Made with Xodo PDF Reader and Editor

STUDIL3 o f t h e r e a c t i v i t y
TEMPERATURE COEFFICIENT
IN LIGHT WATER REACTORS

MALTE EDENIUS

DEPARTMENT OF REACTOR PHYSICS

Made with Xodo PDF Reader and Editor

Studies of the reactivity temperature coefficient in

light water reactors

Malte Edenius

AKADEMISK AVHANDLING

som framlagges till offentlig granskning

vid sektionen for teknisk fysik for av-
laggande av teknisk doktorsexamen i reak-
torfysik den 4 maj 1976, kl 10.00, semi-
narierummet, institutionen for reaktor-
fysik, Gibraltargatan 3.

Goteborg

Mars 1976
 Made with Xodo PDF Reader and Editor

A E —R F —76—3160

STUDIES O F THE REACT IVIT Y T EM PERATURE COEFFICIENT IN

LIGHT WATER REACT ORS

Malte Edenius

AB Atomenergi, Studsvik
March 1976
Made with Xodo PDF Reader and Editor

L I S T OF CONTENTS

1. Introduction 1

2. I n f luence of t e m p e r a t u r e on the neutron

transport 4
2.1 Thermalization 4
2.1.1 N e u t r o n s c a t tering laws 5
2.1.2 S c a t tering m o d e l s for water 7
2.2 The D o ppler effect 9
2.3 D e n s i t y effects 14

3. The cell code A E - B U X Y 17

3.1 N u c l e a r d a t a library 17
3.2 Res o n a n c e treatment 19
3.3 M i cro gro u p c a l c u l a t i o n 21
3.4 M a c r o gro u p c a l c u l a t i o n 23
3.5 Fundamental mode calculation 24

4. C o m p a r i s o n b e t w e e n theor e t i c a l results
and e x p e r imental i n f o r m a t i o n 28
4.1 D e s c r i p t i o n of the m e a s u r e m e n t s 29
4.2 D e s c r i p t i o n of the calcula-
tional m e t h o d s 30
4.3 The r e a c t i v i t y w o r t h of spacers 31
4.4 R e sults of c o m p a r i s o n b e tween
theory and experiment 33
4.4.1 Latti c e s w i t h 1.35 % enriched
U02 33
4.4.2 Latti c e s w i t h 1.9 % enriched
U O 2 rods and P U O 2 rods 35
4.4.3 Comments to the r e s u l t s 39

5. T h e co m p o n e n t s of the t e m p e r a t u r e c o ­
ef f icient 44
5.1 P a r t i a l temperature, c o e f f i ­
cients 46
5.2 C o n t r i b u t i o n s to the t e m p e r a ­
ture c o e f f i c i e n t calculated
b y C O EFF 52
5.2.1 D e s c r i p t i o n of C O E F F 52
5.2.2 Results from calculations with
COEFF 55
 Made with Xodo PDF Reader and Editor

6. The i n f l u e n c e of a ppro x i m a t i o n s in the

theoretical treatment on calculated tempe­
rature coefficients :/o
6.1 The c r y s t a l l i n e b i nding in "0
6.2 T he c y l i n d r i c a l i z a t i o n of pin
cells 74
6.3 C o m p a r i son b e tween pin cell
c a l c u l a tions u s i n g isotropic
and a n i sotropic scattering 7fe
6.4 The c a l cu l a t i o n of leakage 77
6.5 T h e n umber of energy groups
and space points 78
6.5.1 M a c r o g roups and Gauss points
in A E - B UXY 75
6.5 . 2 E n e r g y groups and m e s h points
in the d i f f u s i o n th e o r y c a l c u ­
lation 80
6.6 The i n fluence of thermal ex­
p a n s i o n on r e a c t i v i t y 81
6.7 The p l u t o n i u m f i s s i o n neutron
spectrum 83

7. T he i n f l u e n c e of nuclear d a t a on the
ca l c u l a t e d temp e r a ture coefficient 86
7.1 C o m p a r i s o n of t e m p e r a t u r e c o ­
effici e nts calculated by u s e
of E N D F / B and UKNDL d a t a 86
7.2 The e f f ective r e s o n a n c e i n t e ­
gral of U-238 88
7.3 The r m a l scattering data 94
7-3.1 Scatte r ing in water 94
7.3.2 S c a t t e r i n g in UC^ 94
7.4 Thermal a b s o r p t i o n and f i ssion
cross sections 97
7.4.1 G r o u p cross sections for a
1/v-ab s orber in M a x w e l l i a n
spectra of v a r i o u s t e m p e ratures 97
7.4.2 T h e r m a l d a t a for U-235 and
U-238 98

8. Summary 109

9. Acknowledgement 111

10. References 112

Made with Xodo PDF Reader and Editor

1. INTRODUCTION

One important quantity in determining the operating charac­

teristics and safety of nuclear reactors is the temperature
coefficient of reactivity. The isothermal temperature coef­
ficient in Light Water Reactors (LWR) varies considerably
with the design, the moderator temperature and the boron con­
centration in the moderator etc. Typical values for a fresh
Boiling Water Reactor (BWR) core is -5 pcm/°C (1 pcm = 10
at 20°C and -25 pcm/°C at 280°C. Typical values for a Pres­
surized Water Reactor (PWR) are in the range -5 to -30 pcm/°C.

Much effort has been devoted to the development of proper

methods for the calculation of temperature coefficients. De­
spite this, the methods in current use to calculate the tem­
perature dependence of reactivity in light water reactors are
not altogether successful [l - 4]. As an example measured and
calculated values of dk ,-,/dT from the Swedish Oskarshamn-I
err
BWR reactor are shown in Fig 1.1.

The calculated temperature coefficient is usually 2 - 5

pcm/°C too negative compared to the measured value. Going
from room temperature to operating temperature this means an
error in the predicted reactivity of about 1 %. Thus an in­
centive exists for further studies in this field of reactor
physics. The literature lacks information about clean
(simple, well-defined) and precise high temperature experi­
ments. With the extensive series of experiments which have
been performed in the high temperature facility KRITZ at
Studsvik sample data to compare with calculated ones have
been produced.

In the present paper a survey of the temperature effects in

a nuclear reactor is first given. Then follows a description
of the analysis of the KRITZ experiments and a comparison
between theoretical results and experimental information.
The theoretical methods used in the analysis are described.
 Made with Xodo PDF Reader and Editor

Different components of the temperature coefficient are

studied in chapter 5. Uncertainties in the employed theore­
tical methods are discussed in chapter 6 and the influence
of nuclear data on the temperature coefficient is investiga­
ted in chapter 7. A summary is given in chapter 8.
 Made with Xodo PDF Reader and Editor

( pcm / ° C )
coeff
Temp

Fig. 1.1 Oskarshamn I. The isothermal temperature coefficient

versus temperature [1].
 Made with Xodo PDF Reader and Editor

2. INFLUENCE OF TEMPERATURE ON THE NEUTRON TRANSPORT

The influence of temperature on the neutron transport is

caused by the thermal movement of nuclei influencing the
scattering of thermal neutrons and the Doppler broadening of
resonances and by the thermal expansion of different mate­
rials.

2.1 T h e rmalization

In m a n y materials w h i c h are present in a nuclear reactor the

atoms may be considered to be free. The energy distribution
of the atoms is then the Maxwell distribution and it is poss­
ible to derive an accurate expression for the scattering of
neutrons. This model can be used e.g. for scattering against
heavy nuclei which does not affect the neutron spectrum very
much.

For light nuclei, however, the treatment of scattering re­

quires a consideration of the chemical binding and for a sa­
tisfactory treatment of the thermalization in water it is
necessary to use scattering cross sections computed according
to a relevant model for the scattering process. A number of
approximations are involved in these scattering models. The
accuracy of the model is of great importance w h e n predicting
the temperature dependence of the neutron spectrum and hence
for predicting the temperature effects in thermal reactors.
It may be noted that an accurate treatment of the thermali­
zation is especially important in systems with only partial
thermalization, i.e. w h e n the neutrons do not obtain a M ax­
w e l l i a n velocity d i s tribution before they are absorbed. This
is the case in light w a t e r reactors.

The influence of b i n ding on absorption is negligible and

a bsorption cross sections can be taken to be the same as for
free n u c l e i .
 Made with Xodo PDF Reader and Editor

2 . 1.1 N e u t r o n s c a t t e r i n g laws

Th e s c a t t e r i n g in a m o n o a t o m l c gas is tre a t e d in m o s t b o o k s
o n r e a c t o r p h y s i c s and w i l l not be d i s c u s s e d here. In
a p p l i c a t i o n s all s c a t t e r i n g ex c e p t that in the m a i n c o n ­
stitu e n t of the m o d e r a t o r is u s u a l l y c o n s i d e r e d to obey
the free at o m s c a t t e r i n g law.

For n e u t r o n s u n d e r g o i n g s c a t t e r i n g in a m e d i u m c o n t a i n i n g
b o u n d atoms it has b e e n s h o w n [5] that the s c a t t e r i n g f u n c t i o n
c a n be w r i t t e n as the sum of d i f f e r e n t i a l c o h e r e n t and in­
co h erent cross sections

Z (E-*E 1 1) * Z L (E->E *, 1) +1. (E-*E f 1) ( 2 .1 )

---- coh — me ------

(2 .2 )

(2.3)

E and ft are the e n e r g y and the unit v e c t o r in the d i r e c t i o n

of m o t i o n b e f o r e the collision. E* and ft* are the c o r r e -
s p e n d i n g q u a n t i t i e s a f t e r the collision. Z , and E.
r n coh m
i e
are the m a c r o s c o p i c b o u n d c o h e r e n t a n d i n c o h e r e n t cross
sections, e = E-E* is the e n e r g y c h a n g e of the n e u t r o n and
•tfic = m ( v - v T) is the n e u t r o n m o m e n t u m ch a n g e vector.

Th e f u n c t i o n s S ( k ,c ) and S. (k ,e ) are defined by

— inc — J

(2.4)

i Ur-xt/fi)
e G (r,t)drdt (2.5)
s
 Made with Xodo PDF Reader and Editor

The pair distribution functions G s (£> t) and G^Cr,t) de­

termine the dynamics of the scattering systems. G g (r^t) is
defined as the probability that a nucleus which is at origin
at time t = 0 will be present w i t h in a unit volume at posi­
tion _r at time t. G^(r,t) is the probability that a nu­
cleus other than that at origin at time zero will be at p o ­
sition r at time t.

The interference effects of the scattering are contained in

G^(_r, t). These interference effects are important for elastic
scattering in many materials, but for inelastic scattering
in most liquids and polycrystalline materials they can be n eg­
lected, i.e. G^(r,t) can be put equal to zero. Then

(2 .6 )

which is k n o w n as the incoherent approximation. For light water

the incoherent approximation is especially good because
neutron scattering by hydrogen is almost completely inco­
herent (a = 1.8 b and o. = 80 b ) .
coh me

One further approximation is employed in many scattering

models, viz. the Gaussian approximation. A function inter­
mediate b e t w e e n G(^r,t) and S( k ^,g ) is defined by

(2.7)

The Gauss i a n approximation of the intermediate scattering

function is gi v e n by

(2 .8 )
Made with Xodo PDF Reader and Editor

with

~fiu/2kT
f (w)e -not ,
e -1
2ui s i n h ( W 2 k T )

A is the mass of the scattering atoms, k the Boltzmann

constant, T the temperature, w is the angular oscillation
frequency and f(w) the frequency spectrum normalized so
that

f
f(w)du) = 1

For simple scattering models assumes the Gaussian

form and this form is also applied in many sophisticated
models for scattering in water. In order to determine
>'(t)-y(0) the frequency spectrum, f(ui), is estimated from
physical considerations.

2^1.2 _____ § cattering_models_for_water

One of the early models is the Nelkin model [6]. In this

model the hindered rotational m o tion is approximated by a
torsional oscillation. The incoherent approximation is used
for the scattering by hydrogen and the Gaussian approxima­
tion is employed with a spectrum representing a set of
discrete oscillator frequencies. The spectrum is written

f(u>) = II -±
1 - 6(u>-w.) (2.10)
i=l i L

18 ' = 0
A1 =
2.32 •Ftu>2 = 0.06 eV
5.84 = 0.205 eV
A3 = fiw3
2.92 ■fiu>. = 0.481 eV
ll
Ji­>

4
 Made with Xodo PDF Reader and Editor

The first term in the summation represents the translational

motion of the free gas molecules and the second term the
hindered rotation which is assumed to be a torsional oscilla­
tion. The remaining two terms represent vibrational modes.
Aj, is an effective mass of the i:th quantum state.

Defining

S(a,3) H kTe6/2 S( k ,e ) (2 .11 )

with

2
(2 .12 )

-G
6 (2.13)
kT

the spectral density function can be introduced as

p (8) s 26 sinh(8/2 ) Him S(a,B)] (2.14)

rv-vn
a-’-O a J

It is possible in the framework of the Gaussian approximation

to reconstruct the complete S(a,|3) function from p(6) .
Egelstaff and Schofield [7] have proposed a scattering model,
which is known as the effective width model, with

(2.15)

where K^(x) is the Bessel function of the second kind.

The effective width, q, is a measure of the width of p(3).
The model degenerates to that of a free gas as q-K) . The
values of q to be used are determined by correlation to
integral experimental data.
Made with Xodo PDF Reader and Editor

The Haywood model [8] is based o n experimental values of

p(B) for water.

The three scattering models for water mentioned above are

the ones most frequently used in reactor physics calcula­
tions. They are also the models wh i c h have been used in the
present work.

The complete scattering cross section for the water moleeule

is obtained b y the addition of oxygen as a free gas.

2.2 The Doppler effect

The resonance cross section is for nuclei at rest given by

the Breit-Wigner formula

Ex (E ) = Noo r x/IT E 2
4 (E-E ) +r
2
(2.16)
o

/IT 2 rrF_ 4(E-Eo ) R

I (E) = No V - J ---- — - ■ - -S. + No (2.17)
+ r *
E 4 (E-E ) + r 2 L 1

The index x denotes absorption or fission. N is the atom

number density. T , T and T are, respectively, the
x n
width for reaction x , the wid t h for neutron emission and
the total width of the resonance. E
is the resonance
o
energy. K is the reduced de Broglie wave length of the
neut r o n and R is the nuclear radius, o is the potential
scattering cross section and denotes the peak value of a
o
the total resonance cross section which is given by

. ? T'n
n (oE J
(2.18)
°o = *o 8 r—
 Made with Xodo PDF Reader and Editor

The statistical factor g is expressed by

2J+1
g ~ 2(21+ 1)

where I is the spin of the target nucleus and J the

spin of the compound state.

When the nuclei are in thermal motion, the resonances are

broadened as a result of the Doppler effect. The Doppler
broadened cross section can be written [9 ]

(2 .20 )

S(£,e) is the scattering law function defined in Eq. (2.4)

and £ is the m o m e n t u m of the neutron. In most applications
it is assumed that the absorbing nuclei have a Maxwellian
v elocity distribution at the temperature of the medium. At
temperatures above the Debye temperature this approximation
is a good one, because the velocity distribution of the
nuclei is then insensitive to the chemical binding. The
Debye temperature is about 620 K [10] for UO^ and 200 K
for metallic uranium. The effects of crystalline binding on
the resonance absorption will be discussed in § 6.1.

Assuming that the nuclear velocities have a Maxwellian

distribution one obtains [11]

(2 .21 )

I s (E) = N o q - p * ( C ,Y)+Noo | x(?,Y)+Nap ( 2 . 22 )

where the Doppler functions t|>(£,Y) and x(?>Y) are

defined by
Made with Xodo PDF Reader and Editor

g
2/tT

exp - \ g(X-Y)2 ]
dX (2.24)
/iT ■ i+x2

and

A = / H

2 1 2
X e — (E~E) : E = —m v where v is the
T r o r 2 r r
relative neutron-nucleus speed.

The Doppler width, A, represents the effect of temperature

on the shape of the resonance and is a measure of the width
of the Doppler broadened resonance. A is proportional to
the square root of the temperature, T, and the energy, E.

At low temperature A is small so that C is large and

the integral (2.23) becomes

*(C,Y) » — K (2-25)
1 +Y

Inserting (2.25) into (2.21) one obtains the Breit-Wigner

formula for the unbroadened resonance. At the other extreme,
i.e. at very high temperature, t, is small and

U.(?,Y) « ^ x. exp(- \ C2Y2) (2.26)

so that near the resonance peak

 Made with Xodo PDF Reader and Editor

.. X / O /n
S c (E) = N a o — V E ~ T exp

In this Gaussian expression A determines the width.

The shape of a resonance is changed markedly by the Doppler

broadening but the integral o(E)dE is approximately con-
stant

a (E)dE = - a r 4*(C»Y)dY = - t a r (2.28)

i o z o

In deriving the expressions (2.21) - (2.28) for the Doppler

broadened resonance cross sections it was assumed that the
velocity of resonance nuclei is much smaller than that of the
neutron. Except for low energy resonances in light nuclei
this approximation is justified, and it introduces negligible
errors when considering the U-238 resonance absorption.

