NEACRP-L-323 NEANDC-299"U" General Distribution

OECD NUCLEAR ENERGY AGENCY (NEA) Committee on Reactor Physics (NEACRP) and Nuclear Data Committee (NEANDC)

-

STATUS OF DELAYED NEUTRON DATA 1990

J. Blachot, CEN Grenoble, FRANCE. M. C. Brady, ORNL. USA. A. Filip, CEN Cadarache, FRANCE. R. W. Mills, and D.R. Weaver, Univ. of Birmingham, ENGLAND. December 1990

0.1

- 1%

Largest Fission Yield

Nuclides to the right are Delayed Neutron Precursors

Organisation for Economic Cooperation and Development (OECD) Nuclear Energy Agency (NEA) Committee on Reactor Physics (NEACRP) and Nuclear Data Committee (NEANDC)

STATUS OF DELAYED NEUTRON DATA - 1990 J. Blachot, Centre &Etudes Nucleaires de Grenoble, Grenoble. FRANCE. M. C. Brady, Oak Ridge National Laboratoly, Oak Ridge, Tennessee, USA. A. F i l i ~Centre . &Etudes Nucleaires de Cadarache. St. Paul-lez-Durance. FRANCE. R. W. ~ i l l~ ic h o oofi Physics and Space Research, he university of ~irrningham,ENGLAND. D.R. Weaver, School of Physics and Space Research, The University of Birmingham. ENGLAND.

December 1990

- ABSTRACT Delaved neutron data Dlav a kev role in the reactor ~hvsicsanalvsis of safetv related oarameters. This is ihe case for any 'of reahor. For existingarid &raing ieactok theherest i i for an imorovement of the basic data w ich are used to establish the reactivitv scale and the reduction of the associated uncertainties. In th~scontext delayed neutmn data play a sfgnificant role. Moreover, there is at present a strong trend towards the study and development of new reactor types. For these advanced and innovative reactor concepts, there is a need to establish complete and sound data bases.

6

~

~

I?

This p er carlies out a review of the delayed neutron parameters and their uncertainties as available today. %e review focuses on reactor technology and presents the data in a "consistent structure" having three levels of refinement. Conclusions on the quality of the data together with recommendationsfor improving them through modelling and measurements on each of the three levels are formulated. Improvements in delayed neutron data make it ssibie to establish a more precise reactivity scale for existing reactors. Increased safety margins can achieved by reducing. by about a factor Of 2, the current uncertainty of 5 % (one standard deviation).

Contents 1

2 3

..................................................................................................... 5 BACKGROUND...................................................................................................... 6 MACROSCOPIC DELAYED NEUTRON DATA.....................................................7

INTRODUCTION

3.1 Time-Dependent Parameters............................................................................ 7 3.1 .1 Six-temporal group constants................................................................ 7 3.1.2 Spectra associated with the six-temporal groups...................................8 3.2 Time-Independent (Equilibrium) Delayed Neutron Data.................................... 9 3.2.1 Equilibrium delayed neutron spectra...............................:......................9 3.2.2 Absolute delayed neutron yields from fission.........................................9 4

.........................................................................................

RECOMMENDATIONS 14 4.1 v d Absolute Data............................................................................................ 14 4.2 Temporal Group Parameters and Spectra...................................................... 15 4.3 Integral (level-3) Testing of Delayed Neutron Data.........................................15

6

...................................................................................................... FIGURES...............................................................................................................

7

TABLES................................................................................................................ 23

5

REFERENCES

17 21

Figures 1

Delayed Neutron Parameters................................................................................21

2

vd experiments after 1955 with uncertainty c 10 %.............................................22

.

