fs FS G FS Pf a) 14.4hO.9 19.5zt0.7 -2.9hO.7 -1.9zt0.9 3.1*0.6 1.5f0.8 A@ *) -2.6 t2.3 -5.2 -4.7 to.8 -0.9 Pk 17.0 17.2 2.3 2.8 2.3 2.4

Q, from Bessel-function fits between 0 - 12.5 cm radius 0 *) obtained from whole-reactor calculations

Figure Captions

Vertical sectional view of the PROTEUS reactor.

Reference Test Lattice for PROTEUS-LWHCR Phase II, Cores 7-9. Dimensions are in millimeters.

0 -Fig.3 Photograph showing the insertion of a “void-box” into the &O-moderated PROTEUS- 0 LWHCR Core 7 test zone.

25 Table II: Calculated quantities for various reference materials in the dry Core 8 test lattice and comparison of the derived experimental results for k+ relative to that from the 262Cf-source method. The total reactivity signal (sum of production and removal terms) is normalised to a value of 100 for each material.

Sample Reactivity Contributions (%)

Production Absorption 56Mn -49.5 mu 75.4 -129.4 -46.0 -0.773 =‘Th 7.5 -85.5 -22.0 -1.170 23OPll 149.6 -48.7 -0.9 0.668

a) Only statistical errors in the reactivity and reaction rate measurements have been considered. Neither errors in the calculated factors nor the systematic error due to the Z52Cf-source result are included. *) standard deviation of the mean

Table III: PROTEUS/single-zone factors for the relative a) worths of different control rods in PROTEUS-LWHCR Cores 7 and 8 - i Control rod 1 PROTEUS/Single-Zone Factor Material Diameter FClrIIl l Core 7 (HzO) Core 8 (Air) - BaC-nat 7.4 powder 1.001 1.001 - l BdC-93 % 1°B 8.3 pellet 0.993 0.994 A&Cd 8.9 alloy 1.001 0.987 Hafnium 8.3 metal 1.001 1.020 Tantalum 8.3 metal 1.000 1.016 G&O3 8.3 pellet 1.002 1.013 Snl~OS 7.0 pellet 1.003 1.051 -

a) normalized to the worth of a nat. BdC-pellet rod of diameter 7.5 mm

26 .

Table IV: CalculRtion/Experiment (C/E) values for k, and reaction rate ratios measured in LWHCR-PROTEUS, Cores l-8.

WIMS-D, 1981 KARBUS, KEDAK-4 Hz0 Dowtherm Air Two Rod Lattice, 6: pqias 0 1.005 1.057 1.105 0.999 1.083 1.069 1.006 1.057 1.024

1 1.009 1.008 1.040 0.992 0.974 0.981

Two Rod Lattice, 8! p”fisa 1.084 1.097 1.091 1.051 1.031 0.997 1.044 1.054 1.033 1.039 1.033 1.037 1.044 1.051 1.042

i Single Rod Lattice, 7. 5! % PUfis.. 0.982 1.007 1.036 - 1.082 1.030 - 1.026 1.024 - 1.047 0 0.995 - 1.013 1.012 - 1.006 1.054 - 1.152 1.008 - 0.967 0 1.729 - 0.962 1.552 - 1.138

1.012 - 1.037 I 0.986 - 0.984

27 Table V: The k, void coefficient 01” and its components between 0 and 100% void for the various types of test lattices investigated in PROTEUS (units: lo-*/%)

Calculation WIMS-D, 1981lKARBUS, KEDAK-4. Lattice, 6% Pu~;~, -35.1 I -3a.c: +6.1 f 0.4 -1-6.5 +7.c to.3 i 0.1 to.3 to.4: t11.4 i 2.0a) +16.9 +15.1.

-14.4 f 1.4 -11.4 i -15.6 Two ROI Lattice, 8% Pufiao avi(C8lFQ) I -25.2 I!C 1.7 -24.1 ) -25.f; t5.2 +5.: to.2 to.2 a,; (‘others’) t20.7 t18.5

-23:7 +5.5 tf3.0 to.5 tO.!j -0.7 -1.5 t2.3 +1.9 t21.6 t20.s

t6.7 t4.0 i

a) Difference of experimental net au and sum of measured cr,i’s *J The usual ARi definition [l] in this deduction was found inadequak, and Eqn. A.3zef.2, was applied

28 . c

Table VI: CalculntionlExperiment values for control rod worth ratios a) in PROTEUS-LWHCR Cores 7 and 8. The calculational results are based on cross-sections derived from WIMS-D/1981 library 0 I Control rod I 1 Calculation/Experiment

1.143 0.921 1.012 0.944 1.535 1.085 2.097 1.162 0.917 0.893 0.933 1.090

a) relative to the nat. B4C-pellet rod b, infinite dilute resonance cross-sections used c) data for the less important isotopes missing

29 TEST LATTICE Up? BLANKET

ZONE :LECl-OR . ,.

11% PuO*/UO* (7.5% PUfj&

‘Moderator

Steel +Air

F/M q 2.07