The reaction rate within a resonance is a (E)<|>(E)dE and

although the integral (2.28) does not change with temperature

the resonance absorption is increased at higher temperature
because the absorption rate is changed as a result of less
marked dips in the flux at Doppler broadened resonances.
If <j>(E) is normalized to <f>(E) = 1/E above the resonance

the quantity 0 ^(E)<j>(E)dE is called the effective resonance

integral and is represented by RI

In the narrow resonance approximation [12] the effective

resonance integral is written

£x (E )£p dE
[Link] =
x Z t (E) E
Made with Xodo PDF Reader and Editor

where is the potential scattering cross section and

I (E) the total cross section. If E/E is set equal to
t o
unity and the interference between resonance and potential
scattering is neglected, one obtains for a single resonance

f Na (r /r ) * (c ,Y ) £
RI = — — - — - — ------------------- E
x E I No !{.(£,Y)-!-!
o3 o p

Defining the function

»(C,Y)
dY C •31)
<KC.Y)+B

RI can be written
x

N'RI
x
= I —
p E
o ('■£
• Ji c .32)

For a series of resonances the total resonance integral is

obtained by summation of Eq (2.32) over all individual
resonances.

The behaviour of J(t,f3) versus 8 is shown in Fig 2.1

(reproduced from [13]). For large values of 6=Sp/NaQ the
flux depression by the resonance is small and J(?,8) is
insensitive to £ , i.e. to the temperature. J(^,6) is
independent of ? also when 8 is small. This is due to
the very strong flux dip at the resonance center causing
most of the absorption to occur at some distance from the
resonance center, where the shape of the cross section is
independent of the temperature.
 Made with Xodo PDF Reader and Editor

The Doppler b r o adening is most important when J(C,8)

varies significantly with £ at a given 3 • From Fig 2.1
this is seen to be the case for 7<j<16 , i.e. in the
range 0.001<8<1 . For U-238 this roughly corresponds to
1 00<a <1 0 5 .
P

2. 3_________ Density effects

The effects of m o d erator density on reactivity are important

for the temperature coefficient in water moderated systems.
The x^ater density decreases with higher temperature causing
the leakage of neutrons and the number of fast fissions to
increase, the. resonance escape probability and the absorption
in the water to decrease and the thermal spectrum to become
harder. These effects are studied in chapter 5.

The effect of a change in water temperature is pronounced

aL high temperature due to the variation of the water
d0
density derivative, — , with temperature (Fig 2.2).

A less important component of the temperature coefficient

is due to the thermal expansion of the fuel, the canning
and construction material. These effects have been investi­
gated for some lattices, the results being given in § 6.6.
 Made with Xodo PDF Reader and Editor

J(E.P)

j (where (3 =2^ x 10"5)

Fig.2.1 The Doppler broadening function J(£.P)

versus p for various £ .
Made with Xodo PDF Reader and Editor
Made with Xodo PDF Reader and Editor

3. THE CELL CODE AE-BUXY

AE-BUXY is the AE-version of the code BUXY (Burnup in xy-

g e o m e t r y ) . It is a spectrum code for pin cells as well as
for BWR and PW R fuel assemblies. In the present work, we
have used only the pin cell option and the description b e ­
low will be limited to this option. A flow chart is shown
in Fig 3.1.

3.1_________ N uclear data library

The data library contains microscopic cross sections in 69

energy groups (Table 3.1) divided into 14 groups in the fast
region above 9 keV, 13 groups in the resonance region between
4 eV and 9 keV and 42 thermal groups be l o w 4 eV. Most of the
calculations performed for this work have been made with a
library based on the U K Nuclear Data File [14]. A library
based on ENDF/B III has b e e n used for some calculations de­
scribed in chapter 7.

M axwellian spectra with 1/E-tails have b e e n used for w e ight­

ing thermal group cross sections. Group averaged cross sec­
tions in the resonance region have been obtained assuming a
1/E-specrum and a spectrum typical for a water system has been
used in the fast region. Effective resonance integrals for
U-235, U-236, U-238 and Pu-239 have been provided through
detailed slowing down calculations for homogeneous mixtures
of the absorber and hydrogen. Resonance integrals are tabu­
lated as function of potential scattering cross section and
temp e r a t u r e .

For the principal moderators scattering matrices are avail­

able based upon alternative theoretical models. We have used
the Nelkin model for water w h e n not otherwise stated. Scat­
tering matrices are tabulated for a representative range of
temper a t u r e s .
 Made with Xodo PDF Reader and Editor

Table 3.1 Energy boundaries i the 69 group library

Group Energy Group Energy

Mev eV
1 10 .0 6 .0655 36 1.097-1.071
2 6 .0655 - 3 .679 37 1.071-1.045
3 3 .679 - 2 .231 38 1.045-1.020
4 2 .231 - 1 .353 39 1.020-0.996
5 1 .353 -0 .821 40 0.996-0.972
6 0 .821 -0 .500 41 0.972-0.950
7 0 .500 - 0 .3025 42 0.950-0.910
8 0 .3025 - 0 .183 43 0.910-0.850
9 0 .813 -0 .1110 44 0.850-0.780
10 0 .1110 -0 .06734 45 0.780-0.625
11 0 .06734- 0 .04085 46 0.625-0.500
12 0 .04085- 0 .02478 47 0.500-0.400
13 0 .02478- 0 .01503 48 0.400-0.350
14 0 .01503- 0 .009118 49 0.350-0.320
eV 50 0.320-0.300
15 9118.0 5530.0 51 0.300-0.280
16 5530.0 3519.1 52 0.280-0.250
17 3519.1 2239.45 53 0.250-0.220
18 2239.45 - 1425.1 54 0.220-0.180
19 1425.1 - 906.898 55 0.180-0.140
20 906.898 - 367.262 56 0.140-0.100
21 367.262 - L48.728 57 0.100-0.080
22 148.728 - 75.5014 58 0.080-0.067
23 75.5014- 48.052 59 0.067-0.058
24 48.052 - 27.700 60 0.058-0.050
25 27.700 - 15.968 61 0.050-0.042
26 15.968 - 9.877 62 0.042-0.035
27 9.877 - 4.00 63 0.035-0.030
28 4.00 - 3.30 64 0.030-0.025
29 3.30 - 2.60 65 0.025-0.020
30 2.60 - 2.10 66 0.020-0.015
31 2.10 - 1.50 67 0.015-0.010
32 1.50 - 1.30 68 0.010-0.005
33 1.30 - 1.15 69 0.005-0
34 1.15 - 1.123

35 1.123 _ 1.097
Made with Xodo PDF Reader and Editor

3. 2_________ Resonance treatment:

A special calculation is performed to determine the effective

resonance integrals in the energy region between 4 eV and
9118 eV. Resonance absorption above 9118 eV is regarded as
being unshielded. The 1.0 eV resonance in Pu-240 and the
G.3 eV resonance in Pu-239 are adequately covered by the
concentration of thermal groups around these resonances and
are consequently excluded from the special resonance treat­
ment. Four nuclides viz. U-235, U-236, U-238 and Pu-239 are
treated as resonance absorbers.

The subroutine for calculation of effective cross sections

in the resonance region is based on an equivalence theorem
[15] which relates the tabulated resonance integrals to the
particular heterogeneous problem. The equivalence theorem is
derived using a suitable rational approximation for the fuel
to fuel collision probability.

The resonance integral in the fuel is in the narrow resonanct

approximation [16]

- 0 (E)
RI =
a (E) *'(1-Pff-)ot + ap P ff-* E ~

where is the absorption cross section, o the total

cross section, j the potential scattering cross section
of the fuel with admixed moderator, and P^ the self col­
lision probability in the fuel region.

Using a rational expression for P^

x b. a
p ff = y——- ; x = — and J S. = 1(3
ff r x+u. o lt 1
l i e i

equation (3.1) can be expressed as a sum of homogeneous

resonance integrals.
 Made with Xodo PDF Reader and Editor

ef f
a = c +a.a
P p i e

and

a = 1 / (NX,) .
e

Z is Che average chord length in the fuel and N the num­

ber density of the absorbing nuclide.

To obtain an expression of the form (3.2) for an infinite

uniform pin cell lattice we write

V . P fb P b f O M
P ff = P ff + l-pKK ( 3 *4)
DD

p^ is the fuel-to-fuel collision probability for an isola­

ted rod, p^ is the probability for neutrons entering the
cell isotropically through the cell boundary to suffer their
first collision in the fuel, p^£ is the probability for
neutrons born uniformly and isotropically in the fuel to
reach the cell boundary uncollided and p., is the probabil-
bb
ity for neutrons entering the cell isotropically through the
cell boundary to traverse the cell without collision.

The probabilities p,, , p. , and p., are calculated

lb br bb
using formulae given by Bonalumi [17] and the final expres­
sion for can be written in the form of Eq (3.2) with
two terms if p^ is given by the Wigner approximation [18]

Pff = x+a (3'5)

a is the so-called Bell factor and is calculated by a poly­

nomial fit [19]. The final expression for the resonance in­
tegral is then
Made with Xodo PDF Reader and Editor

R I = R I, (a +a_o ) + (1--&)RI, (a +aa )

h p £ e n p e

where is a parameter depending only on the materials between

the fuel regions and

nT / \ f P
RI, (o )= j~ ---a =—
dE
h P J ot E

is the resonance integral of a homogeneous medium.

In its original form Eq (3.6) gives the narrow resonance

approximation of the resonance integral. By modifying cr^,
the intermediate resonance approximation [20] is obtained.

The effective cross section in energy group g is given by

[21 ]

RI
° 8 --------- f e - « - 7>
T “
g o eff
P>g

t is the lethargy width of group g.

8

The resonance integral is then corrected for the overlap ef­

fect as described in [21]. The influence of this correction
on the Doppler coefficient is very small and it will not be
discussed here.

3.3_________ Micro group calculation

The mi c r o group calculation is made by use of collision pro­

babilities for the three-region (fuel, canning and modera­
tor) cylindricalized pin cell. It provides 69 group spectra
which are used for energy condensation to broad group cross
sections. These are then used in the macro group and funda­
mental mode calculations.
 Made with Xodo PDF Reader and Editor

I. V. <j>. = I P. .V. ( I I. , <t>. ,+ S. \(3.8)

i,g x r i,g ^ l-s-j.g 3 j.g^-g' j,g' j,g)

S. is the fission source in the fuel region, g is the

J »g
energy group index and i and j the spatial region index
referring to fuel, canning and moderator.

Accurate collision probabilities for the cylindricalized pin

cell with flat source in each of the three regions are cal­
culated using the FLURIG method developed by Carlvik [22].

In evaluating the collision probabilities we start by eli­

minating the z coordinate. With notations according to Fig
3.2

TI
P(x) da = i da j exp(- ■^r^)sin0d8
0

is the probability that neutrons isotropically emitted from

a line source will travel the optical length x within the
azimuthal angle da without colliding.

Defining the Bickley functions

-x cosh u
----------- du
K i»(x) = ,n
cosh u

we get

P ( t) = K i 2 (T ) (3.9)

From (3.9) the collision probabilities in cylindrical geo­

metry are derived
Made with Xodo PDF Reader and Editor

E.V.P. . = •=- | da dy[Ki ( t ..)-Ki ( t . . + t . )

J J 2 IT 3 i] 3 ij j

- Ki ( t . ,+t.)+Ki ( t . . + t . + t . ) ]
3 ij l 3 ij l j

where . , t ..
t and t . are optical lengths within and
i ij J
between the regions under consideration. The numerical
evaluation of P is an integration over the system in the
spatial variable y and in the angular variable a.

Let Pj«-j '3e t'le f i rst flight collision probability within

the three-region pin cell, p, . is the probability for a
neutron born in region j to reach the boundary of the pin
cell and the probability for neutrons isotropically
reflected at the boundary to have their next collision in
region i . p
is the probability for the isotropically
bb
reflected neutrons to pass the pin cell without collision.
The final expression for the
t collision probability
in an infinite lattice is

P. . = p. . + ! i ± V, i (j.U )
i-J i-J l"Pbb

3.4_________ Macro group calculation

Using the spectra from the micro group calculation mac r o ­

scopic cross sections are condensed to a m a x i m u m of 25 energy
groups to be used in the macro group calculation. The condensa­
tion is made by flux weighting of the cross sections.

The macro group calculation is made in the same annular

geometry as the micro group calculation. The DIT method [23],
i.e. a point formalism is used to solve the integral
transport equation
 Made with Xodo PDF Reader and Editor

X is the eigenvalue and x is the fission spectrum. The

transport matrices T give the flux in energy group
>g
g at the point k due to a unit source at £ . The formal
volumes are given by

V2 =

are Gauss weights and r^ is the radius of a point

situated in the interval bounded by and ■

3.5_________ Fundamental mode calculation

The aim of the fundamental mode calculation is twofold.

First, the leakage and the migration area are determined
assuming a fundamental buckling mode. Secondly, the infinite
lattice results obtained from the transport calculation are
modified to include the effects of leakage in predicted
group constants.

The fundamental mode calculation is carried out by use of

the B^-leakage method. P^-scattering matrices are available
in the library for the principal moderators and the P^~
scattering terms are explicitly represented in the B^-
equations.

The fundamental mode equation solved by AE-BUXY is in

matrix form

a-E°)<j> + B 2d(f> = J T* (3.13)

S K
Made with Xodo PDF Reader and Editor

k = the eigenvalue

T , = x , where x is the fission spectrum

88 8 * >8 8
I = diag(I ) (the total cross section)
t >8

£° = £ , (the zeroth moment of the scattering matrix)

® 8 8

d = (3a Z-E1)-1
s'

. (the first moment of the scattering matrix)

s g ^8

ar
(B \
f I fL .)2 . ; B > 0
3 \lJ arctgj

. <
2 2
; K = -B > 0
3 \z J A -1
x % o»8

£ I+ k /Z
A = - l i n ______&
o,8 2k 1 - tc /E
s

B 2 = -K2

All quantities are cell integrated values.

Three fundamental mode calculations are made, namely:

. 2
k00 is obtained with B =0

ke^ is obtained for a given geometrical buckling

2 2
Bis varied so that k ,, = 1. B is then
eff 2
equal to the material buckling B^ .

Equation (3.13) is solved in the macro group structure.

 Made with Xodo PDF Reader and Editor

Fig.3.1 Flow chart of the pin cell calculation in AE-BUXY.

Made with Xodo PDF Reader and Editor

Fig.3.2 Elim ination of the z co-ordinate

in cylindrical geometry.
 Made with Xodo PDF Reader and Editor

4. COMPARISON BETWEEN THEORETICAL RESULTS AND

EXPERIMENTAL INFORMATION

In order to compare the theoretical model for calculation

of the temperature coefficient with experimental informa­
tion, it is desirable to use measurements on simple geomet­
ries, i.e. uniform pin cell lattices. There are not many
papers published about reactor physics experiments on light
water moderated systems at high temperature. We have looked
at some early exponential experiments but found that the ac­
curacy is not good enough for detailed comparisons with theo­
ry, one of the m a i n problems being the nonexistence of an
asymptotic region due to the small size of the subcritical
lattices.

Measurements on light water systems in critical facilities

at temperatures up to 90°C have been performed at several
laboratories, but at temperatures above 200°C we know only
of the experiments made in the KRITZ facility [24] at Studs-
vik and Soviet measurements [25] on highly enriched (m 80 %)
rods in H^O. Power reactor measurements provide valuable in­
formation on the adequacy of the calculational methods, but
the conditions are usually very complex including a great
number of different absorber and fuel rods in different as­
semblies, which requires two- or three-dimensional core cal­
culations. We have therefore chosen to analyse some KRITZ-
experiments in detail. It is very valuable that these m e a ­
surements have b e e n made at temperatures up to 245°C. Com­
parisons have b e e n made on both uniform pin cell lattices
and cores containing BWR or PWR fuel assemblies. Here we
will, however, refer only to the uniform lattices since thest
are the most appropriate ones for the purpose of revealing
fundamental discrepancies b e t ween theory and experiment in
determining temperature effects in reactors.
Made with Xodo PDF Reader and Editor

4.1 D e s c r ip t io n o f t h e m e a s u r e m e n ts

The critical facility KRITZ is used for reactor physics m ea­

surements on water-moderated cores at temperatures up to
245°C. KRITZ consists of a pressure tank with an insert v es­
sel of wh i c h the outer wall is a circular cylinder loosely
fitting into the pressure tank, whereas the inner wall con­
sists of a square cylinder. Vertical and horizontal cross
sections of the reactor tank and insert vessel are shown in
^igs 4.1 and 4.2.

Coarse reactivity control is achieved b y poisoning the water

w i t h bo r i c acid and fine reactivity control is made by ad­
justing the water level. No control rods are involved. It
means that uniform lattice arrangements can be loaded w i t h ­
out any interference with heterogeneous devices except for
spacers wit h i n the core volume.

Critical water levels and flux distributions in the vertical

and horizontal directions are measured and experimental
values of the material buckling are evaluated. The flux
distributions are obtained from activation of copper wires.

Critical measurements have b e e n performed on a large number

of uniform lattices of UC^ and PuC^ rods as well as on more
complex BWR and PWR geometries [26]. The experiments used
in the present study are all on uniform lattices containing
U O 2 enriched to 1.35 % or 1.9 % or mixed oxide (1.5 % PuO^ -
depl U C ^ ) • The moderator to fuel ratio, bor o n contents and
core size are varied.

The fuel rods with b o ttom extensions were standing on stain­

less steel beams and their radial positions were determined
by spacers in form of horizontal straight wires placed in
different planes. The wires in each plane were turned per­
pendicular to those in the neighbouring planes. The distance
b e t w e e n the spacer layers was in the cores with the 1.35 %
 Made with Xodo PDF Reader and Editor

UC> 2 rods 36.0 cm which means that they affect the reacti­
vity. In the cores with 1.9 % U O 2 or Pu ( > 2 rods the spacers
were placed above the critical water level and below the
active length of the rods.