Tables -

1

Evaluations of v per 100 fissions ....................................................................... 23

2 3

Nuclides for which vd has been measured .......................................................... 24 Uncertainties (in %) for the Evaluation of vd ....................................................... 24

4

235 238 239 vd per 100 fissions for U, U and Pu ...................................................... 25

5

Review of vd values per 100 fissions by summation........................................... 26

d

-

a

INTRODUCTION

1 INTRODUCTION

@

Trends in reactor technology reported at meetings such as PHYSOR'SO include the increased use of mixed-oxide fuels and the move to increased enrichment and burnup in current light water reactor designs as well as the development of fast reactor designs such as the actinide burning reactor. Accurate predictions of the kinetic response of these new reactor fuels require reliable data concerning delayed neutron production. These trends illustrate the necessity for improving the delayed neutron data available for transuranic nuclides and resolving the discrepancies existing in the current data, particularly those for '%. The current delayed neutron data have been shown to contribute significantly to the large uncertainties' in the reactivity scale for fast reactors; improvements in the basic delayed neutron data and integral benchmarking are necessary to reduce these uncertainties. This paper will briefly review the status of delayed neutron data, identify prominent discrepancies, and identify and prioritize areas for improvement in the data.

'

unwxtainties are given tbmughout as one standard deviation.

-5-

BACKGROUND

2 BACKGROUND The fundamental goals of every basic data library are precision and general applicability; the long term objective of the delayed neutron evaluation effort must be focused toward these same goals. A straightforward method for doing this is related to an effort to ensure consistency between 3 levels of treatment of delayed neutron parametsrs (ref. 1). These, as described in figure 4 , are: e

Levol 4 . The individual precursor, or microscopic, level; this is tho most extensive and physically based of the three levels.

s

Level 2. The aggregate precursor, or macroscopic, level which is more synthetic b& contains global parameters which are directly measurable and these can be used iin reactor applications.

e

Leve! 3. Tha integral level; this provides important and stringent tests related to ~.1.3 ! i e ~ ~ t i osuch n s as reactor kinetics calculations and includes corngarison of values f ~ Fe8 r from measurements with values from calculations using iower level data.

0

l3aorotically, a complete knowledge of the information at level 4 would be sufficient as Pha parameters at the other 2 levels can be deduced from Phis. Practically and historically the 3 levels were unequally and nearly independantly developed in terms of their Measurement, Modelling, and Evaluation (MMdiBE). The data initially produced in the first decades (1950-4970), were principally at level 2 wiih some at level 4 , such as fission product yields. The basic work at this time was timi performed by Keepin (ref. 2) which resulted in the now familiar six-group modelling of delayed neutron parameters as used in all later fission reactor design. The last tws clecades (1970-1990) have focused on the MM&E for level 4 data, primarily fission yields, the delayed neutron emission probabilities (P,) and spectra (xd). These advances in the level 1 data have been based upon work in nuclear physics investigating the detailed structure of nuclei. Some level 3 data have also been developed in this period, and small advances in level 2 parameters such as 7 and spectra have also been made. With the evaluated microscopic data now availab e the macroscopic parameters of level 2, and consequently the integral pararneiers of level 3 (namely Pefl), can or already have been calculated with satisfactory precision except (see section 3.2.2). for

F'

vd

In the next section the status of macroscopic delayed neutron data (level 2) is reviewed, focused on the fissile nuclides of principal importance for nuclear reactor 23s na U and ='Pu. *me suggestions for future experimental and technology, U, theoretical work to improve the data are made.

@

MACROSCOPIC DELAYED NEUTRON DATA

3 MACROSCOPIC DELAYED NEUTRON DATA These data are conventionally classified as time-dependent and time-independent (equilibrium) parameters.

3.1 Time-Dependent Parameters The time-dependent parameters such as delayed neutron activity, DN(t), and the spectra ~ ( t )which , result from the decays of individual precursors are currently obtained by "summation" from level 1 microscopic data and they have been shown to perform in general good agreement with level 2 data obtained experimentally. Specifically, this is done by starting with the fission yields of delayed neutron precursors, in either independent or cumulative form (Yi or Yc), and then combining these results with decay data: P, values and the spectra from individual precursors. See the scheme in figure 1. The DN(t) parameter is usually represented in a 6-temporal group form (a and k) normalized to an absolute delayed neutron yield, Spectra corresponding to the 6-temporal groups may also be derived from the level 7 data using indirect summation techniques (Brady, ref. 3).