4. 2_________ Description of the calculational methods

The cell code AE-BUXY was used to calculate the material

buckling and for a geometrical buckling given in input.
When the input geometrical buckling is put equal to the
experimental material buckling, the deviation of from
unity gives a measure of the d i s crepancy between AE-BUXY
and experiment. This comparison b etween theory and experi-
2
ment is relying on the experimental material buckling, B ^ e x p )
obtained from copper activation. Alternatively, cal­
culations may be carried out on the whole core using material
composition, geometry and critical water level as input. An
accurate calculational model requires a three-dimensional
code. Three-dimensional calculations are, however, expensive
and have b e e n avoided in this work. We have instead simulated
the experimental configuration w ith a two-dimensional diffusion
theory code in xy-geometry, DIXY [271, using the measured
axial buckling to account for the axial leakage. The largest
2
uncertainty in the experimental B^ is due to the deter­
m i n a t i o n of the extrapolation length from the flux mapping.
The m e a sured extrapolation lengths have been compared with
calculated ones. These calculations were carried out in
one d i m ension by use of the S
code DTF-4 [28]. DTF-4 was
n
also used to calculate the reactivity worth of the spacers.
Few-group constants for DIXY and DTF-4 were generated by
AE-BUXY.

Cross sections for the radial water reflector were also

generated by use of AE-BUXY. The who l e core was cylindricalized
and a transport calculation was carried out on the cluster
Made with Xodo PDF Reader and Editor

(containing 1500-2000 fuel rods) surrounded by a thick

water region. Few group reflector cross sections are p ro­
duced at different distances from the fuel region. It was
found that the 4 and 8 group cross sections do not vary
much with the distance from the fuel region. Average reflec­
tor cross sections were therefore used in the core calcula­
tions .

All pin cell calculations were carried out w i t h 17 energy

groups in the macro group calculation of AE-BUXY and with
3 Gauss points in the fuel, 2 in the canning and 3 in the
cylindricalized water region. The same 17 groups were used
in the fundamental m o d e calculation. The choice of the above
parameters is discussed in chapter 6.

The DIXY calculations were made in 4 energy groups (10 MeV-

- 0.5 MeV - 9118 eV - 0.625 eV - 0) in the 1.35 % U 0 2
lattices and with 6 groups (10 MeV.- 0.5 M e V - 9118 eV -
- 4 eV - 0.625 eV - 0.140 eV - 0) in the 1.9 % U 0 2 and P u 0 2
lattices. The number of mesh points was chosen from a few
calculations where the distribution and total number of mesh
points were varied (see chapter 6).

Axial calculations with DTF-4 were made in the S^-approximation

w i t h 4 energy groups and including anisotropic scattering
explicitly.

4• 3_________ The reactivity worth of spaces

The reactivity worth of spacers need be taken into account

only in the cores w i t h the 1.35 % U 0 2 fuel. In these lattices
horizontal straight wires of stainless steel, 0.40 cm in
diameter, placed in different planes were used to determine
the radial positions of the fuel rods. The spacer plane
separation was 36.0 cm.
 Made with Xodo PDF Reader and Editor

The reactivity worth of the spacers and its temperature

dependence was calculated by use of DTF-4 and AE-BUXY. An
example of the geometry used in DTF-4 is shown in Fig 4.3.
The axial direction of the core and bottom- and top-reflec-
tors are considered. The radial leakage istaken into accouut
by use of a transverse buckling equal to the measured radial
buckling B2 = B2 + B2 .
r x y

Homogenized macroscopic cross sections were calculated with

AE-BUXY. The calculation of cross sections for the pin cell
region (material 4 in Fig 4.3) and the region above the water
level (where the [Link] between the rods consists of H^O-
steam) (material 6) is straight forward. To be able to
handle the spacer region (material 5) the SS-wire has to be
homogenized with the water.

Some results from calculations of the spacer reactivity

worth are given in Table 4.1.

Table 4 . 1 : The influence on reactivity from spacers in

1.35 % U ( > 2 lattices calculated by DTF-4

Boron Temperature AB2

Akeff m
conc
(ppm) C°c) (pcm) (nf2)

O 90 - 500 - 1.3
210 - 560 - 1.3

175 90 - 410 - 1 .0
210 - 480 - 1.1

Experimental buckling effects of spacers were determined

only at 20 °C and 0 p p m boron. The experimental result is
2 -2
[24] AB
= -1.3+0.2 in . This is in good agreement with
m —
the calculated value.
Made with Xodo PDF Reader and Editor

^ 4 ^ 1 ___ Lat tices_wi th_l_135_%_enr iched_UQ 2

A great number of measurements has been performed with the

fuel rods containing 1.35 % enriched UC^ [24]. Two cores
have been analysed in detail. The first one contains 39x39
pins and is without boron in the water. The seconc core con­
tains 46x46 pins and the boron concentration is 175 ppm.
The moderator to fuel volume ratio, V A ' , is 1.4 and the
m o
measurements cover the temperature range 20-210 C.

Results of the analysis are collected in Tables 4.2-4.4.

Table 4.2: Calculated k and k for the 1.35 % U0„ lattices

---------- co e ff 2

Core size Boron Temp. AE-BUXY DIXY

cone.
(ppm) (°C) k 03 k k eff
«
eff

20 1.16294 0.99402 0.99381

90 1.15566 0.99091 0.99144
39x39 0
160 1.14677 0.98773 0.98876
210 1.13872 0.98587 0.98673

20 1.12172 0.99636 0.99569

90 1.11621 0.99365 0.99383
46x46 175
160 1.11003 0.99186 0.99198
210 1.10448 0.99100 0.99102

It is seen in Table 4.2 that the reactivity is slightly

underpredicted. The deviation of k e££ from unity increases
with temperature, i.e. the predicted temperature coefficient
is too negative.
 Made with Xodo PDF Reader and Editor

Ak _j-/AT in Table 4.3 is defined as the difference between

eff
kg ££ at two temperatures divided by the temperature differ­
ence. Thus A k ^ ^ / A T gives the discrepancy between the cal­
culated and m e a sured temperature coefficient.

Table 4.3: Ak _C/AT derived from Table 4.2

---------- eff

Core size Boron Temp. A k g ^^/AT (pcm/°C)

conc. interval
(ppm) (°C) AE-BUXY DIXY

20- 90 -4.4 -3.4

39x39 0 90-160 -4.6 -3.8
160-210 -3.6 -4.2

20- 90 -3.9 -2.7

46x46 175 90-160 -2.6 -2.7
160-210 -1.8 -2.0

The calculated temperature coefficient is (2-4) pcm/°C

more negative than the experimental one. There are only
small discrepancies between AE-BUXY and DIXY.

k ^ £ was calculated using both DIXY and DTF-4 at 90 °C and

210 °C. In DIXY the experimental axial buckling was used,
and in DTF-4 calculations were carried out using the
experimental radial buckling to give the transverse
leakage. As seen in Table 4.4 the agreement between the
different calculations is very good. This means that m e a ­
sured and calculated axial and radial bucklings are consistent.

Table 4 . 4 : Comparison between calculated by use of

AE-BUXY, DIXY and DTF-4

Bcron Code k
eff S k « f f /S T
conc.
(ppm) 90 °C 210 °C (pcm/°C)

AE-BUXY 0.9909 0.9859 -4.2

0 DIXY 0.9914 0.9867 -4.0
DTF-4 0.9898 0.9845 ■ -4.4 '
Made with Xodo PDF Reader and Editor

The cores containing 1.9 % enriched IK^ rods and PuO^ rods
are regular square lattices with either only UC^ or PuC^
rods or with a central square zone of PuC> 2 rods surrounded
by a U O 2 zone. Three different volume ratios were covered
by the experiments [26] with temperatures from 20 °C up to
245 °C. Lattice data are given in Table 4.5.

In the AE-BUXY calculations the plutonium was assumed to be

homogeneously distributed in the fuel. In reality the plu­
tonium fuel used in these experiments contains Pu 0 2 _particles
whose average size is about 25 pm. This non-homogeneity of
U 0 2 ~Pu 0 2 needs to be considered because of the self shielding
in the P u 0 2 ~ p a r t i d e s . As a result of the shielding the
Pu-239 fission and capture reaction rates will be reduced.
Because of the larger proportionate increase in the shielding
of the 0.3 eV resonance the a-value of Pu-239 is reduced.
The change of the fission rate results in a reduction in
reactivity and the change of the a-value results in an
increase in reactivity. Further, the reactivity is increased
because of the shielding of the 1 eV resonance of Pu-240.
The net reactivity change is in all our cases negative.

The self shielding in Pu-239 was estimated assuming that

the incurrent into the Pu 0 2 _particles has the same spectrum
as the average flux in the fuel. The corrections for
self shielding in Pu-239 are in our cases -(300-600) pcm
and for self shielding in Pu-240 about +100 pcm. Our correc­
tions were found to be in good agreement with calculations
for room temperature presented by Liikala et a l . [29].
Finally the statistical weight of the plutonium zone was
 Made with Xodo PDF Reader and Editor

Table 4 .5 . L a t t i c e d a t a and c o r r e c t io n s f o r therm al expansion and p a r t i c l e s iz e p lu s c a lc u la t e d

i
1
Ak ,,/A T
Fuel Core Pu zone Boron Temp C o rr e c tio n s eff
m r ef f
s iz e conc
therm al p a rtic le
expan* s iz e
(ppm) (°C) (pcm) (pcm) (pcm/°C/

1.2 34x34 _ _
1.9% U0o 0 22.3 0 0.99547
c. 1! - 0 90.0 +65 - 0.99466 -1.2
44x44 _ 0 204.9 +175 - 0.99320 -1.3
n — 0 245.2 +214 — 0.99228 -2.3

1.7 27x27 — 0 20.0 0 — 0.99430 -

ii — 0 88.8 +12 — 0.99309 -1.8

1.7 36x36 — 300 23.0 0 — 0.99551

40x40 - 200 208.3 +30 - 0.99193 -1.9
n — 200 247.2 +38 — 0.99147 -1.2

0.99662 __
PuC2- 1.8 40x40 24x24 300 18.1 0 -280
rt If
1.9% U0o 300 90.2 +33 -260 0.99670 +0.1
z ii 1!
0 211.3 +83 -140 0.99481 -1.6
i: II
0 245.5 +104 -130 0.99568 +2.5

34x34 24x24 350 21.8 0 -410 0.99791 —

2.5
ii ii
350 90.5 +17 -380 0.99810 +0.3
i» u
250 247.0 +54 -300 0.99563 -1.6

26x26 26x26 50 23.0 0 -500 0.99691 ii

Pu0o 3.3 i
JL m ii
50 93.4 -16 -460 0.99726 +0.5 !
M it
5.0 208.4 -43 -400 0.99541 -1.6
II if
50 239.2 -51 -380 0.99571 +1.0
<

* c f § 6.6
Made with Xodo PDF Reader and Editor

used to find the corrections in cores containing both PuC^

and U O 2 pins. The corrections are shown in Table 4.5. The
temperature coefficient is increased by about 0.5 pcm/°C
due to the particle size reactivity effect.

Table 4.6 shows the results from AE-BUXY.

Table 4 . 6 : Results from AE-BUXY for the 1.9 % UO 2 and

the PUO 2 lattices

Fuel Vm v 1f Boron Temp. k 00 k Ak /AT

ef £ eff
konc.
(ppm) (°C) (pcm/°C)

1.9 % U 0 2 1.2 0 22.3 1.23514 0.99657

-5.2
90.0 1.22490 0.99309
-0.9
204.9 1.19884 0.99205
-2.1
245.2 1.18559 0.99123

1.7 0 20.0 1.26328 0.99938

-5.3
88.8 1.25663 0.99575

1.7 300 23.0 1.19120 0.99746

-3.7
200 216.2 1.19693 0.99035
-1.1
200 247.2 1.19125 0.99000

Pu0„ 1.8 300 18.1 1.22217 _

z
300 90.2 1.22039 -
0 211.3 1.25546 -
0 245.5 1.24654 -

2.5 350 21.8 1.20305 -

350 90.5 1.20867 -

250 247.0 1.23400 -

3.3 50 23.0 1.25822 0.99795

-2.9
50 93.4 1.26545 0.99594
-3.0
50 208.4 1.27912 0.99254
-1.0
50 239.2 1.28120 0.99224

Comparing the results in Tables 4.5 and 4.6 one finds that
the calculated k ^ ^ - v a l u e s at room temperature given by
AE-BUXY are higher than those given by DIXY and at high
temperature the AE-BUXY values are lower. Thus, the experi-
2
mental B corresponds to a less negative temperature
m
coefficient than what is obtained from the diffusion theory
calculations on the whole core.
 Made with Xodo PDF Reader and Editor

At room temperature DIXY gives « 0.995 for the clean

UO^ lattices and k ^ ^ « 0.997 for lattices containing PuO^.
Corresponding reactivities at high temperature are 0.992
and 0.996, respectively. The spread in k ^ ^ is small and the
trend in k ^ ^ versus temperature is smaller than for the
1.35 % UO^ lattices.

There are several sources to experimental uncertainties.

The uncertainty in the m e a sured bor o n concentration is 1 %
w h e n there is m o r e than 100 p pm boron in the water. 1 ppm
boron corresponds to about 20 pcm in reactivity. The error
2 .
in the measu r e d B is expected to correspond to less than
z
100 pcm in the absolute value of k ^ ^ . Further, there is
the uncertainty in material composition and geometry. The
experimental uncertainty in the difference between two cal­
culated k ^ ^ - v a l u e s for the same core (but at different
temperatures) is expected to be about 50 pcm.

The last column of Tables 4.5 and 4.6 gives the discrepancy
between the measured and calculated temperature coefficient.
When looking at A k ^ ^ / A T it should be remembered that the
uncertainty in A k ^ ^ / A T due to experimental errors m a y be
1 pcm/°C or more. The uncertainty becomes less if one cal­
culates A k g j^/AT for the whole temperature range (Table 4.7).

Table 4 . 7 : A k ^ ^ / A T for the temperature intervall 20-245 °C

Fuel V m V tf A k e f f /AT (pcm/°C)

AE-BUXY with exp B 2 DIXY

m
1.9 % U 0 o 1.2 -2.4 -1.4
1.7 -3.3 -1.8

P u 0 2- U 0 2 1.8 - -0.4
2.5 -1.0

Pu02 3.3 -2.6 -0.6

 Made with Xodo PDF Reader and Editor

The present calculations have been performed neglecting

the difference between the uranium and plutonium fission
neutron spectra. The latter one has a higher average
energy. This gives an increased leakage (which causes a
negative correction to anc^ increased fast fission
(positive c o r r e c t i o n ) . The importance of the plutonium
fission neutron spectrum is discussed in § 6.7. The influ­
ence on the temperature coefficient is negligible in all
lattices studied in the present report.

4^4^3___ Comments to_the_resuits

The spread in predicted values is small. The average

value is 0.995 for cold UO^ lattices and 0.997 for PuO^
lattices. versus temperature is shown in Fig 4.4.
AE-BUXY predicts a temperature coefficient which is
(1-4) pcm/°C too negative. This discrepancy is somewhat
smaller than what has been observed in power reactor applica­
tions t1]. The magnitude of the temperature coefficient is
(10-20) pcm/°C at room temperature and about 40 pcm/°C at
high temperature. The discrepancy between measured and
2
calculated values is slightly larger w h e n measured are
used to calculate k than what is obtained from two-
eff
dimensional diffusion theory. Concentrating o n the DIXY
results it is seen that Ak ri./AT varies little with temper-
etr
ature.
 Made with Xodo PDF Reader and Editor

TO COVER
GAS SUPPLY

NON-RETURN
VALVE

IN-LET FOR WATER

CIRCULATION

U 02 REGION

PROBE OF SQUARE-FORMED
•WATER SPACE FOR
LEVEL EXPERIMENTS
METER

DUMP SPACE
ADJUSTABI
WATER (4 SEGMENTS)
LEVEL

POSITION OF
NEUTRON SOURCE
(SB - B E )

SPRING-LOADED
SAFETY SHUTTER
(7 TOTALLY)

DRAINAGE OF DUMP
SF*CE

Fig 4.1 Vertical cross section of reactor tank with insert vessel
 Made with Xodo PDF Reader and Editor

Pressure tank

Dump space
(four communi­
Safety shutter
cating segments) (seven totally)

Fig 4.2 Horizontal cross section of the insert vessel.

The lines of dots and dashes give the boundaries of the
two core sizes investigated for the 1.35 % U 0 2 rods.
(The dash circles indicate the size and position of
sealed openings in the tank lid. The shaded circular
areas show the position of the neutron detectors.)
 Made with Xodo PDF Reader and Editor

Axial representation
in DTF -4

Mat. 7 (SS ♦ Steam)

Mat. 6 (U02 + Zr ♦ Steam)

Mat. 4
Mat. 5
Mat. U
-Mat. 5
Mat. U

Mat. 5 ( H,0 +U0, +Zr+:

Mat. 4 (H20 ♦ U02+Zr)

Mat. 3 (H20 +SS+Zr)

Mat. 2 (H20 +SS)

Mat. 1 (H20 )

Fig .L 3. Axial representation of and reflectors in DTF-4.

 Made with Xodo PDF Reader and Editor

keff

Fig. 4. A keff v e r s u s te m pe ra tu re c a lc u la t e d by DIXY.

 Made with Xodo PDF Reader and Editor

5. The components of the temperature coefficient

Many phenomena in a reactor system contribute to the temper­

ature coefficient of reactivity. Lack of agreement between
theoretical and experimental values of the temperature coeffi­
cient, dk/dT, may therefore be due to errors in the calcula­
tion of one or several of the different components. Further­
more, agreement obtained in a limited study does not neces­
sarily mean that the calculations are correct in all respects.
Compensating errors may appear and care must be exercized in
drawing conclusions.

Neglecting the effect of thermal expansion, except the water

density effect, dk/dT may be written

dk(T ,T ,T ,0)
3k 3k d0
(5.1)
dT 3T 36 dT

T^, T £ and T ^ = fuel, canning and moderator temperature,

respectively.
0 = moderator density.

Theoretical values of the terms on the RHS of Eq (5.1) may

be obtained by using a suitable advanced code such as
AE-BUXY. A corresponding experimental determination is
difficult to achieve. Attempts have been made to measure
the last term of (5.1) using flashing experiments in KRITZ.
The accuracy of these measurements, especially the determina­
tion of the void content and void distribution, is, however,
not high enough to allow a detailed comparison with the
theoretical results.

Other experimental methods to separate the total temper­

ature coefficient into temperature and density effects have
been used elsewhere, e.g. simulating void by use of aluminium
Made with Xodo PDF Reader and Editor

tubes in the moderator. The introduction of aluminium

causes, however, other problems, e.g. the not negligible
fast scattering in A1 and streaming effects in the tubes.

Another way to obtain experimental information of the

separate terms in (5.1) is to perform critical measure­
ments within a wide temperature range. The factor d0/dT
in the last term is much more sensitive to the temperature
(cf Fig 2.2) than 8k/3T and 3k/30 , so the last term in
(5.1) gives a significantly higher contribution to dk/dT
at high temperature than at low temperature. The KRITZ
measurements of dk/dT which cover a wide range of temper­
atures therefore provide important information, which may
help to reveal whether the discrepancy between theory and
experiment is due to pure temperature effects, water
density effects or both.

Theoretical values of the terms in Eq. (5.1) are given

in § 5.1.

In order to gain some insight into the physical significance

of the calculated coefficients, it is helpful to express them
in terms of the derivatives

dn df dp dL
_ & & — i. and — £.
dT dT dT dT

n is defined as \>Z,/£ in the fuel, f is the ratio

g f a g
between fuel absorptions and cell absorptions, p is the
2 . .®
resonance escape probability and L the diffusion length.
Contributions to dk/dT from each of the derivatives (di­
vided into four energy groups, g, have been calculated using
a special code, COEFF, with four-group cross sections from
AE-BUXY as input. Results from these calculations are dis­
cussed in § 5.2.
 Made with Xodo PDF Reader and Editor

Looking at E q . (5.1) we note that the term 9k/8Tc in a light

water moderated system with zircaloy canning can be neglected
compared to the other terms. In the 1.35 % UO 2 lattice for
example, AE-BUXY gave the very small contribution

3k^
Y f ~ = “0*02 pcm/°C
c

and this term will be neglected in the following discussion.

Tables 5.1-5.3 show the terms of Eq. (5.1) as predicted by

AE-BUXY. Results are given for the temperature coefficient
of the infinite multiplication constant, Ak^/AT , the
reactivity for a critical lattice, Ap/AT , the material
2 2
buckling, ABm /AT , and of the migration area, AM /AT .
The temperature intervals used in the calculations are 20 °C
to 90 °C and 160 °C to 210 °C. In this temperature range
experimental values of the total temperature coefficients
are available.

The fuel temperature coefficient is in all lattices less

negative in the temperature interval 160-210 °C than in the
interval 20-90 °C. This should be expected due to the
approximate T
- 1/2 . .
dependence of the Doppler coefficient.
8p/3T^ is built up of two components. The dominating part
is caused by the Doppler broadening of resonances and a
minor contribution is given by the thermal scattering in the
oxygen. In the 1.35 % UO^ fuel these two components were
determined separately and we obtained

(I£_\ = -0.3 pcm/°C

\3Tj
f'thermal

for the temperature interval 20-90 °C and

 Made with Xodo PDF Reader and Editor

Boron
co n te n t 9X/£Tf 3X/3T 3X/36 *d0/dT dX/dT
m
(ppm)

Temp i n t e r v a l (°C) 20-90 160-210 20-90 160-210 20-90 160-210 20-90 160-210

X=k00 0 -4.8 -4.3 -3.8 -3.6 -1.8 -8.2 -10.4 -16.1

(pcm/°C) 175 -4.7 -4.1 -3.3 -3.3 +0.1 -3.7 -7.9 -11.1

X=p 0 -4.1 -3.6 -5.1 -4.7 -10.6 -29.0 -19.8 -37.3

(pcm/°C) 175 -4.1 -3.7 -4.3 -4.0 -6.9 -20.2 -15.3 -27.9

X=B2 0 -.0114 -.0088 -.0137 -.0108 -.0261 -.0598 -.0512 -.0794

(n fV c .) 175 -.0113 -.0086 -.0113 -.0092 -.01.64 -.0406 -.0390 -.0584

X=M2 0 -.0004 -.0004 .0049 .0048 .0253 .0732 .0298 .0776

(cm2/°C ) 175 -.0004 -.0004 .0054 .0046 .0251 .0730 .0301 .0772
 Made with Xodo PDF Reader and Editor

Vm/Vf Boron
m t
conc 3X/3Tf 3X/3T 3X/30‘ d0/dT dX/dT
(ppm) m

Temp i n t e r v a l (° C) 20-90 160-210 20-90 160-210 20-90 160-210 20-90 160-210

X=k 00 1 2 200 -5.8 -5.1 -2.9 -2.8 -4.0 -13.8 -12.6 -21.7
(pcm/°C) 1 7 200 -4.6 -4.1 -2.6 -2.9 +0.7 -2.4 -6.5 -9.4

X=p 1 2 200 -4.8 -4.2 -3.7 -3.5 -13.2 -34.3 -21.7 -42.0
(pcm/°C) 1 7 200 -3.8 -3.4 -4.1 -4.1 -10.9 -30.3 -18.9 -37.8

X=B2 1 2 200 -.0124 -.0096 -.0095 -.0080 -.0322 -.0724 -.0542 -.0900
m
(m-2/°C) 1 7 200 -.0110 -.0086 -.0117 -.0100 -.0291 -.0690 -.0518 -.0876

X=M2 1 2 200 -.0007 -,0008 +.0032 +.0034 +.0255 +.0728 +.0281 +.0754
(cm2/°C) 1 7 200 -.0004 -.0006 +.0042 +.0042 +.0261 +.0768 +.0300 +.0804
 Made with Xodo PDF Reader and Editor

Boron
i Vm Vfr
i conc 3X/8Tf 3X/3T dX/ZO ‘ dQ/dT dX/dT
j
t (ppm) tn
i i
Temp i n t e r v a l (°C) 20-90 160-210 20-90 160-210 20-90 160-210 20-90 160-210
it
X=k00 1.8 300 -5.1 -4.6 +4.0 +5 .0 -1.3 -9.5 -2.4 “ 9.1

(pcm/°C) 2.5 250 -4.1 -3.7 +7.6 +8-7 +2.7 +0.1 +6.2 +5.1
3.3 0 -3.5 -3.1 +7.7 +9.0 +3.7 +2.5 +8.0 +8.4

X=p 1.8 300 -4.1 “ 3.7 +2.5 +3.8 -13.7 -39.5 -15.3 -39.4
(pcm/°C) 2.5 250 ■3.3 -3.0 +5.1 +6.7 -11.4 “ 36.5 -9.6 -32.8
3.3 0 -2.7 -2.5 +4.1 +6.2 -14.4 -44.3 -13.0 -40.5

J
i
X=B2 1.8 300 -.0124 -.0098 +.0075 |+.0100 -.0394 -.0972 -.0443 -.0970
m
(m 2/°C) 2.5 250 -.0107 .-0082 +.0164 1+.0186 -.0351 -.0952 -.0294 ! -.0848
i
3.3 0 -.0092 -.0072 +.0140 !+.0182 -.0467 -.1210 -.0420 | -.1100
1 i
r i
X=M2 1.8 300 -.0004 -,0006 +.0018 +.0012 +.0258 I +.0756 +.0272 +.0762
i
(cm2/°C) 2.5 250 -.0002 -.0002 +.0025 +.0016 +.0262 | +.0784 +.0285 +.0798
1
3.3 0 -.0002 -.0002 +.0034 +.0020 +.0263 j +.0812 +.0300 +.0830
—. ___ i__ ____ _
 Made with Xodo PDF Reader and Editor

v3 :f7
1 Doppler
, = - 3 -3 ^ ° c

GO j. thermal
= -0.3 pcm/°C

for the interval 160-210 °C.

The moderator temperature coefficient is negative in the

U O 2 lattices and positive in the PUO 2 lattices. The different
signs of 9p/3Tm for the two types of fuel are due to the
0.3 eV resonance in Pu-239 and the non-l/v-dependence of
the thermal U-235 cross section. Thus, an analysis of both
UO2 and PUO 2 systems constitutes a severe test of the
ability to predict the thermal neutron spectrum. An error
in the calculated temperature dependence of the spectrum will
give different errors in the predicted temperature coefficient
for U O 2 and P U O 2 lattices. Our analysis of the KRITZ experi­
ments has, however, given approximately the same discrepancy
between calculated and measured temperature coefficients in
both U O 2 and P U O 2 systems. This strongly indicates that the
change of the thermal neutron spectrum with temperature has
been correctly calculated.

The large temperature dependence of the water density

coefficient is a consequence of the variation of d0/dT with
temperature. The water density, 0, and d0/dT versus
temperature are shown in Fig. 2.2. The average value of
d0/dT is -0.470*10"3 g/cm3 *°C and -1.096-10_3 g/cm3 *°C
in the temperature intervals 20-90 °C and 160-210 °C,
respectively. 3Bm /30*d0/dT is roughly proportional to
d0/dT , whereas the changes of 3p/30»d3/dT and
3M2/30'd0/dT are larger than they would be if they
were proportional to d0/dT . This is due to spectrum
effects. 3 ^ / 3 0 - d0/dT has a more complex variation
with temperature because Bk^/ST is built up from several
positive and negative contributions (cf § 5.2).
Made with Xodo PDF Reader and Editor

The discrepancy between the theoretical and the experimental

reactivity can be described by a function f(T,0) defined
by

[Link](T' 9> ’ ‘W r y (T-e) + £« ’ 6> <5-«

We found in chapter 4 that f(T,0) , within reasonable

error limits, is independent of temperature, i.e.

3f , 3 f d0 _ ,
3T 36 dT " const* (5.3)

There are no physical reasons why the twoterms ofthe LHS

should be correlated so we can expect both terms to be con­
stant. This means that

3f /de\ 1
— = const^— j (5.4)

There is also no reason to believe that 3f/30 is inversely

proportional to d0/dT so (5.4) leads to

(If 3f/33 f 0 we would have 3f/30-*<» at a small extrapola­

tion in the temperature to T = 4 °C where d0/dT = 0.)
Thus we find that if (5.3) is true, which is approximately
the case in the investigated lattices, then the discrepancy
between calculated and measured temperature coefficients
must be expected to be a pure temperature effect and not a
secondary effect due to the temperature dependence of the
water density.
 Made with Xodo PDF Reader and Editor

5.2______ Contributions to the temperature coefficient

_________ calculated by COEFF_____________________________

5.2.1_ _Descrigtion_of COEFF

The p r o g r a m COEFF was writ t e n as a tool for studying the

contributions from various parameters (n, f, P and L^) to
the temperature coefficient. Input data consist of few-group
parameters produced by AE-BUXY. These data refer to the
criticality spectrum obtained [Link] fundamental mode cal-
2 2
culation w i t h B = B
m

The following nomenclature will be used to describe COEFF.

G number of energy groups

g group index
fission neutron spectrum
Xg
Z cell averaged absorption cross section
ag
vE cell averaged v - f i s s i o n cross section
fg
cell averaged scattering cross section (from g to g')
g *-g
D cell averaged diffusion constant
g
cell integrated neutron flux
‘
"g

B2 material buckling
m

B2 geometric buckling

<j> and y are normalized such that

g g

E
g
xg = 1
s

?(I“s + D
g
B 2)<|> = 1
g
Made with Xodo PDF Reader and Editor

Define

R = f" Y (E .. ./d> , — E , „ d> „)

8 g^ = l g H_g+ i g ' ^ ' V g g

R = 0
o

rg =o

vE
nf =
fg (5.7)
ag

P = (5.8)
8 R g“l + Xg

PG = 0

T he n e u t r o n b a l ance for group g can then be w r i t t e n

(E
ag
+ D B ) <J) = x
g m *g Ag
+ R
g-1
i “R
g
(5.9)

N o t i n g that R = P (R + x ) w e g et
g g g g

(E
ag + D g B m') *g
<j> = (1 - P„) (x
g ' VAg + R g-1'
,) =

D g- 1
: = (i+ b 2 ) L (i - p ) ) x . fl P„u (5.10)
38 8 ag m 8 g ^= l 8 g "= g ' 8

whe r e

g-1
n P „ = 1 for g' = g
g^g' 8

Wr i t e

k = ------ S------ T------ = I v Z ( <f> =7’

nf z <J> (5.11)
g ag + g "> Og 8 8 8 8 8 a§ 8
 Made with Xodo PDF Reader and Editor

k = E d - B2 ) " 1 nf (1 - P ) f X , V P „ (5.12)
g ag 8 8 g'=l 8 g " = g ’ 8

This expression may be written

8-1
n p „„ (5.13)
V
g 1 + L 2g B 2m g ’
-l 8 g"=g

with

D
(1 - P )
L 2 = ^ — --------- (5.14)
8 e 2
1 + -=2- P B
£ g m
ag

P 1 = (1 + L 2 B 2 ) P (5.15)
g g m g

Pg is the slowing down probability in an infinite lattice.

In the special case with P 1 = ---- — (£ = V £ , )
g £ ag + £ rg rg g V jl
Fg g •‘
-g
we get

, 2 ______
8 Eag + Erg

2 2
i.e. the conventional definition of L . Replace B with the
2 m
geometrical buckling, B , in (5.13). W e obtain

k = T 5k (5.16)
8 S

nf (1 - P ) g g-1 P en
Sk = — §— y x . n ---- &=— =• (5.17)
8 1 + L2 B2 g^l 8 g'-g' 1 + L „ B
o o

(5.16) now gives the effective multiplication constant, k,

2 . 2
at a given geometrical buckling, B , (k^k^ if B =0) and (5,
gives the contributions to k from each energy group.
Made with Xodo PDF Reader and Editor

To see how a change in n . f , p1 and L effects k we diffe-

g g g g
rentiate (5.16)

Ak = T An f X+5"Af n X +y'AP1 Y +^AL2 Z (5.18)

“ ggg“ g g g ^ g g g g g

with the coefficients X , Y and Z given by Table 5.4.

g g g

Table 5.4: Coefficients in equation (5.18)

1 i r- 8 1 P g"
X
g 2 2 (1- y I v „n , 2 2
1 + LT B 8 g^=l 8 g"=g' 1 + l/„ B
©. O

1 ^ 8~^ PE »
Y
g 2 2 11 2 ? +
1 + i/ B 8 g'=l 8 g"=g' 1 + l/„ B •
8 8

! G g g"'-l Pg"
+ i L 2 2 I ^c1 n 2 2
g"’
=g+l 1+L‘„BZ 8 8 gV =l 8 g"=g' l+l/„B
8 8 8

b2 i 8 8 -1 pi"
Z
g 2 2 2 nfg (1_Pg) I V " *
(l+LgB ) 8 8 gV =l 8 g"=g' 1+L2 B
o o
B2 ° 1 ~i ^ 8
2 2 L 2 2 P o"') C.
1+l V g"'-g+l l+L2,,,B 8 8 gV =l 8
o o
g’
"-l pi,
n —
g"=g' l+lr,,B
o

G
]T which appears in the second term of Y and Z is put
g - 'o g + i § 8
equal to zero for g = G.

5^2^2___ Results_from_calculations_with_COEFF

The coefficients An /AT , Af /AT , Ap /AT and A L 2/AT

g g g g
and their influence on the partial and total temperature
coefficients have been calculated using four energy groups
with the boundaries 10 MeV - 0.5 MeV - 9118 aV - 0.625 eV -
 Made with Xodo PDF Reader and Editor

The results have been collected in Figs 5.1-5.12. There is

no experimental information available which corresponds to
these theoretical results. The value of this kind of calcula­
tions is that they give a detailed picture of the temperature
dependence of the neut r o n balance and its influence on the
reactivity coefficient.

The fuel temperature coefficient is mainly caused by changes

in n and p in the resonance group (Figs 5.1, 5.5 and
5.9). A small contribution to 3p/3T^ comes from 3n/3T^
and 3f/3Tf in the thermal group due to the thermal scatter­
ing in the oxygen. The thermal contribution to 3p/3T^ is
0.2-0.3 pcm/°C. The increased resonance absorption gives
rise to a very small positive contribution due to decreased
leakage. This effect is seen in 31,2/3T^. The numerical
value is about 0.1 pcm/°C. It is also seen in the diagrams
that the Doppler effect is less in the high temperature
interval than in the low one.