3.1.1 Six-temporal group constants

Current sets of six-group parameter data from the literature are: the (classical) Keepin (1965, ref. 2), which are essentially the data proposed by Cox (1974, ref. 4) for ENDFIB-IV (Evaluated Nuclear Data File) and retained for ENDFB-V, the Waldo et al (1980, ref. 5) measured data. the Manevich et al (1988, ref. 6) evaluated summation data, and the Brady and England (1989, ref. 3), summation data. The differences between these data sets are evident, but the repercussions of using a particular set on the calculated reactor kinetics parameters (a stringent test) are less noticeable (Stevenson, ref. 7), although some significant uncertainties in the reactor reactivity (inverse kinetic equation) were pointed out by Manevich and co-workers (1988, ref. 5). The most apparent differences are seen in the groups representing short decay times.

MACROSCOPIC DELAYED NEUTRON DATA

More specific sensitivity studies are needed to determine the importance of these parameters relative to delayed neutron activity, DN(t), and the kinetic response to reactivity changes. 3.1.2 Spectra associated with the six-temporal groups Measured rou s ctra prior to 1986 were available only for groups 1-5 for 2 5 p2s', and only 1-4 for U and Pu. Measurements by Lowell (ref. 8) and Birmingham (ref. 9) have-oxpaded the measured spectra to include all six delayed neutron groups E8t1, and 2 3 9 ~ u for

Tho more recent and complete data sets for six-group spectra are:

1. ttra surnrnaiion results of Brady et al. (1989, ref. 3) which utilized a great deal of level 1 individual precursor data (measured by Rudstam (ref. lo), Mratz (ref. 1I), Shalev (ref. 22), Reeder (ref. 13). Greenwood (ref. 14) and many others). These were augmented for the unmeasured energy ranges by theoretical predictions (iliiann, ref. 25) and rely on systematics models for precursor nuclides with no maasured cia& (England et al. 1986, ref. 16). level 2 macroscopic data. These include both time-dependent 2. directly ~:?~asuif;sf spectra and those arranged into six temporal groups. The results include earlier measui.ornenCs (Batchelor (ref. 17), Feig (ref. 18), Shalev and Cunler (ref. 19), and others) plus the more recent (1988-1989) measurements at the University of Lowell (ref. 8)and those of the University of Birmingham (1986, ref. 9). Tho cansistency of the two entirely independent methods (4) and (2),has been demonsf~atedby Brady et al. (1989, ref. 3), and the set of results they proposed are probably the most complete and accurate data available on aggregate (level 2) spectra. The sensitivity of fast reactor kinetic behavior to variations in delayed neutron energy spectra has been studied by Das and Walker (1986, ref. 20) who concluded "no striking consequences of spectral changes have been observed". sensitivity to delayed neutron spectra are in progress in Further studies of $ Casaccia (D'Angelo, &o, ref. 21).

0

MACROSCOPIC DELAYED NEUTRON DATA

3.2 Time-Independent (Equilibrium) Delayed Neutron Data 3.2.1 Equilibrium delayed neutron spectra

Equilibrium energy spectra are readily deduced from the time-dependent results noted in Sect. 3.1.2. The results of Los Alamos (ref. 3) are to be utilized %onsistentlywith the six-group spectra. 3.2.2 Absolute delayed neutron yields from fission

This parameter is the most important from two perspectives:

cd

1. as a stringent test of overall consistency of the two levels of data; e.g., resulting from the summation calculations using the microscopic data of level 1 in comparison with experimental values for vd on the macroscopic level (level 2), and 2. as a fundamental parameter in nuclear reactor kinetics and dynamics studies, via parameters derived at level 3. Specifically the integral peff is particularly sensitive to the parameter (D'Angelo, 1990, ref. 21).

vd

In connection with this second perspective, it is useful to distinguish between three classes of fissile systems:

0

Firstly, those concerning the short term goals associated with nuclear energy technologies, e.g. thermal and fast fission in 2 3 5 ~ thermal , and fast fission in ?'PU and fast fission in (labeled for convenience a 5 (T~and F). u and F) ,. 2 3 9 ~(T and 2 3 8 ~(F)). secondly, fissile systems related to intermediate or long term goals (such as improved or new concepts for nuclear reactors); particularly =?h (F), =U (F and T), "OPU (F), 2 4 1 P(F ~ and T) etc. finally, the higher transactinides and some exotic svstems immrtant for fundamental physics or particular applications (e.g. space reactors 1990, ref. 22)).