Figs 5.2, 5.6 and 5.10 show the contributions to the m o d e r a ­

tor temperature coefficient. In the IK^ lattices there is a
large negative effect caused by 3n^/3Tm and a smaller
positive effect caused by 3f./3T > so that the total thermal
4 m

temperature because the water density is lower. In the P u 0 2

lattices there is also a negative contribution to 3p/3T
m
due to 3ri,/3T , but the positive influence of 3f,/3T is
4 m '* m
larger than the negative contribution from Sn^/ST , so
that the coefficient, Sp/ST^, is positive. The thermal leakage
increases with higher temperature due to the harder spectrum.
This leakage effect is nearly proportional to B and is
m
(1-2)pcm/°C.

The water density influence on the temperature coefficient

is shown in Figs 5.3, 5.7 and 5.11. This effect is, as we
already have seen in § 5 . 1 , larger at high temperature than
at low one due to the temperature dependence of d0/dT .
Made with Xodo PDF Reader and Editor

The density change contributes to the reactivity coeffi­

cient in all energy groups, the largest contributions being
from 9n^/36'd0/dT (positive due to less absorption in 1^0
and boron), 3p^/30-d0/dT (positive due to increased fast
fission), 3p../30•d0/dT and 3p„/30'd0/dT (negative due to
decreased resonance escape probability) and from 3L /30-d0/dT
(in all energy groups negative d ue to the larger leakage at
low water density). The negative contribution to 3p/30-d0/dT
from the change of is much smaller than the corresponding
contribution to 3p/3T .
m

Figs 5.4, 5.8 and 5.12 show the contributions to the total
temperature coefficient, i.e. the sum of the contributions
in the other diagrams.

We have seen that the temperature coefficient is built up

f r o m a number of negative and positive contributions, each
of which has to be correctly predicted. Small errors in
several of the calculated contributions can give a signifi­
cant error in the total calculated coefficient or compensate
each other, so that the total coefficient is in better agree­
ment with experiments than should be expected from the
accuracy of the theory and the nuclear data. Since we do
not have detailed experimental information about the sepa­
rate contributions, we must rely on careful theoretical
analyses of the approximations made in the calculations
and on estimated uncertainties in the nuclear data in order
to determine the accuracy in the theoretical methods.
 Made with Xodo PDF Reader and Editor

AT! Af Ap AL Total
g=1 2 3 i* 2 3 A 2 3 2 3

-5 .0.0 0 0
S 2 SS
' M IN
o i o I
° i to
o '<
o
JD
CQ i=- OQ
c CD £ CD
Q- e Q-e
Q- a a a
- a ioa
to

Fig. 5.1 Contributions to the fuel temp, coeff.

1.35% UO2 lattices
 Made with Xodo PDF Reader and Editor

AT) Ap A^ Total

g=1 2 3 U 1 2 3 1 2 3 /
Hi
1 2 3 A
Af

-5

•o •<_>oo
• •
o O O o

o oo o
CN ID CN ID

03 m m m
£ E Q. Q.
Q . Q . CL CL
a. cun in
o
 Made with Xodo PDF Reader and Editor
5jo d0_
69 dT
pcm/*C

15 «o •u •u •u
oooo
cn cn *-
, CN
i i icn
o oo o
CN to CM UD
10
'c o m
CD CD £ £
cE r£ Q-Q-
CL CL
Ap
CL Q . _ _
2 3

AL Total

1 2 3 A
=0=0
g=1 2 3 I* 2 3 U rat
All Af
-5

-10

-15

Fig. 5.3 Contributions to the water density coeff.

1.35% U02 lattices
 Made with Xodo PDF Reader and Editor

20

15

10

Ap

2 3
AL Total
1 2 3 A
=OUDL
2 3 U
Af

-10 -
o oo
o o Oo
r n i— CD
(Nl CM

o ooo
(N ID CN ID

-15 CD CQCQ CD

l l in
l in
l
O O
 Made with Xodo PDF Reader and Editor

At| Af Ap AL2 [Total

9=1 2 3 2 3 A 1 2 3 1 2 3 A
UM I

o ooo

CN (N
I i I I
O O O Q
-5 CM tO ( N U 5
CN ( n W i" ; '

II II H >1

Fig.5.5 Contributions to the fuel temp, coeff.

1.9% U02 lattices.
 Made with Xodo PDF Reader and Editor

At] Ap AL Total
g=1 2 3 4 1 2 3 1 2 3 A
JlJ
2 3
Af

yyy[
O o
• •
-5 O o cn o o,
cn CN r;
o o'
CN CD

CN CN

"h h ’ m ’
n
:> :> > >

- bi -
 Made with Xodo PDF Reader and Editor
5p d8
50' dT
pcm/ G

Total

Fig. 5.7 Contributions to the water density coeff.

1.9% UO2 lattices
- 64 -
 Made with Xodo PDF Reader and Editor

+10

Ap
1 2 3
A t) AL2 |Total

g=1 2 3 4 1 2 3 I*
=D=cr
W \ 1 2

Af
3
m
-5 O o O O

CT) CN CM
I I ' *
O O O O
CN tO CN lO
-10

II II II II
> > > >
-15 w w
>£>E>e>E

-20

-25

F ig.5.8 C o ntributions to the total temp, coeff.

1 .9 % UO 2 la ttic e s .
 Made with Xodo PDF Reader and Editor

Total

Fig. 5.9 Contributions to the fuel temp, coeff.

1.5% Pu02 la t t ic e s
 Made with Xodo PDF Reader and Editor

+10

Ail Ap | ALZ

g=i 2 3 4 1 2 3 M 2 3
3 4 Total
TF
Af o 0 0 .0
O O O o
I I * 1
c3 §R §
oo oo co oo
-5
^ ^ ro n
ii n n )■
> >> >

Fig. 5.10 Contributions to the moderator temp, coeff.

1.5% U02 lattice s
Made with Xodo PDF Reader and Editor
Made with Xodo PDF Reader and Editor

Total

Fig. 5.12 Contributions to the total temp- coeff.

1.5% Pu02 lattices.
 Made with Xodo PDF Reader and Editor

[Link] INFLUENCE OF APPROXIMATIONS IN THE THEORETICAL

TREATMENT ON CALCULATED TEMPERATURE COEFFICIENTS

Practical reactor physics calculations have to be performed

in a limited number of energy groups and spatial regions.
We will, show that in our calculations the energy groups and
the spatial representation were chosen so that the calculated
temperature coefficient is insensitive to our specific
choice. Other effects that for practical purposes are often
neglected in the calculations will also be discussed here.
The influence of uncertainties in nuclear data will be dealt
with in the next chapter.

6. L___________The crystalline binding in U P ,,

Due to crystalline binding the velocity distribution of

the fuel atoms is not Maxwellian as it was assumed when
calculating the Doppler broadened line shape of the reso­
nance cross section in Eqs (2.21) and (2.22). The uranium
atoms in U 0 2 vibrate w i t h an average kinetic energy larger
than that in a free gas state. L a m b [30] has shown that for
weak binding (or the Short Compound Nucleus Lifetime
approximation, SCNL) the atoms behave like a free gas with
an effective temperature which is higher than the temper­
ature of the medium.

The SCNL approximation is applicable when

r + A >> 2 k e D (6-D

where

9^ is the Debye temperature

/ AkTE
A = / — - is the Doppler width
A

r is the total width of the resonance at half maximum.

Made with Xodo PDF Reader and Editor

The effective temperature, T e ££» is i-n the SCNL approxima­

tion given in terms of the average kinetic energy per mode
of oscillation as [31].

CO

where v is the phonon frequency and g(v) the phonon

f requency distribution.

It is convenient to describe the crystalline effects in

terms of the Debye temperature of the medium. This descrip­
tion is approximate, because in general the crystal does
not have a Debye phonon distribution. However, an effective
Debye temperature, which varies with the temperature of the
medium, can be defined so that it produces the correct
crystalline effects in the medium. Dolling [32] has deter­
mined the effective 0Q versus T for U 0 2 . For temperatures
above 300 K, 0p is fairly constant (e^ ^ 620 K ) .

Shenter [33] has used the phonon distribution measured by

Dolling [32] to calculate T ... He has also given results
eff
for a Debye distribution

2
g(v) = 3 ^ 3 for \> < v
D
v_

g (v ) = 0 for v > v
D

kO
D D

with Q_ = 620 K. T obtained with the Dolling distribu-

D eff
tion and the Debye distribution agree very well for temper­
atures above about 300 K.
 Made with Xodo PDF Reader and Editor

The S C N L a p p r o x i m a t i o n d o e s not hold for the low energy

r e s o n a n c es of U-238. F is 0 . 0 2 5 - 0.1 eV for re s o n a n c e s in
U-238 b e l o w 200eV. The Doppler width, A, is for the 6.68
eV r e s o n a n c e 0.05 eV at 3 00 K and 0.07 eV at 600 K and for
the 190 eV r e s o n a n c e the v a l u e s of A are 0.3 eV and 0.4 eV
at 300 K and 6 00 K, respectively. 2k©D is equal to 0.1 eV
in U O 2 so r e l a t i o n (6.1) is fulf illed for the 190 eV r e ­
sonance and r e s o n a n c e s at h i g h e r ener g i e s but not for the
6.68 e V resonance.

The c h e m i c a l b i n d i n g effects o n r e s o n a n c e s have been studied

by A d k i n s [31]. He used several m o d e l s for the pho n o n f r e ­
q u ency d i s t r i b u t i o n , among o t h e rs the one meas u r e d by
Dolling, and r e p orted results for the 6.68 eV and 190 eV
resonances. In T a b l e 6.1 the r e s o n a n c e integral change be­
tween 300 K and 500 K is shown as obta i n e d by A d k i n s using
a d e t a i l e d c r y s t a l l i n e model, CRYS, the SCNL a p p r o x i m a t i o n
wi t h the D o l l i n g f r e q u e n c y s p e c t r u m and the free gas model.
R e g a r d i n g the CRYS m o del as g i v i n g the correct re s u l t his
r e s u l t s show that the SCNL a p p r o x i m a t i o n r e d u c e s the error
in A R I / A T by a factor 3 compared w i t h the free gas model.
T he small o v e r e s t i m a t i o n of the D o p pler effect in low energy
r e s o n a n c e s that rema i n s can in m o s t a p p l i c a t i o n s be accepted,
b e c a u s e the d o m i n a t i n g c o n t r i b u t i o n to the D o p pler c o e f f i ­
cient in n ormal LWR lattices c o mes from the en e r g y ran g e
above 100 eV for w h i c h energies A > 2k0^.

Tab l e 6 . 1 : C a l c u l a t e d U - 2 3 8 r e s o n a n c e integral change

b e t w e e n 300 K and 500 K [31] u s ing v a rious
crystal m o d e l s

Model 6.68 eV, 0 =17.52 b 190 eV, 0 =40 b

p P
A RI/RI Error ARI/RI Err o r
(%) (%) (%) (%)

CRYS .2534 0 .7791 0

SCNL Dolling .2739 8 .8152 5

Free gas .3092 22 .8981 15

Made with Xodo PDF Reader and Editor

Thus, for calculating the Doppler effect in U-238 it may be

recommended to use an effective Doppler temperature above
the physical temperature of the fuel. This is a very simple
way to account for the crystal binding effects without
having to consider complicated calculations of the reso­
nance cross section line shape. Well established codes
using the ip and x formalism can be used without any
modification. In cell programs which use tabulations of the
effective resonance integral as function of the temperature
Te^ should then be used instead of T when interpolating
in the tables. This procedure has been built into the
A.E-BUXY code w ith T g ££ given by Eq (6.2) w i t h the Debye
frequency distribution and eD = 620 K

T (6.3)
eff !»»

AT = T . -T as function of T is plotted in Fig 6.1.

eff

The influence of the effective temperature on calculated

fuel temperature coefficients in three of the lattices
which were analysed in chapter 4 is shown in Table 6.2.

Table 6 . 2 : Comparison between fuel temperature coefficients

calculated using T ,, = T and T defined
efr eff
by Eq (6.3)

Case Ok to/3T, (pcm/°C)

T e m p .interval(°C) 20-90 160-210

1.35 % U02
-4.7 -4.1
V /V =1.4 T e£f-T
in f T ef £ -E q (6.3) -3.8 -3.5
175 ppm B

1.9 % U02
T =T -4.6 -4.1
V /V =1.7
m f T e f f =Eq ( 6.3) -3.8 -3.5
200 ppm B

Pu02
-4.1 -3.7
1e f f =T
V /V =2.5
m f -3.5 -3.2
T e f f = E q '6,3)
250 ppm B
 Made with Xodo PDF Reader and Editor

The cylindrrealization of pin cells

The lattice cell is in AE-BUXY replaced by a circular

Wigner-Seitz cell with white boundary conditions. In order
to investigate the validity of this approximation we
carried out calculations on the actual square cell in two
lattices and compared calculated values of and its
temperature dependence with results obtained in the corre­
sponding circular cell.

Similar investigations have been done previously by Honeck

[34, 35], Fukai [36, 37], Sauer [38], Carlvik [39], Dudley
[40], Newmarch [41] and Weiss [42] among others. These
investigations are, however, limited to the study of the
flux distribution in one-group calculations with a flat
source in the moderator. We found it therefore worthwhile
to carry out multi-group calculations where the spatial
variation of the source in each group is taken into account.

The two square lattices contain 1.9 % IK^ rods and the
moderator to fuel ratios are 1 arid 2. To simplify the cal­
culations the fuel has no canning. Calculations were done
for 20 °C and 245 °C.

The multi-group transport codes BOCOP [43] and FLUCAL [44]

were used for the calculations. BOCOP is a two-dimensional
collision probability code in xy-geometry and the circular
fuel region was approximated by a polygon. The polygon was
chosen so that the volume of the rod is preserved. The outer
square boundary of the cell is correctly represented in
BOCOP with reflecting boundary conditions. FLUCAL solves
the transport problem in annular geometry by use of the
DIT-method. The cell boundary was cylindricalized preserving
the cell volume and white boundary conditions were applied.
Made with Xodo PDF Reader and Editor

All calculations were made with 8 energy groups with the

group boundaries 10 MeV - 9118 eV - 4 eV - 0.625 eV -
-0.14 eV - 0.058 eV - 0.030 eV - 0. Group cross sections
were generated by AE-BUXY.

Table 6.3 shows the comparison of and temperature

coefficients, Ak OO/AT, calculated in the square and the
circular cell.

Table
■"
6.3: Comparison
■ ■— ■ ■■■■ - ■ ■ *
of kOO and Ak oo /AT calculated
in square and circular cell

V /V, Temp. k OO Ak^/AT (pcm/°C)

m f
(°C). Square Circ. Square Circ.

20 1.23366 1.23427
1.0 -24.66 -24.78
245 1.17817 1.17851

20 1.27605 1.27654
2.0 - 8.02 - 8.09
245 1.25800 1.25834

The difference between the results obtained in the two

geometries is small, k^ is overestimated by about 50 pcm
when the square cell is cylindricalized. Ak^/AT is about
0.1 pcm/°C more negative in the circular geometry. This
small discrepancy may, however, be due to numerical uncer­
tainties in the calculations which are about 0.1 pcm/°C.

We conclude that the error introduced by the cylindricaliza-

tion of the pin cells is negligible. This conclusion is in
agreement with the results of [34-42], where the authors
in most cases found that the flux distribution obtained
from a one-group calculation in a Wigner-Seitz cell with
white boundary conditions is a good approximation for the
solution in the exact geometry.
 Made with Xodo PDF Reader and Editor

Comparison between pin ceil calculations using

isotropic and anisotropic scattering___________

In cell calculations transport corrected cross sections

are usually used to account for anisotropy in scattering.
The transport cross section used in AE-BUXY is derived by
expanding the transport equation in Legendre polynomials.
P^-row sum and weighted P^-column sum corrections [21] are
used in the thermal and epithermal regions, respectively.

In order to check the adequacy of the use of a transport

corrected total cross section, pin cell calculations were
performed using S -theory. These calculations were made
O
with DTF-4 in 6 energy groups with the boundaries 10 MeV -
- 0.5 M e V - 9118 eV - 0.625 eV - 0.140 eV - 0.058 e V - 0.
Cross sections were generated by AE-BUXY and DTF-4 was run
on a 1.35 % U0£ pin cell at 20 °C and 40 °C with either P^-
scattering and transport corrected diagonaT’elenents or P^-
scattering. The results are shown in Table 6.4. It is seen
that the use of transport corrected isotropic scattering is
a very good approximation for the P^-scattering. The very
small discrepancy between the P^- and P^-results is not
significant and we conclude that transport corrected cross
sections give the same k^ as calculations using explicit
P^-scattering.

Table 6 . 4 : Comparison of DTF-4 calculations using P -

and P -scattering. 1.35 % U0„, V / V c = 1 . 4
I I m f

k 00 Ak 00/AT
o

20 °C (pcm/°C)
o
o

Pg-scattering 1.16024 1.15834 -9.50

P^-scattering 1.16028 1.15837 -9.55
Made with Xodo PDF Reader and Editor

The prediction of leakage in small cores is a crucial part

of the reactor physics calculations. The calculated leakage
is sensitive to nuclear data in the fast energy region
(fission neutron spectrum, inelastic scattering cross
sections for U-238, hydrogen and oxygen scattering data),
and a proper treatment of the anisotropic scattering is re­
quired. One must therefore expect that in most cases there
may be an uncertainty of several per cent in the calculated
leakage. The error in the predicted leakage due to erroneous
fast data or an inadequate treatment of the anisotropic
scattering is, however, insensitive to the temperature and
the influence on calculated temperature coefficients is less
significant than the influence on the k ^ ^ - v a l u e .