o on en,

In this paper we focus our review on the first class of fissile systems. Summation calculations using level 1 data have been performedby several groups. We have referred here to the most recent andlor complete results. There are three evaluation efforts effectively utilizing data on level 1:

MACROSCOPIC DELAYED NEUTRON DATA

* ENDFIB-VI (ref. 3)

a

based on preliminary data from ref. 23a for fission yields and on data from ref. 23b for P,,.

JEF-2 (ref. 24) (Joint Evaluated File)

based on data from ref. 25 for fission yields and mainly on data from ref. 26 for P,,.

Soviet work

based on data from ref. 5.

On the otier hand, there are many proposed evaluated from MM&E on level 2. We will address these first.

iddata resu%ing

A selection among the many proposed evaluated data sets is ~>i.~s.~;???d in table 1. In figwe 2 are displayed the results of MM&E (evaluations by TuMIe 1979, ref. 27b) for ali tho measurements with uncertainties concerning the three most irnprtant fission systoms I'or nuclear reactor technology - thermal and fast fission of U and 239P~ and ~ , fast fission of 2 3 8 ~ .

It is considered that the evaluations of Tuttle (1975 ref. 27a,1979 ref. 27b) are ?Re most careful and exhaustive and, therefore, recommendable provided that more recent rmaswements by Synetos, Waldo, Benedetti, and other (post 1979) results be included. Table 2 lists nuclides with measured data at one or more incident neutron energies.

vd

I-lowever, some remarks are to be made: 1. The uncefiainties appear to be too optimistic, as evident in figure 2. For a more quantitative illustration see table 3 where the uncertainties (associated with the mean of all measurements having uncertainties less than 15% in W 0 evaloation by Tuttis) haw txm calculated in several classical ways.

It is apparoiii that the Tuttle uncertainties on the mean approach tho standard deviation of the mean, relevant to the central limit theorem, which is not rigorously applicable since the measurements considered are not actually independent (they utilize some common techniques) and not always sufficiently numerous.

vd

Consider the data for for fast fission in ='u. The inverse variance weighted mean calculated using the measured data reported in ref. 27b is 0.Q4354. The uncertainty in the value may be calculated as the internal error (2.96%) or the external error (4.01%). Note that the internal and external estimates of error are different methods of calculating the uncertainty on an average. Tho internal estimate is based on the determination of a mean of a set of values using the inverse of its variance as a weight for each value. The internal estimate of the variance of the mean is then the inverse of the sum of these weights, hence the internal estimate of the standard deviation of the mean is derived as the square root of this variance. The internal error thus represents the expesLci achievable standard deviation based upon the given uncertainty information on the input data.

MACROSCOPIC DELAYED NEUTRON DATA

@

The external estimate of the error of the mean is derived from the weighted sum of squares of deviations of the data values from the weighted mean. It thus represents the standard deviation actually arising from the distribution of the data about the mean. If the original data is normally distributed, and the original estimates of errors on these values are reasonable, then the internal and external estimates of the standard deviation of the mean should be close to being equal. With respect to the data for obtained from ref. 27b, the estimates of these errors are quite different indicating that either the data are not normally distributed or the estimates of the errors on the original data are not reasonable. In this case the external estimate, the more conservative error is the one to be recommended. The resultant X-squared test for these pre- 1979 values is 94% representing a good distribution about the mean. Inclusion of the Waldo 1981 (ref. 5) value only slightly changes the mean (.043577) and the internal and external errors are 1.95% and 4.04%, respectively. The X-squared test for these data is 96%. The Tuttle 1979 evaluation (ref. 27b) gives a value of 0.0439 for (F) with an uncertainty of 2.3%. It is considered that the real uncertainties lie between the standard deviation of the mean and the standard deviation of the individual typical measurement of that is: +IS% for P 5 (T,F) ~ +I-4% for n g (T,F) ~ ~