We have seen in chapter 4 that the fundamental mode calcula-

2
tion using the m easured B and the two-dimensional diffusion
m
theory calculation give approximately the same reactivity,
i.e. the predicted leakage is the same. DIXY-calculations
have also been compared with S Q-calculations using DTF-4 in
O
one-dimensional cylindrical geometry [45]. The agreement in
calculated temperature coefficients is good. The axial
2
leakage obtained using the measured agrees well with that
calculated by DTF-4 (Table 4.6). Thus, we do not expect that
the observed discrepancies between experimental and theoret­
ical temperature coefficients can be explained by errors
in the leakage calculation.

This conclusion is supported by the observation that the

error in predicted temperature coefficients is the same at
low and high temperature although the influence of the
leakage on the coefficient is much larger at elevated
temperatures. Further, we note that the leakage in the
cores of chapter 4 varies from 10 % to 25 but no trend
is seen in the discrepancy, Ak ^ / A T , as function of the
1cakage.
 Made with Xodo PDF Reader and Editor

6.5^1_____ Macro grougs_and Gauss H 2 iD£®_iD_^E~BUXY

AE-BUXY calculations were made on the 1.35 % UC> 2 lattice

with 175 ppm boron in the water using 17 macro groups and
8 Gauss points (3 in UO^, 2 in canning, 3 in H^O) or 25
groups and 16 Gauss points (6 in UC^, 4 in canning, 6 in
1^0). Condensation vectors for the group structures are
(cf Table 3.1).

17 groups 2, 4, 6, 14, 21, 25, 26, 27, 35, 38,

45, 48, 51, 55, 59, 63, 69.

25 groups 2, 4, 6, 14, 21, 23, 25, 26, 27, 32,

35, 38, 41, 45, 48, 51, 54, 55, 56,
57, 59, 61, 63, 66, 69.

The two group-structures differ mainly in the thermal region.

Results from calculations at two temperatures are given in
Table 6.5. The discrepancies between the 17 and 25 group
results are negligible.

Table 6 . 5 : Comparison between AE-BUXY calculations with

17 macro groups, 8 Gauss points and 25 macro
groups and 16 Gauss points. 1.35 % 175 ppm B

2
Temp Number Number ko k c. B2 M
eft m
of of
macro­ Gauss
(°C) groups points (m'2) ( ^)
(cm

20 17 8 1.12172 0.99640 29.91 40.7

25 16 1.12171 0.99630 29.88 40.7 '

210 17 8 1.10448 0.99060 20.77 50.3

25 16 1.10459 0.99060 20.77 50.3
Made with Xodo PDF Reader and Editor

To investigate h ow the leakage depends on the number of

fast groups, calculations were made on the 1.35 % UO^
lattice without boron in the water using 17 and 23 groups
in the fundamental mode calculation. The 17 group structure
is the same as given above. 6 groups were added in the fast
region to obtain the 23 group structure which is

23 groups 1, 2, 3, 4, 5, 6, 8, 10, 12, 14, 21,

25, 26, 27, 35, 38, 45, 48, 51, 55,
59, 63, 69.

The 17 and 23 group structures are identical below 9118 eV

(gr 14). Results are given in the tables below.

Table 6.6: AE-BUXY calculations with 17 and 23 macro groups.

1.35 % U 0 2 , 0.8 ppm B

r
2 ;
Temp Mumber k 00 k B2 M
ef f m
of
ma cro­
(°C) groups (m “) (cm )

90 17 1.15566 0.99585 36.01 43.2

23 1.15567 0.99415 35.61 43.7

21.0 1.7 L. 1.3872 0. 99147 27.28 50.9

23 1.13874 0.98983 26.97 51.4

Table 6.7: Temperature coefficients derived from Table 6.6.

Number of Alc^/AT Ak ff/AT

macro groups (pcm/°C.) (cpm/°C)

17 -14. 1. -3.65
23 -14.1 -3.60
 Made with Xodo PDF Reader and Editor

It is seen from Table 6.6 that the leakage is underestimated

by about 1 % in the 17 group structure compared to the 23
group structure. This corresponds to about 0.2 « in k
eff
The temperature coefficients were changed very little when
the number of groups was changed (Table 6.7).

The results in Tables 6.5-6.7 are in agreement with previous

experience and justify the use of 17 macro groups and 8 Gauss
points in AE-BUXY.

6.5.2 Energy groups and mesh points in the diffusion

theory calculation

The number of energy groups and mesh points was varied in

diffusion theory calculations on the 1.35 % U 0 2 lattices
with surrounding reflectors. Results are given below.

Two group-structures were compared with the condensation

vectors

4 groups 6, 14, 45, 69.

8 groups 2, 6, 14, 45, 55, 59, 63, 69.

Table 6.8: k rr from DIXY. 1.35 % U0„ lattice with 46x46

---------- err 2
fuel pins, 175 ppm B

Number of Number of k eff Ake ff/AT

groups m e s h points
20°C 210°C (pcm/°C)

4 28x28 0.99691 0.99301 -2.05

4 39x39 0.99730 0.99349 -2.01
8 39x39 0.99730 0.99350 -2.01
Made with Xodo PDF Reader and Editor

It is seen from Table 6.8 that k __ and Ak AT are

eff eff
insensitive to the number of energy groups and mesh points
in the diffusion theory calculation. The exact agreement
between the 4 and 8 group calculations is accidental. Similar
comparisons made for other purposes have given discrepancies
of about 100 pcm between 4 and 8 group calculations.

6.6_________ The influence of thermal expansion on reactivity

For practical reasons the thermal expansion of fuel, canning

and construction material is often neglected in calculations,
because this effect in most cases gives only a small correc­
tion.

The influence of thermal expansion on the reactivity was

estimated separately for all the KRITZ lattices analysed
in chapter 4, and the corrections have been applied in the
results given in Tables 4.2-4.7.

The following coefficients of thermal expansion were used

i i -io'Vc

7•10_6/°C
SS 18‘
10~6/°C

The variation of the lattice pitch with temperature is due

to expansion of stainless steel in which the spacers are
fastened.

In Table 6.9 the correction for thermal expansion is split

up into contributions from the change of fuel radius, can
radius, cell radius, fuel density and can density. It is
seen that the effects on dk /dT
(X>
are very small except for
*

the contribution due to l\r ,- . The dominating effects on

cell
dk rp/dT are due to changes in the leakage in the fast
 Made with Xodo PDF Reader and Editor

energy region. The individual contributions compensate

each other to some extent. In the 1.35 % UO^ lattice with
0 ppm boron in the water the total correction to the
temperature coefficient is only -0.05 pcm/°C.

Table 6 . 9 : Individual contributions to the temperature

coefficient due to thermal expansion.
1.35 % ITC>2 lattice with 0 pp m boron in the water

dk OO /dT dk /dT d B 2 /dT d M 2 /dT

eff m

(pcm/°C) (pcm/°C) (m"2 /°C) (cm2 /°C)

-4
-.04 + .18 5-10_A - 5-10
“uo2 +

-.03 -.27 - 7 + 5
SrZr-2
Ar ., + .20 + .38 + 10 - 5
cell
-.04 -.34 - 9 + 10
AeU 0 2

+ .01 0 0 0
A0Zr-2

Total + .10 -.05 - 1 + 5

The expansion of the spacers increases the lattice pitch

at high temperatures. This affects not only the cell cal­
culations 'but also the core calculation since the cor- size
is changed. For the core with 46x46 pins we get
A B 2 /AT = -6*1 0 ~ 4 m _ 2 /°C . A B 2 /AT is -0.05 m ~ 2 /°C so the
g m
correction is in this lattice about 1 % of the total temper­
ature coefficient wh i c h corresponds to 0.2-0.4 pcm/°C.

Table 6.10 shows corrections for thermal expansion cal­

culated using AE-BUXY.
Made with Xodo PDF Reader and Editor

Fuel Boron dk OO/dT dk _./dT dB2/dT dM2/dT

m f eff m
conc.
(ppm) (pcm/°C) (pcm/°C) (m'2/°C) (cm2/°C)
-4
0 + .10 -.05 - 1-10"4 + 5-10
1.35 % U02 1.4
175 + .02 -.14 - 3 + 10

0 + 1.05 + .65 +13 + 9

z9.
1♦
300 +1.01 + .65 +13 + 9
1.9 % U0o
z
1.7 300 + .13 -.09 - 2 + 7

1.8 300 + .82 + .50 +10 + 6

Pu0o 2.5 300 + .10 -.14 - 3 + 5
3.3 j - .10 -.29 - 8 + 5

6.7_________The plutonium fission neutron spectrum

The fission neutron spectrum used in the analysis is a

Maxwellian with ~ 1*33 MeV. The average energy of
fission neutrons from fissions in plutonium is higher than
that given by this spectrum. The use of a plutonium fis-
2
SLon ne u t r o n spectrum increases the p redicted and M ,
but has a negligible influence on the temperature coeffi­
cient. The table below shows results from AE-BUXY calcula­
tions with two different fission neutron spectra represen­
tative for fissions in uranium and plutonium. The Maxwellian
temperatures are 1.33 MeV and 1.385 MeV, respectively. The
change in k is due to an increased number of fast
. 2
fissions m U-238. M is 1.4 % larger with the plutonium
spectrum.
 Made with Xodo PDF Reader and Editor

and p l u t o n i u m fis s i o n spectra.

PuO- lattice w i t h V /V, = 2.5
2 m f

T ’
mp . Fission neutron k oo k M2
eff
(uc) spectrum (cm )

U 1.22588 1.00000 36.60

20 Pu 1.22685 .99781 37.11
diff +97 p cm -219 pcm + .51

U 1.23829 1.00000 46.31

210 Pu 1.23938 .99775 46.96
diff +109 pcm -225 pcm + .65
 Made with Xodo PDF Reader and Editor

Fig. 6.1 Te ff as function of T

 Made with Xodo PDF Reader and Editor

7. THE INFLUENCE OF N U C L E A R DATA ON THE CALCUL A T E D

T E MP E RATI IRE COE F F 1CIE NT

7.L C o m p a r i s o n of: Lemper;i lure c o e f ficients calculated

__ ___________by use of ENDP/B and UKNI)1^..lata______________________

The s t a n d a r d library in A E - BUXY is b a sed on UKNDL but a data

l ibrary in the BUXY format has also b e e n gen e r a t e d from
E N D F /B-III. E p i t h e r m a l cross sections and e f f ective reso­
n a n c e integrals w e r e p ro c e s s e d by Che S P ENG and DORIX
codes [46]. Thermal group cross sections were generated
by use of F L A NGE-2 [47] whi c h was m o d i f i e d to calculate
group s c a t t e r i n g m a t r i c e s instead of p o i n t - t o - p o i n t matrices.

Thr e e of the lattices studied in c h a p t e r 5 were recalculated

u s i n g the data library g e n erated from ENDF/B-III. Partial
and total temp e r a t u r e coefficients ca l c u l a t e d with the two
libraries are c o m p a r e d in T a ble 7.1.

The a g r e e m e n t b e t w e e n the t e mperature c o e f f icients obta i n e d

w i t h the two d i f f e r e n t libraries is remarkable (the levels
2
of c a l c u l a t e d k and M differ, however, more than the
c o e f f i c i e n t s in Table 7.1). The two A E - B U X Y libraries are
e n t i r e l y independent. Both the bas i c n u c lear data and the
p r o c e s s i n g codes used for g e n e r a t i n g group cross sections
differ.

S c a t t e r i n g m a t r i c e s for water are based on the Nelkin

m o d e l in the "UK.\'DL"-library and on the H a ywood model in the
" E N D F / B " library (cf § 7.3).

The good a g r e e m e n t in the m o d e r a t o r temperature coefficients

shows that the thermal, s c a t t e r i n g m a t r i c e s and thermal
r e a c t i o n cross sections are in agreement. The wat e r density
c o e f f i c i e n t s o b t a i n e d with the two data libraries are
p r a c t i c a l l y the same.
 Made with Xodo PDF Reader and Editor

f ......
Case Data l i b 3X/3Tf 3X/3T 3X/36‘ dQ/dT dX/dT
m
Temp 20-90 160-210 20-90 160-210 20-90 160-210 20-90 160-210
in te r v a l
(°C)

1.35% UO9 X=km UKNDL -4.7 -4.1 -3.3 -3.3 +0.1 -3.7 -7.9 -11.1
Vm/Vf-1.4 (n r m ^VDF/R -4.8 -4.2 -3.3 -3.3 0 -4. 1 -8.3 -11.6
175 ppm B X=p UKNDL -4.1 -_J. 1 “ *4. 3 -4.U “ D. 9 -20.2 -15.3 -27.9
(pcm/°C) I'NDF/R -4.2 -3.8 -4 . 3 -4. 3 -6.9 -20.2 -15.4 -28.3
- . U113 -.0086 -.0113 * .0092 -.0164 -.0406 -.o3yo -.0584
(m-2/°C) ENDF/B -.0121 -.0094 -.0120 -.0100 -.0164 -.0406 -.0405 -.0600
X=M2 UKNDL -.1)004 -.0004 .0046 .0046 .0251 .0730 .0293 .0772
(cm2/°C) ENDF/3 -.0004 -.0004 .0046 .0048 .0239 .0692 .0281 .0736

1-9% U09 X=k01 UKNDL -4.6 -4.1 -2.6 -2.9 +0.7 -2.4 -6.5 -9.4
Vm/Vf =l77 (pcm/°C) ENDF/B -4.7 -4.2 -2.7 -3.0 +0.8 -2.2 -6 . 6 -9.3
200 ppm B X=p UKNDL -3.8 -3.4 -4.1 -4.1 -10.9 -30.3 -18.9 -37.8
(pcm/°C) ENDF/B -3.8 -3.4 -4.2 -4.3 -11.0 -30.5 -19.1 -38.2
UKNDL -.0110 -.0086 -.0117 -.0100 -.0291 -.0690 -.0518 -.0876
X=BI
(m**2/oc) ENDF/B -.0116 -.0090 -.0129 -.0112 -.0300 -.0706 -.0544 -.0908
X-M2 UKNDL -.0004 -.0006 +.0042 +.0042 +.0261 +.0768 +.0300 +.0804
(cm~/°C) ENDF/B -.0003 -.0004 +.0043 +.0044 +.0250 +.0734 +.0290 +.0774

Pu09 X=kOT UKNDL -4.1 -3.7 +7.6 +8.7 +2.7 +0.1 +6.2 +5.1
ENDF/B -4.1 -3.8 +7.5 +8.4 +2.7 +0.2 +6.1

00
V V 2*5 . (pcm/°C)

+
t
250 ppm B |X*o UKNDL -3.3 -3.0 +5.1 +6.7 -11.4 -36.5 -9.6 -32.8
(pcm/°C) ENDF/B -3.3 -3.0 +5 .0 +6.4 -11.5 -36.5 -9.8 -33.1
■
X=B& UKNDL -.0107 -.0082 +.0164 +.0186 -.0351 -.0952 -.0294 -.0848
i (m~2/oc)
ENDF/B -.0110 -.0086 +.0160 +.0176 -.0354 -.0956 -.0304 -.0866
X=M2 UKNDL -.0002 -.0002 +.0025 +.0016 +.0262 +.0784 +.0285 +.0798
1
(cm2/°C) ENDF/B -.0003 -.0004 +.0027 +.0018 +.0257 +.0766 +.0281 +.0780
i ! 1
 Made with Xodo PDF Reader and Editor

7 •2 _ The e f f e c t ive-' resonance i»tegral of U-238

Shielded resonance integrals for U-238 processed from UKNDL

of ENDF/B-IT.1 give an overestimation of the resonance
absorption in cell calculations [48-50]. Tt is there­
fore necessary to reduce the resonance integral and corre­
late it to integral experiments.

In the data library based on UKNDL which was used in our

analysis of experiments, the U-238 absorption cross section
was reduced uniformly by 0 . 2 0 barn in the energy interval
4 eV - 9118 eV. This correction was originally suggested
by Askew [48]. The effective resonance cross section for
energy group g is given by Eq (3.7)

RI (a ,T)
■ W w ” ■ — L J ;; »t )
x - __ 5__ P.?.5___
p»g

where RI (o ,T) is the tabulated effective resonance

g P»g
integral as function of the background cross section, ,
and the temperature, T .
is the lethargy width of
t
g
group g . D i fferentiating Eq (7.1) one obtains

ARI„(o„ „,T) RI„( o „ -T) 2

1 _____________ g .P.1« . ]
£ ■.P »g_____ [ Ao (7.2)
a ,g t o
8 8 P>g

which with Ao = -0.20 b for all e is the correction used

a »g
in the tables of the library.

In the last column of Table 7.3 resonance integrals cal­

culated by AE-BUXY using the Wigner approximation for the
fuel self collision probability

Et 1
P ff = ——- ; x = — and E = — (7.3)
ff a+x Ze2rv;
e
 Made with Xodo PDF Reader and Editor

and the above c orrec t i o n are compared w i t h resonance inte­

grals m e a s u r e d by H e l l s t r a n d [51] in isolated fuel pins con­
taining UO^ or U-metal. The c o m p a r i s o n is done for UO^ rods
w i t h two d i f f e r ent radii. The smaller radius is typical for
a B W R rod and the larger one gives b a c k g r o u n d cross sections
c o r r e s p o n d i n g to a Dancoff factor of about 0.5, w h i c h is
r e p r e s e n t a t i v e for a B W R assembly w i t h average void contents
in the coolant. Resonance integrals for U-metal rods are
also c o m p a r e d at two radii. The expe r i m e n t a l values are
g i v e n in T a b l e 7.2. The tables also show the temperature
de p e n d e n c e of the resonance integrals. The calculated reso­
nan c e inte g r a l s are here compared w i t h e x p e r imental results
fr o m r e f e r e n c e s [51, but a c o m p a r i s o n w i t h others e x p e r i ­
ments w o uld give essentially the same results. See e.g. the
r e v i e w by H e l l s t r a n d in [53]. The e x p e r imental un c e r t a i n t i e s
are about 4 % in the room temperature integrals and about
10 % in the D o p pler coefficients.

It is seen that AE-BUXY, '■

Lth the resonance integral c o r r e c ­
tion w h i c h w as used in the a n a l y s i s in chapter 4, o v e r e s t i ­
ma t e s the D o p p l er effect by about 10-15 %, co r r e s p o n d i n g to
0 . 4 - 0 . 8 pcm/° C in the fuel t e m perature coefficient. In order
to see h o w the c o r r e c t i o n in itself affects the t e m perature
co e fficient, three d i fferent c o r r e c t i o n s were tested, viz.

Correction la : .'a = - 0 .20 barn

a
11 _ __
2a : ;'o = -0. 20[ i *-0.007 (/T-/300) ] barn

" 3a : .MU = -0.06 RI

A rational a p p roximation Cor tin? fuel self c o llision pr o ­

b a b i l i t y suggested by [Link] ivik [54]

was u s e d instead of the Wigner a p p r o x i m a t i o n because the

C a r l v i k a p p r o x i m a t i o n is more accurate.
 Made with Xodo PDF Reader and Editor

Using uncorrected data the calculated resonance integral

is overestimated by about 6 % in UC^ rods and 5 % in
U-metal rods (Table 7.3). The temperature dependence is
predicted within about 5 %. The correction la is somewhat
too large giving a slightly underestimated resonance integral
in the UO^ rods. For the U-metal rods, which have a smaller
resonance integral, this correction is about twice the
required one. The Doppler coefficient is increased by 9 %
due to the correction. In order to preserve the temperature
dependence of the uncorrected resonance integral correction 2a
was tested. This correction is identical with la at 300 K.
but does not change the /F-dependence of the resonance inte­
gral. As seen in the table ARI/AT remains the same as
with the uncorrected data. Correction 3a reduces the reso­
nance integral by & % at all temperatures and o^-values. This
correction also leaves the Doppier coefficient unchanged.

Table 7.4 shows the corresponding comparison for the

library based on ENDF/B-III. The results are similar to
those for the U K N D L - l i b r a r y . The uncorrected resonance inte­
gral is about 10 % too large, compared to 6 % for the UKNDL
data, and the following corrections were tested

Correction lb : Ao -0.27 barn

a
2b : Aa = -0.27[l+0.007(/T-/300)]barn
a
" 3b : ARI = -0.10 RI

The u n i f o r m reduction of a is larger than in the UKNDL

library and the discrepancy in ARI/AT is consequently
larger. Corrections 2b and 3b reproduce the measured Doppler
coefficient better.
Made with Xodo PDF Reader and Editor

The corrections la and lb have the advantage that the un­

shielded resonance integral is reduced by 1.5 b and 2.1 b,
respectively, which is within the experimental uncertainties.
They do, however, change the temperature dependence of the
resonance integral. Corrections 3a and 3b do not affect the
Doppler coefficient but give a value of the unshielded reso­
nance integral which lies outside the experimental uncer­
tainty. Corrections 2a and 2b combine the advantages of
corrections 1 and 3. A drawback is, however, that although
ARI/AT over the whole resonance energy region becomes
correct, ARI /AT in individual energy groups will be
§
erroneous. For example, in a group without any resonance
where ARI /AT « 0 one obtains ARI /AT < 0 .
g g

The measured temperature dependence of the resonance inte­

gral was fitted to the expression [52].

RI(T)-<5 = [RI(T )-5 ] -11+3 (v^T'-v/T—)] (7.5)

o o

where 6 is the 1/v-part of the resonance integral and

the coefficient 6 varies slightly with the surface tp mass
ratio, S/M, but is assumed to be independent of T . The
measurements cover the temperature range up to 1000 K.

The theoretical values of the Doppler coefficients in

Tables 7.3 and 7.4 were determined using the real temperature,
T, of the [Link] the crystalline binding effects are taken
into account using an effective Doppler temperature, Te j£>
defined by Eq (6.3) one finds that the calculated Doppler
coefficient in the UO^ rods is reduced by about 9 % over
the temperature interval 300 - 1000 K. The influence of
crystalline effects in the U-metal rods can be expected to
be small.
 Made with Xodo PDF Reader and Editor

Table 7 . 2 : Measured U-238 resonance integrals [51] and

Doppler coefficients [52] for isolated fuel rods

Fuel Radius RI 6 '*10“

exp
(cm.) 300 °K

u°? 0.52 21.6 0.77

uo2 1.04 16.8 0.67
u 0.50 16.1 0.73
u 1.00 12.6 0.62

cf Eq (7.5)

Table 7 . 3 : Comparison between calculated U-238 resonance integrals

and Hellstrand's measurements. UKNDL data

Fuel Radius Carlvik approximation Wigner

(cm) No corr Corr la Corr 2a Corr 3a Corr la

U°2 0.52 +5.7 - 0.9 -0.9 0.0 - 1.2

U02 1.04 +6.4 - 1.7 -1.7 +1.1 - 1.8
RI
U-metal 0.50 +4.9 - 3.7 -3.7 -0.4 - 4.0
RI - 1 <% >
exp
U-metal 1.00 +4.4 - 5.9 -5.9 -0.5 - 6.0

uo2 0.52 +6 +15 +6 +6 +14

RI (ARI/AT)
exp th 1
uo2 1.04 -4 + 5 -4 -4 +11
RI., (ARI/AT)
th exp
U-metal 0.50 +1 +11 0 ~i
L +15
(%) U-metal 1.00 -2 + 9 -5 -3 +11
 Made with Xodo PDF Reader and Editor

Table 7.4: Comparison between calculated U-238 resonance integrals

and Hellstrand's measurements. ENDF/B-III data

Fuel Radius Carlvik approximation Wigner ;

(cm) No corr Corr la Corr 2a Corr 3a Corr la

U°2 0.52 + 9.2 + 0.4 +0.4 -0.5 0.0

RI UCL 1.04 +10.1 - 0.7 -0.7 +0.8 0.0

2
RI - 1 «> U-metal 0.50 + 10.8 - 0.6 -0.6 +1.4 - 0.2
exp
U-metal 1.00 +12.1 - 1.3 -1.3 +3.1 - 1.0

U0„ 0.52 + 6 +6 +4 + 16
RI (ARI/AT) 2
exp th i
U0„ 1.04 - 1 +12 0 -3 +15
RI (ARI/AT) 2
tn exp
U-metal 0.50 + 5 + 18 +3 +3 + 21
(%) U-metal 1.00 - 2 +14 -4 + 11
-4

The analysis in chapter 4 was performed using the Wigner

a pproximation and correction la to the U-238 resonance inte­
gral. Table 7.3 shows that using this option the Doppler
coefficient is overpredicted by about 10-15 %. This over-
estimation is reduced to 0-5 % when the crystalline binding
effects are accounted for. It should, however, be remembered
that the experimental uncertainty is of the magnitude 10 %
and one can therefore not draw too detailed conclusions
about the ability of the theory from these comparisons.

W h e n the crystalline binding effects are neglected the shape

of the calculated RI versus temperature can be fitted to
expression (7.5) with a 0-value which is different from the
experimental one. Using the effective Doppler temperature,
T^j;, the calculated Doppler coefficient is reduced more
at low temperature than at high, thus giving a slightly
different shape of RI(T). The precision in Doppler coefficient
measurements has usually not been hi°h enough to detect any
temperature dependence in g . There are, however, some
experimental results indicating a deviation from the in­
dependence, e.g. [55].
 Made with Xodo PDF Reader and Editor

7._3I_1_____ Scatter ing_i.n_water

The Nelkin scattering model was used for water in the

analysis in chapter 4. Below, temperature coefficients
calculated by use of the Nelkin, the effective width and
the Haywood model are compared. Scattering matrices derived
from the effective wi d t h model were tabulated at 293 K w i t h q =
4.3, 450 K with q = 2.8 and at 600 K with q = 2.1 (cf eq 2.15).
Scattering cross sections at intermediate temperatures wer'i
obtained by linear interpolation. The Haywood scattering
cross sections are those obtained from ENDF/B-III by use
of FLANGE-2.

Moderator temperature coefficients for three lattices using

the three different scattering models are listed in Table 7.5.
The agreement in calculated coefficients is very good con­
sidering that three entirely different models have been
used.

Table 7.6 shows calculated reaction rate ratios in the fuel

region using the different scattering models. The reaction
rates obtained with the Nelkin and the effective width model
are equal, whereas the Haywood model gives a somewhat harder
spectrum at both low and high temperature.

7 ^ 3 ^ 2 __________ § £ 2 t t e r i n g _ i n _ U 0 2

Calculations using the IK^ scattering model in ENDF/B-III

have been compared w i t h calculations using the free gas model
for both u r a nium and oxygen in • Although the scattering
cross sections differ considerably in the two cases, the
influence on calculated reactivities and coefficients is
negligible. This is due to the small importance of the
thermal scattering in the fuel in light water reactors. (It
was shown in chapter 5 that the thermal contribution to the
fuel temperature coefficient is only about -0.3 pcm/°C).
 Made with Xodo PDF Reader and Editor

Table 7 , 5 : C a lc u la t e d m oderator tem perature c o e f f i c i e n t s u s in g d i f f e r e n t

s c a t t e r i n g models f o r water

Case S c a t t e r in g model dk /9T 3p/3T 3B2/3T 3M2/3T

00 m m m m m
Temp i n t e r v a l (°C) 20-90 160-210 20-90 160-210' 20-90 160-210 20-90 160-210

1.35 % U09 N e lk in -3.3 -3.3 -4.2 -3.9 -.0113 -.0092 +.0046 +.0046
V /V *1.4 E ff.w id th -3.3 -3.2 -4.5 -4.1 -.0124 -.0096 +.0060 +.0058
m f
175 ppm B Haywood -3.2 -3.2 -3.8 -.0114 -.0094 +.0046 +.0044

r-
H |
i |
•
1
1.9 % U09 N e lk in -2.6 -2.9 -4.0 -3.9 -.0117 -.0100 +.0042 +.0042
V / V _ = l.7 E ff.w id th -2.6 -2.8 -4.6 -4.3 -.0136 -.0110 +.0057 +.0054
m r
1 U1

200 ppm B Haywood -2.8 -3.9 -3.9 -.0119 -.0104 +.0041 +.0042
i
•

Pu09 N e lk in +7.6 +8.7 +5.0 +6.3 +.0163 +.0184 +.0025 +.0016

V /V =2.5 E ff.w id th +8.9 +9.2 +5.5 +6.4 +.0179 +.0182 +.0037 +.0024
m f
250 ppm B Haywood +7.4 +8.6 +4.7 +6.2 +.0158 +.0186 +.0026 +.0016
 Made with Xodo PDF Reader and Editor

Table 7 . 6 : C a lc u la t e d r e a c t i o n r a te s f o r U-235, Pu-239, Pu-241 and Dy-164

r e l a t i v e to the r e a c t i o n r a t e i n a 1/v-absorber

Case S c a t t e r in g modal (2 3 5 /l/v )^ (2 4 1 /1/v ) . ( D y / i / v ) a bs

(2 3 9 /1 /v )f is s f 1SS

Temp# (°C) 20 210 20 210 20 210 20 210

1.35 % U02 . N e lk in .993 .968 1.288 1.498 1.109 1.216 .926 .902
V /V =1.4 E ff.w id th .994 .968 1.286 1.500 1.110 1.218 .926 .902
m f
175 ppm B Haywood .990 .966 1.307 1.514. 1.116 1.220 .924 ,901_

1.9 % JJO N e lk in .993 .969 1.305 1.515 1.116 1.222 .920 .897
V /V =1.7 ^ ff.w id th .994 .969 1.304 1.517 ' 1.117 1.224 .920 .897
tn f
200 ppm B Haywood .990 .967 1.326 1.531 1.123 1.226 .918 .895

Pu02 N e lk in .997 .973 1.185 1.357 1.081 1.174 .926 .894

V / V =2.5 E ff.w id th .997 .974 1.184 1.357 1.082 1.175 .927 .895
m i
250 ppm B Haywood .994 .971 1.197 1.365 1.085 1.175 .924 .893
Made with Xodo PDF Reader and Editor

Thermal absorption and fission cross sections

7.4.1 Group cross sections for a 1/v-absorber in

____________ Maxwel_Han_sgectra_of_various_temperatures

The AE-BUXY data library contains group cross sections which

have been generated by energy condensation of the cross
sections using typical w e i ghting spectra. The energy groups
have been chosen so that group constants may be considered
to be insensitive to the weighting spectra used to generate
them. In the thermal region Maxwellian spectra have been
used for weighting. Scattering matrices are tabulated for
a representative range of temperatures, whereas absorption
and fission cross sections for most nuclides have been
assumed to be temperature independent.

In order to verify this assumption, group cross sections for

a 1/v-absorber were calculated for E <_ 0.625 eV using M a x ­
w ell i a n spectra with various temperatures for weighting.
A n analytical expression for the group cross sections of
a 1/v-absorber is easily derived

(7.6)
M(E,T)dE
E£g

where

The evaluation of the integrals in (7.6) gives

a g (T) = (7.7)

g+ and g- are the upper and the lower energy boundaries of

group g .
 Made with Xodo PDF Reader and Editor

Calculated a (T) in the AE-BUXY group structure are

listed in Table 7.7 for the temperatures 300, 600, 900 and
1200 K and for E 0.625 eV. CT2200 ecIua^ to uHity- It
is seen that the group cross sections are insensitive to the
temperature of the Maxwellian spectrum used for weighting.
This conclusion may be expected to hold for all smooth cross
sections. Thus, the error in predicted reactivities and
reactivity coefficients caused by the use of a limited
number of thermal energy groups is negligible as far as
reaction cross sections are concerned (if all materials
are containing only nuclides with smoothly varying cross
sections).

Z •--- _Thermal_data_for_U-235_and_U-238

Figures 7.1-7.4 show the thermal U-235 data, a , a ,

sl r
a = o /o, - 1 and n = vo,/o in the libraries generated
a f r a
from UKNDL and ENDF/B-III. The agreement between the two
libraries is very good and in the scale used in Figs 7.1
and 7.2 the cross sections coincide. There are, however,
other evaluations giving different data and it has been
pointed out, e.g. in [56]> that the spread in measured data
for U-235 is relatively large.

In order to study the effect of modifications in U-235 data

on calculated reactivities and temperature coefficients a
was changed keeping a unchanged. The change in the shape
cl
of a versus energy is shown in Fig 7.3. This modification
lies within the scatter of experimental points and is the
same one as has been considered by Askew L563• Our choice
for this modification of a is arbitrary. It represents
one among several modifications which may be considered.
The influence on calculated reactivities and temperature
coefficients is shown in Table 7.9.
Made with Xodo PDF Reader and Editor

Table 7.7: Group cross sections for a L/v-absorber in the .

AE-BUXY group structure using Maxwellian weighting
spectra of various temperatures.

Q 1
Group

00
300 K 600 K 900 K 1200 K

46 2. 21447E-01 2.16559E-01 2.15074E-01 2.14321E-01

47 2.45213E-01 2.41125E-01 2.39788E-01 2.39105E-01
48 2.62920E-01 2.61079E-01 2.60619E-01 2.60392E-01
49 2.76325E-01 2.75378E-01 2.75185E-01 2.75088E-01
50 2.86520E-01 2.85957E-01 2.85864E-01 2.85817E-01
51 2.96194E-01 2.95673E-01 2.95571E-01 2.95517E-01
52 3 . 10725E-01 3.09782E-01 3.09507E-01 3.09367E-01
53 3.30109E-01 3.20967E-01 3.28735E-01 3.28566E-01
54 3.60025E-01 3.57808E-01 3.57049E-01 3.56666E-01
55 4.03698E-01 4.00596E-01 3.99528E-01 3.98991E-01
56 4.68427E-01 4.63581E-01 4.61926E-01 4.61100E-01
57 5.33721E-01 5.31824E-01 5.31190E-01 5.30874E-01
58 5.88684E-01 5.87 594E-01 5.87233E-01 5.87053E-01
59 6.37430E-01 6.36764E-01 6.36544E-01 6.36435E-01
60 6.85634E-01 6.8497 9E-01 6.84764E-01 6.84655E-01
61 7.43046E-01 7.42215E-01 7.41941E-01 7.41802E-01
62 8.12023E-01 8.11195E-01 8 . 10919E-01 8.10779E-01
63 8.83177E-01 8.82635E-01 8.82452E-01 8.82358E-01
64 9.60238E-01 9.59539E-01 9.59302E-01 9.59182E-01
65 1.06175E+00 1.06080E+00 1.06048E+00 1.06032E+00
66 1.20412E+00 1.20273E+00 1.20226E+00 1.20203E+00
67 1.42486E+00 1 .42256E+00 1.42179E+00 1.4214IE+00
68 1.83773E+00 1.83282E+00 1.83120E+00 1.83040E+00
69 3.03882E+00 3.0J 881E+00 3.01227E+00 3.00901E+00
 Made with Xodo PDF Reader and Editor

The absorption cross section of U-238 is generally assumed

to obey the 1/v-law at thermal energies. This is the case
also in UKNDL and ENDF/B-III. Table 7.8 shows the lowest
positive energy resonances of U-238. There are no significant
discrepancies between data from different references.

Table 7.8: U-238 resonance parameters in BNL-325

E S rn r r I
y
(eV) (meV) (meV) (meV)

4.41 .000111 1

6.67 1.52 26 27.5 0

10.25 .00156 1
11.32 .00035 1
16.3 .00005 1

19.50 .0014 1

20.90 8.7 25 34 0
36.80 32.0 25 57.0 0
66.15 26.0 22 48 0

Extrapolating from the positive energy resonances one

finds that the first negative energy resonance may be
expected at approximately -10 eV, which means that its
contribution to at thermal energy is of the 1/v-shape.
The contribution from this negative energy resonance is
of importance when deriving the thermal . At
E = 0.0253 eV the first 22 positive levels contribute
2.38 barns [57] and with °2200 = the remaining 0.35 b
usually is attributed to a single negative energy resonance.
When generating U-238 data for ENDF/B-III a single
negative energy resonance was placed at -15 eV [57].

The possibility of a resonance close to zero energy can,

however, not be excluded. If such a resonance exists, its
contribution to the thermal absorption cross section will
be such that the gradient of a versus energy will be
more negative than for a 1/v-cross section.
 Made with Xodo PDF Reader and Editor

In order to investigate how a non-l/v cross section affects

the predicted temperature coefficient a change of the absorp­
tion cross section was made below 0.3 eV. The modification
was made so that the reaction rate in a 20 °C Maxwellian
spectrum was unchanged. Modified and unmodified cross
sections are shown in Fig 7.5 and results from the calcula­
tions with the modified data are collected in Table 7.9

Table 7.9: k and temperature coefficients calculated by AE-BUXY using modified

thermal U-235 and U-238 data

k OO Ak OO /AT Ak ,,/AT
k
eff f ett
Lattice Modification 20°C 210°C (pcm/°C) 20°C 210°C (pcm/°C)
i
1.35% U 0 2 Reference 1.16299 1.13875 -12.76 1.00000 1.00000 I 0
1
1

V m /Vf=1.4 +483 +741 + 1.36 +412 1 +645 + 1.23

r A°f235
0 ppm B +500 + 1172 +3.54 +380 i +966 +3.08
A°a238 i

1.9% U 0 2 Reference 1.21010 1.19573 -7.56 1.00000 1.00000 0

Vm /V,=1.7
r +519 +785 +1.40 +426 +650 + 1.18
A °f235
200 ppm +392 +909 + 2.72 +338 +765 +2.25
A0a238

In the 1.35 % U 0 2 lattice the temperature coefficient was

predicted about 4 pcm/°C too negative (cf Table 4.3) and in
the 1.9 % UO^ lattice 2 pcm/°C too negative (cf Table 4.7).
The modification of U-238 gives the correction +3 pcm/°C
and +2 pcm/°C in the 1.35 % U02 and 1.9 % U 0 2 lattices,
respectively. The U-235 modification gives about +1 pcm/°C
in both lattices.

Data for some lattices which were analysed in chapter 4 are

collected in Table 7.10. The thermal absorption in U-235
and U-238 as calculated by AE-BUXY is given in percent of
the total cell absorption plus leakage. Assume that the
change in calculated Ak^^/AT due to modified U-235 and
U-238 cross sections is proportional to the relative absorp­
tion in the nuclide. Then corrections to Ak ,.r/AT are
eff
 Made with Xodo PDF Reader and Editor

obtained by combining the results of Table 7.9 with the

relative absorptions listed in Table 7.10. These estimated
corrections are given in the last two columns and are com­
pared with the discrepancies obtained in the comparisons
between theory and experiment.

It is seen that for all these lattices a close agreement

between theory and experiment can be obtained by modifying
the shape of the thermal U-238 absorption cross section as
function of energy.

The influence of a modification in the shape of the thermal a

for U-235 on the temperature coefficient has previously been
studied elsewhere, e.g. in [56, 58]. Askew et al. [4]
conclude that nuclear data uncertainties indicate that
extreme changes within the uncertainties of nuclear data
in the energy dependence of r) for U-235 could contribute
about +1 pcm/°C to the temperature coefficient. Our results
are in agreement with his conclusion.

Basiuk et al. [59] have made calculations on a PWR lattice

assuming a p-wave resonance in U-238 at 0.1 eV. This did
not change the calculated temperature coefficient and they
conclude that a p-wave resonance at a lower energy would
not influence the results.
 Made with Xodo PDF Reader and Editor

T a b l e 7.10: E s timated c o r rections to Ak £ C /AT d u e to m o d i f i e d a anu

------------ eff a238 f235
for some K R I T Z lattices.

Therma abs (%) (pcm/°C

i k e f £ /4T
D i s c r e ­ Correct ion du e to
V / V f and Temp
m f pancy modifie i
Fuel bor o n conc °C U235 U238 th-exp
°f235 °a238

1.35% U 0 9 V /V,=1.4 20 41.1 12.5

m f
-4.4 + 1.2 +3.1
0 ppm B 210 40.6 12.7

V / V f =1.4 20 40.9 12.4

m f
-3.3 + 1.2 +3.1
175 p pm B 210 40.1 12.5

1.9% U 0 2 V / V f =1.2 20 39.1 8.6

m i
-1.4 + 1.1 +2.1
200 p p m B 210 38.0 8.6

V /V,=l.7 20 4 1.2 9.0

m f
-1.8 + 1.2 +2.3
200 p pm B 210 40.4 9.1

1.5% P u 0 2 V /V,=l.8 20 2.5 5.7

m 1
-0.4 +0.1 +1.3
300 p p m B 210 1.8 4.9

V /V,=2.5 20 2.3 6.1

m t
- 1.0 +0.1 + 1.4
250 p pm B 210 1.9 5.2

V /Vf=3.3 20 2.5 6.4

m r
-0.6 +0.1 + 1.5
0 p pm B 210 2.3 5.5
Made with Xodo PDF Reader and Editor

Fig. 7.1 oQ for U-235 in the AE-BUXY library-

Made with Xodo PDF Reader and Editor

Fig.7.2 o f for U-235 in the AE-BUXY library

 Made with Xodo PDF Reader and Editor

106

Fig. 73. a for U - 2 3 5 in the A E - B U X Y lib r a r y and in ENDF/B m .

Made with Xodo PDF Reader and Editor
 Made with Xodo PDF Reader and Editor

s.

^ M o d fied shape of cfa in the present study

....

>vS s
\

............. ....
v v

0.001 0.01 0.1 eV E

Fig. 7.5 orQ for U-238 in the AE-BUXY library

Made with Xodo PDF Reader and Editor

Our comparison between calculated and measured temperature

coefficients shows that the employed theory predicts a too
negative coefficient, the discrepancy being 1-4 pcm/°C.

The partial temperature coefficients - the fuel temperature

coefficient, the moderator temperature coefficient and the
water density coefficient - were theoretically determined.

The fuel temperature coefficient is approximately proportional

- 1/2
to T . The dominating component is caused by the Doppler
b roadening of the U-238 resonances and an accurate calcula­
tion of the Doppler coefficient requires that the crystalline
binding in the U0„ is taken into a c c o u n t ./This effect is
^ V
largest at low temperature. It can be accounted for by using
an effective Doppler temperature which is higher than the
true temperature of the medium. The correction of the
D oppler coefficient due to the crystalline binding is
about 15 % at room temperature and 3 % at 1000 K.N

The moderator temperature coefficient consists mainly of three

components 3p/3rr3n/3T , 3p/3f-3f/3T and 3p/3L2 -3L2 /3T
m m m
in the thermal energy group. 3p / 3 f •3f/3Tm is in all. lattices
. . /.
positive and the other two components are negative./The m a g ­
nitude of 3p/3f'3f/3Tm is much larger in Pu02 lattices
than in U O 2 lattices. The discrepancy between calculated
and measured temperature coefficients is approximately the
same for all lattices, however, and one may therefore expect
that the employed model for the thermal scattering in water
predicts the temperature dependence of the spectrum correctly.
This conclusion is supported by the comparison between r e ­
sults using the Nelkin, the effective width and the Haywood
models for scattering in water. All three models give the
same temperature coefficient although the Haywood model p re­
dicts a harder spectrum than the other two models.)
 Made with Xodo PDF Reader and Editor

The water density coefficient varies considerably with

temperaturefwhereas the discrepancy between theoretical
and experimental temperature coefficients is nearly the same
at low and high temperature. This observation indicates that
the inconsistency between theory and experiments is not due
to effects caused by the water density variation with tem­
perature. '

Various approximations in the theory have been validated. We

have found that the cylindricalization of pin cells, the
utilization of transport corrected cross sections to account
for anisotropic scattering in the cell calculation and the
use of a limited number of energy groups and spatial meshes
in the calculations introduce very small errors in the pre­
dicted temperature coefficients.

The comparison between results using the UKNDL and the

ENDF/B III data shows that these data sets provide the same
temperature coefficient.!

The temperature coefficient is sensitive to the energy

dependence of the thermal U-235 and U-238 cross sections.
An error of % 1 pcm/°C is possible due to the uncertainty
in the capture to fission ratio as function of energy for
U-235. The thermal U-238 absorption cross section is usu­
ally assumed to obey the 1/v-law. However, if a negative
energy resonance exists close to zero energy its influence
on a will be such that a decreases with higher energy
& a
faster than the 1/v-law prescribes. Such a shape of a
a
versus energy would produce a less negative temperature
coefficient. It was found that the calculated temperature
coefficients for all investigated cores can be brought in
agreement with experimental values by a modification of the
thermal U-238 absorption cross section shape. This modifi­
cation would also give calculated values of closer to
unity and with a smaller spread.
 Made with Xodo PDF Reader and Editor

The w o r k presented here was carried out at the section

for Reactor Physics of AB Atomenergi, Studsvik. Part of
the w o r k was sponsored by the Swedish Board for Technical
Development (STU).

I would like to thank my colleagues for many fruitful d i s ­

cussions. My thanks are particularly due to Drs E Hellstrand,
H H a g g b l o m and R Persson at a B Atomenergi and Prof N G
Sjostrand at Chalmers University of Technology.

- Ill -
 Made with Xodo PDF Reader and Editor

1. 0 BERNANDER et al.,
"Asea-Atom BWR Development and Core Performance
Experience", NUC L E X 72, Technical meeting No 2/3.

2. A F McFARLANE,
Nuv-1 Appl & Tc c h n j), 634 (1970).

3. R I HELLENS,
"The physics of PWR reaoturs", National Topical
Meeting on New Developments in Reactor Physics
and Shielding, Sept 1972, CONF-720901, p 3.

4. J R ASKEW, F JPAYERS and W FOX,

"Thermal reactor temperature coefficient studies
in the UK". EACRP-U-49 (1973).

5. L VAN HOVE,
Phys Rev 95^ 249 (1954).

6. M S NELKIN,
Phys Rev 119, 741 (1960).

7. P A EGELSTAFF and PSCHOFIELD,

Nucl Sci Eng J_2, 260 (1962).

8. B C HAYWOOD,
J Nucl Energy 21_, 249 (1967).

9. M S NELKIN and D E PARKS,

Phys Rev ]19, 1060 (1960).

10. G DOLLING et al,

Can J of Physics, 4J3, 1397 (1965).

11. See for example,

G BELL and S GLASSTONE,
"Nuclear Reactor Theory", Van Nostrand Reinhold
Company, 1970, pp 401 et seq.

12. See e.g. ref 11, p 423.

13. L DRESNER,
"Resonance Absorption in Nuclear Reactors",
Pergamon Press, 1960, p 43.

14. D S NORTON,
"The UKAEA Nuclear Data Library February 1968",
AEEW-M-824 (1968).

15. See e.g. ref 11, p 447.

 Made with Xodo PDF Reader and Editor

J CHERNICK,
"The theory of uranium water lattices"
Proc First UN Conf on Peaceful Uses of Atomic
Energy, 5, 215 (1955).

17. R BONALUMI,
Energia Nucleare £i, 326 (1961).

18. E P W I G N E R et a l ,
J Appl Phys 2, 257 (1955).

19. J R ASKEW,
"The calculation of resonance captures in a few-
group approximation", AEEW-R-489 (1966).

20 . B R SEHGAL and R GOLDSTEIN,

Nucl Sci Eng 25, 174 (1966).

21 . J R A S K E W et al,
JBNES 5, 564 (1966).

22. I CARLVIK,
"A method for calculating collision probabilities
in general cylindrical geometry", Proc Third UN
Conf on Peaceful Uses of Atomic Energy, 2, 225
(1964) .

23. I CARLVIK,
N u k l e o n i k JJO, 104 (1967).

24. R PERSSON et al,

"High temperature critical experiment with t^O-
moderated U O 2 rod lattices in KRITZ". 1972 Reactor
M e e t i n g of Deutsches Atomforum, Sektion 1 (paper
128).

25. G A B A T ’ et al,
A t o m n a y a Energiya 30i, 354 (1971) .

26. R PERSSON,
Private communication.

27. A AHLIN,
Internal AE-report (1973).

28. K D LATHROP,
"DTF-IV A Fortran-IV program for solving the
mul t i g r o u p transport equation w i t h anisotropic
scattering", LA-3373 (1965).

LI I KALA et al,
Nu c l Techn 15, 272 (1972).
 Made with Xodo PDF Reader and Editor

30. W LAMB,
Phys Rev 55, 190 (1939).

31. C R ADKINS et al,

"The chemical binding effects on the resonance
line shapes of uranium 238 in a IK^ lattice",
Conf on Neutron Cross Section Techn, March 1966,
Washington D C, p 134.

32. G DOLLING et al,

Can J of Physics, 43, 1.397 (1966).

33. R SHEUiEn.
"Effects of U0„ crystalline binding on Doppler
coefficient calculations for fast reactor systems",
BNWI-442 (1967).

34. H C HONECK,
Trans A m _5, 350 (1962) .

35. H C HONECK,
Nucl Sci Eng 18, 49 (1964).

36. Y FUKAI,
Reactor Sci Te c h n 17, 115 (1963).

37. Y FUKAI,
J Nucl Ene r g y A/B 1_8, 241 (1964) .

38. A SAUER,
J Nucl Ene r g y A/B 18, 425 (1964).

39. I CARLVIK,
"Calculations of neutron flux distributions by
means of integral transport methods". AE-279 (1967).

40. T E DUD L E Y and P B DAITCH,

Nucl Sci Eng 25, 75 (1966).

41. D A NEWMARCH,
"Errors due to the cylindrical cell approximation
in lattice calculations". AEEW-R34 (1960).

42. Z WEISS and R J J STAMM'LER,

Nucl Sci E n g 19, 374 (1964).

43. A AHLIN, A JONSSON,

Internal AE-report (1967).

44. A AHLIN,
Internal AE-report (1971).

45. G ANDERSSON,
Pr i vate communication.
 Made with Xodo PDF Reader and Editor

46. H HAGGBLOM,
"The ETOS and DO RIX programmes for processing
ENDF/B data", Seminar on Nuclear Data Processing
Codes, Ispra, Dec 1973, EACRP-U-52.

47. H C H O N E C K and D R FINCH,

"FLANGE-II, A code to process thermal neutron data
from an ENDF/B tape". USAEC Report DP-1278,
Savannah River Laboratory, E I du Pont de Nemours
& Co, Aiken, S C (1971).

48. f J FAYERS et al,

JBNES 6 , 161 (1967) .

49. J HARDY Jr,

Trans A m Nucl Soc JJ3, 351 (1974).

50. "Seminar on U-238 resonance capture" edited by

S Pearlstein, M a r c h 1975. BNL-NCS-50451 (ENDF-217).

51. E HELLSTRAND and G LUNDGREN,

Nucl Sci Eng 12, 435 (1962).

52. P BLOMBERG, E HELLSTRAND and S HORNER,

Nucl Sci Eng J3, 497 (1960).

53. E HELLSTRAND,
"Measurements of resonance integrals", Reactor
physics in the resonance and thermal regions.
Proc of the National Topical Meeting, San Diego,
Febr 1966, Vol II, p 151.

54. J CARLVIK,
Internal AE-report (1962).

55. A B REYNOLDS and S L STEWART,

"Application of the SEFOR critical experiments
at ZPR-3 to SEFOR", International Conference on
Critical Experiments and Their Analysis, Oct 1966,
p 41.

56. R CHAWLA,
"An assessment of methods and data for predicting
integral properties for thermal reactor physics
experiments". A E E W - R 797 (1972).

57. W A WITTKOPF et al,

"U-238 neutron cross section data for ENDF/B".
BAW 316 (1967).

58. P B KEMSHELL,
"Some integral properties of nuclear data deduced
from WIMS analyses of well thermalized uranium
lattices". A E E W - R 786 (1972).

59. M J BASIUK, G LE COQ and P REUSS,

"Suggestion to measure the U-238 capture cross
section in the thermal region". EACRP-A-203 (1973).

A t :s kontorstryckeri